TH2019PESC2039

Joint PhD degree in Environmental Technology
Docteur de l’Université Paris-Est

Spécialité: Science et Technique de l’Environnement
Dottore di Ricerca in Tecnologie Ambientali
Degree of Doctor in Environmental Technology
Thesis for the degree of Doctor of Philosophy in Environmental Technology
PhD thesis –Väitöskirja – Proefschrift – Tesi di Dottorato – Thèse
Bikash Chandra Maharaj

ADM1 Based Mathematical Models For Assessing The Effect Of Trace
Elements Dynamics On Solid Waste Anaerobic Digestion
To be defended on 10/12/2019, Paris

In front of the PhD evaluation committee
Prof. Patrice Chatellier Chairman, Reviewer

Prof. Lorenzo Bertin Reviewer
Prof. Blaz Stres Reviewer
Prof. Giovanni Esposito Promotor
Prof. Eric D. van Hullebusch Co-Promotor
Prof. Piet N. L. Lens Co-Promotor
Prof. Jukka Rintala Co-Promotor
Marie Skłodowska-Curie European Joint Doctorate Advanced Biological Waste-to-Energy Technologies (ABWET)
i
Evaluation Committee
Chair
Prof. Patrice Chatellier
French Institute of Science and Technology in Transportation, Planning and Networks,
Marne-la-Vallee Cedex 2, Bienvenue, Paris, France
Reviewers/Examiners
Prof. Patrice Chatellier
French Institute of Science and Technology in Transportation, Planning and Networks,
Marne-la-Vallee Cedex 2, Bienvenue, Paris, France
Prof. Lorenzo Bertin

Department of Civil, Environmental and Materials Engineering, Engineering Faculty
of the University of Bologna, via Terracini 28, I- 40131, Bologna, Italy
Prof. Blaž Stres

University of Ljubljana, Biotechnical Faculty, Department of Animal Science,
Jamnikarjeva 101, 1000 Ljubljana, Slovenia
University of Ljubljana, Faculty of Civil and Geodetic Engineering, Jamova 2, 1000
Ljubljana, Slovenia
Jozef Stefan Institute, Department of Automation, Biocybernetics and Robotics,
Jamova 25, 1000 Ljubljana, Slovenia
University of Innsbruck, Department of Microbiology, Technikerstrasse 25d, A-6020
Innsbruck, Austria
Thesis Promotor
Prof. Giovanni Esposito
Department of Civil and Mechanical Engineering, University of Cassino and Southern
Lazio, Cassino, Italy
Thesis Co-promotors
Prof. Eric D. van Hullebusch
Laboratoire Géomatériaux et Environnement (LGE), Université Paris-Est, Marne-la-
Vallée, Paris, France
Prof. Piet N. L. Lens

Department of Environmental Engineering and Water Technology
IHE Delft Institute for Water Education, Delft, The Netherlands
Prof. Jukka Rintala

Faculty of Engineering and Natural Sciences, Tampere University, Finland
ii
Supervisory Team
Thesis Supervisor
Prof. Giovanni Esposito
Present: Department of Civil, Architectural and Environmental Engineering,
University of Naples Federico II, via Claudio, Naples, Italy
Former: Department of Civil and Mechanical Engineering, University of Cassino and
Southern Lazio, Cassino, Italy
Thesis Co-supervisors
Prof. Eric D. Van Hullebusch
Present: Universitei de Paris, Institut de Physique du Globe de Paris, CNRS, F-75005
Paris, France
Former: Laboratoire Géomatériaux et Environnement (LGE), Université Paris-Est,
Marne-la-Vallée, Paris, France
Asst. Prof. Luigi Frunzo

Department of Mathematics and Applications "Renato Caccioppoli", University of
Naples Federico II Naples, Naples, Italy
Dr. Maria Rosaria Mattei

Department of Mathematics and Applications "Renato Caccioppoli", University of
Naples Federico II Naples, Naples, Italy
iii
ABSTRACT
English
Anaerobic digestion (AD) is one of the most trace element (TE) rich metabolic processes in
biology. Indeed, TEs are important structural components of various enzymes in the AD process.
The role and fate of TEs (Fe, Ni and Co) in AD is poorly understood due to their complex
biogeochemistry. The low detection limit of analytical instruments and the time consuming and
challenging nature of the experimental procedures are major obstacles to the quantification of TEs
in AD. In this thesis, three separate mathematical models based ADM1 have been developed to
simulate the TEs dynamics and speciation pattern in an AD reactor. In particular, a TE
precipitation/dissolution model, a TE complexation model and a TE adsorption model have been
progressively developed to predict the effect, role and fate of TEs in an AD batch system. In all
the models the extent of microbial activity in the AD process is a function of the free TE
concentration in the liquid phase, which is in equilibrium with the physicochemistry of the AD
reactor. The precipitation/dissolution model considers the interactions of TEs with inorganic
carbon (e.g. HCO3- and CO32-), phosphorous (e.g. PO43-, HPO42-, H2PO4-) and sulfur (e.g. HS- and
S2-) components. New chemical equilibrium acid-base and precipitation reactions have been
implemented to study the interactions of Fe, Ni, Co with carbonate, phosphate and sulfide
components. The effects of deficiency, stimulation, inhibition and toxicity of TEs on microbial
activity have been modelled based on a hormesis type TE dose-response inhibition function. The
microbial uptake of TE and the TE inhibition on special microbial activities have been defined as
well. Release of TEs as a disintegration product has been also considered to account for the TE
content of the organic substrate. Model scenarios have been simulated to analyze the dynamics of
TEs, starvation of TEs and the effect of initial sulfur-phosphorus ratio. In the complexation model,
the interactions of TEs with organic chelators have been predicted. TE complexation reactions
with VFAs and EDTA have been incorporated in the extended ADM1 model in addition to TE
precipitation/dissolution processes. New acid-base chemical equilibrium reactions have been
incorporated to model the dynamics of EDTA species. Complexation process rates have been
defined as well. The model is able to quantify the effect of EDTA/VFA -TE complexation on
methane production. Further, effect of initial Ca and Mg concentration on TE complexation has
been predicted in a separate modelling scenario. Finally, a general framework able to take into
account the precipitation/dissolution and complexation reactions, as well as the interaction of TEs
with various surfaces available in the AD system has been developed. The model tracks the TEs
dynamics in a batch anaerobic digester and as an extension of the previous contributions,
incorporates the adsorption reactions of TEs with biomass, inert and precipitate (FeS). The
concepts of free and occupied binding sites, and binding site density for the various surfaces have
been incorporated into the model. Simulation scenarios were able to predict the effect of various
organic matter concentrations, initial TE concentrations, initial Ca-Mg concentrations, initial
EDTA concentrations and change in TE binding site density for biomass, inert and precipitate on
cumulative methane production and TE speciation.
iv
Italian
La Digestione Anaerobica (DA) rappresenta uno dei processi biologici che coinvolgono il maggior
numero di elementi in tracce (TE). In effetti, i TE costituiscono fondamentali componenti
strutturali di una moltitudine di enzimi operanti in DA. Il ruolo e l’evoluzione che i TE (Fe, Ni e
Co) assumono nel contesto della DA non sono ancora stati completamente definiti e compresi, a
causa della loro complessa biogeochimica. La limitata capacità di rilevazione degli strumenti
analitici, i tempi di attuazione e la complessità delle attuali procedure sperimentali costituiscono
le principali difficoltà per la quantificazione dei TE in DA. Nel presente lavoro di tesi, sono stati
messi a punto e sviluppati tre differenti modelli matematici, basati sull’ADM1, al fine di simulare
il comportamento ed i pattern di speciazione degli TE in un reattore di DA. Nello specifico, sono
stati progressivamente sviluppati un modello riguardante la precipitazione/dissoluzione, uno sulla
complessazione ed uno sull’adsorbimento, al fine di predire l’effetto, il ruolo e l’evoluzione degli
TE in un sistema batch di DA. In tutti i modelli, l’attività microbica del processo di DA è
condizionata del contenuto di TE liberi nella fase liquida, in equilibrio con le altre fasi fisico-
chimiche presenti nel reattore di DA. Il modello di precipitazione/dissoluzione considera le
interazioni dei TE con composti inorganici del carbonio (ad es. HCO3- e CO32-), del fosforo (ad es.
PO43-, HPO42-, H2PO4-) e dello zolfo (ad es. HS- e S2-). Sono state implementate nuove reazioni di
equilibrio acido-base e di precipitazione per studiare le interazioni di Fe, Ni, Co con i suddetti
componenti. Gli effetti della limitazione, stimolazione, inibizione e tossicità dei TE sull'attività
microbica sono stati modellati sulla base di una funzione di inibizione dose-risposta di tipo ormesi.
Sono stati anche modellati il consumo microbico dei TE e l'inibizione di specifiche attività
microbiche condizionate dalla presenza di tali elementi. Il rilascio dei TE quali prodotti della
disintegrazione è stato considerato al fine di tenere conto del contenuto di TE del substrato
organico. Sono stati simulati differenti scenari per analizzare le dinamiche dei TE, la loro presenza
in condizioni limitanti e l'effetto del rapporto iniziale zolfo-fosforo. Nel modello di
complessazione, sono state inglobate le interazioni dei TE con chelanti organici. Le reazioni di
complessazione con VFA ed EDTA sono state incorporate nel modello ADM1 esteso, oltre ai
processi di precipitazione/dissoluzione. Nuove reazioni di equilibrio chimico acido-base sono state
incorporate per modellare le dinamiche della specie EDTA. Inoltre, sono state definite le velocità
specifiche del processo di complessazione. Il modello è in grado di quantificare l'effetto della
complessazione EDTA/VFA-TE sulla produzione di metano. Inoltre, l'effetto della concentrazione
iniziale di Ca e Mg sulla complessazione dei TE è stato studiato in simulazioni numeriche
aggiuntive. Infine, è stato sviluppato un modello generale in grado di tener conto delle reazioni di
precipitazione/dissoluzione e complessazione, nonché dell'interazione dei TE con varie superfici
di adsorbimento disponibili nel sistema DA. Il modello riproduce la dinamica dei TE in un
digestore anaerobico batch e come estensione dei precedenti contributi, incorpora le reazioni di
adsorbimento dei TE con le biomasse, materiale inerte e precipitati (FeS). I concetti di siti di
adsorbimento liberi ed occupati e di densità dei siti per le varie superfici di adsorbimento sono stati
incorporati nel modello. Le simulazioni numeriche mostrano che il modello è in grado di prevedere
l'effetto di varie concentrazioni di sostanza organica, concentrazioni iniziali di TE, concentrazioni
v
iniziali di Ca-Mg, concentrazioni iniziali di EDTA e cambiamenti nella densità dei siti per
biomasse, inerte e precipitati sulla produzione cumulativa di metano e la speciazione dei TE.
French
La digestion anaérobie est l’un des processus métaboliques les plus exigeants quant au besoin en éléments
traces métalliques essentiels (ETM). En effet, les ETM essentiels jouent de cofacteur enzymatique pour
diverses activités enzymatiques primordiales dans le processus digestion anaérobie. Cependant, le rôle et
le devenir des ETM (e.g. Fe, Ni et Co) dans les digesteurs sont mal connus en raison de processus
biogéochimiques complexes. Les limites analytiques et la complexité de procédures expérimentales
chronophages constituent des obstacles majeurs à l’étude de la spéciation des ETM dans les digesteurs.
Dans cette thèse, trois modèles mathématiques distincts basés sur le modèle ADM1 ont été développés pour
simuler la dynamique des ETM et leur spéciation dans les réacteurs anaérobies. En particulier, un modèle
de précipitation / dissolution de ETM, un modèle de complexation des ETM et un modèle d'adsorption des
ETM ont été progressivement développés pour prédire l'effet, le rôle et le devenir des ETM essentiels dans
un digesteur fonctionnant en mode discontinu. Dans tous les modèles, l’activité microbienne dans le
procédé de méthanisation est fonction de la concentration de l’ion libre de chaque ETM dans la phase
liquide. La concentration de l’ion libre étant régulée par les équilibres physicochimiques majeurs
habituellement décrits. Le modèle de précipitation / dissolution prend en compte les interactions de ETM
avec les carbone (par exemple HCO3- et CO32-), les phosphore (par exemple PO43-, HPO42-, H2PO4-) et les s
(par exemple HS- et S2-). De nouvelles réactions acide-base et de précipitation à l'équilibre chimique ont
été mises en rentré dans le modèle pour étudier les interactions de Fe, Ni, Co avec les carbonates, phosphates
et sulfures. Les effets de la déficience, de la stimulation, de l'inhibition et/ou de la toxicité des ETM sur
l'activité microbienne ont été modélisés sur la base d'une fonction d'inhibition de la réponse à la dose d’ETM
essentiels de type hormèse. L'acquisition microbienne des ETM essentiels et l'inhibition de ETM sur des
activités microbiennes spécifiques ont également été définies. La libération de ETM en tant que produit de
désintégration a également été prise en compte pour tenir compte de la teneur en ETM du substrat
organique. Des scénarios modèles ont été simulés pour analyser la dynamique des ETM, la déficience des
ETM et l'effet du rapport soufre-phosphore initial. Dans le modèle de complexation, les interactions des
ETM avec des ligands organiques ont été prédites. Les réactions de complexation des ETM avec les acides
gras volatils (AGV) et l'EDTA ont été incorporées dans le modèle ADM1 étendu en complément des
processus de précipitation / dissolution des ETM. De nouvelles réactions d’équilibre chimique acide-base
ont été incorporées pour modéliser la dynamique des espèces d’EDTA. Les taux de complexation ont
également été définis. Le modèle est capable de quantifier l'effet de la complexation EDTA / AGV-ETM
sur la production de méthane. De plus, l'effet de la concentration initiale en ions Calcium et de Magnésium
sur la complexation des ETM a été prédit dans un scénario de modélisation séparé. Enfin, un cadre général
capable de prendre en compte les réactions de précipitation / dissolution et de complexation, ainsi que
l’interaction des ETM avec diverses phases solides disponibles dans les réacteurs a été mis au point. Le
modèle prédit la dynamique des ETM dans un digesteur anaérobie fonctionnant en mode discontinu et, en
prolongement des contributions précédentes, intègre les réactions d'adsorption des ETM sur la biomasse,
des particules inertes et les précipités (i.e., FeS). Les concepts de sites de liaison libres et occupés et de
densité de sites de liaison pour les différentes surfaces ont été intégrés au modèle. Les scénarios de
simulation ont permis de prédire l’effet de diverses concentrations de matière organique, concentrations
initiales en ETM, concentrations initiales en calcium et magnésium, concentrations initiales en EDTA et
modification de la densité de sites de liaison des ETM pour la biomasse, les particules inertes et les
précipitées sur la production cumulative de méthane et la spéciation des ETM.
vi
Dutch
Anaërobe vergisting (AD) is een van de meest sporenelementen (SE) rijke metabolische processen
in de biologie. SE zijn inderdaad belangrijke structurele componenten van verschillende enzymen
in het AD-proces. De rol en het lot van SE (Fe, Ni en Co) in AD wordt slecht begrepen vanwege
hun complexe biogeochemie. Ook de lage detectielimiet van analytische instrumenten en de
tijdrovende en uitdagende aard van de experimentele procedures vormen belangrijke obstakels
voor de kwantificering van SE in AD. In dit proefschrift zijn drie afzonderlijke wiskundige
modellen gebaseerd op ADM1 ontwikkeld om de SE dynamiek en hun speciatiepatroon in een
AD-reactor te simuleren. In het bijzonder zijn een SE neerslag / oplossing model, een SE
complexatie model en een SE adsorptive model geleidelijk ontwikkeld om het effect, de rol en het
lot van SE in een AD batch systeem te voorspellen. In alle modellen is de mate van microbiële
activiteit in het AD proces een functie van de vrije SE concentratie in de vloeibare fase, die in
evenwicht is met de fysicochemie van de AD reactor. Het neerslag / oplossing model houdt
rekening met de interacties van SE met anorganische koolstof (bijvoorbeeld HCO3 - en CO32-),
fosfor (bijvoorbeeld PO43-, HPO42-, H2PO4-) en zwavel (bijvoorbeeld HS- en S2-) componenten.
Nieuwe chemische evenwicht zuur / base en neerslagreacties zijn geïmplementeerd om de
interacties van Fe, Ni en Co met carbonaat, fosfaat en sulfidecomponenten te bestuderen. De
effecten van deficiëntie, stimulatie, remming en toxiciteit van SE op de microbiële activiteit zijn
gemodelleerd op basis van een SE-dosis/remming functie van het hormese type. De microbiële
opname van SE en de SE-remming van speciale microbiële activiteiten zijn ook gedefinieerd. Het
vrijkomen van SE als een desintegratieproduct is ook meegenomen voor het SE gehalte van het
organische substraat. Modelscenario's zijn gesimuleerd o,m de dynamiek van SE, depletie van SE
en het effect van de initiële zwavel / fosforverhouding te analyseren. In het complexatiemodel zijn
de interacties van SE met organische chelatoren voorspeld. SE-complexatiereacties met VFA's en
EDTA zijn opgenomen in het uitgebreide ADM1-model naast SE precipitatie / oplos processen.
Nieuwe chemische zuurevenwichtsreacties zijn opgenomen om de dynamiek van EDTA speciatie
te modelleren. Complexatie proces snelheden zijn ook gedefinieerd. Het model is in staat het effect
van EDTA / VFA - SE complexatie op de methaanproductie te kwantificeren. Verder is het effect
van initiële Ca en Mg concentratie op SE complexering voorspeld in een afzonderlijk
modelleringscenario. Ten slotte is een algemeen raamwerk ontwikkeld dat rekening kan houden
met de neerslag / oplos- en complexatiereacties, evenals de interactie van SE met verschillende
oppervlakken die beschikbaar zijn in het AD systeem. Het model volgt de SEs dynamiek in een
batch anaërobe vergister en als een uitbreiding van de vorige bijdragen, zijn ook adsorptiereacties
van SE op biomassa, een inerte fraktie en neerslagvorming (FeS) meegenomen. De concepten van
vrije en bezette bindingsplaatsen en bindingsplaatsdichtheid voor de verschillende oppervlakken
zijn in het model opgenomen. Simulatiescenario's konden het effect van verschillende organische
stofconcentraties, initiële SE concentratie, initiële Ca-Mg concentratie, initiële EDTA concentratie
en verandering in SE-bindingsplaatsdichtheid voor de biomassa, de inerte reactie en de neerslag
op de cumulatieve methaanproductie en SE speciatie voorspellen.
vii
Preface
This thesis identifies the problem of TE quantification in anaerobic digestion environment and
considers the development of a series of ADM1 based mathematical models as probable solution.
The models developed are analyzed through suitable model scenarios. The first publication
consolidates the scientific background for modelling TE physicochemical processes. The second
publication formulates and implements a new mathematical model for dynamic TE
precipitation/dissolution in the frame work of ADM1. The next paper focuses on the development
of a mathematical model for EDTA/VFA-TE complexation during AD. The fourth paper is
dedicated to the development a of TE- adsorption model in AD systems. All the proposed models
were analyzed by means of extensive numerical simulations.
The thesis is comprised of the following four publications:
1) Maharaj, B. C., Mattei, M. R., Frunzo, L., Hatzikioseyian, A., Van, E. D., Lens, P. N. L., &
Esposito, G. (2019). Mathematical modelling of trace element dynamics in anaerobic
digestion environments. In Trace Elements in Anaerobic Biotechnologies (pp. 101–152).
IWA Publishing. https://doi.org/10.2166/9781789060225
2) Maharaj, B.C., Mattei, M.R., Frunzo, L., van Hullebusch, E.D., Esposito, G., 2018. ADM1
based mathematical model of trace element precipitation/dissolution in anaerobic
digestion processes. Bioresour. Technol. 267, 666–676.
https://doi.org/10.1016/j.biortech.2018.06.099
based mathematical model of trace element complexation in anaerobic digestion
processes. Bioresour. Technol. 276, 253–259.
based mathematical model of trace element speciation in anaerobic digestion
processes. (to be submitted)
viii
Author’s Contribution
Publication I Bikash C Maharaj performed a descriptive literature survey, wrote the

manuscript and is the corresponding author. Maria Rosaria Mattei, Luigi
Frunzo, Artin Hatzikioseyian, Eric D. van Hullebusch, Piet N. L. Lens and
Giovanni Esposito helped in structuring, organizing, interpreting, and
thoroughly revising the manuscript.
Publication II Bikash C Maharaj carried out the mathematical model formulation,

implementation, verification (chemical equilibrium calculations) and
simulations. Further, Bikash C Maharaj wrote the manuscript and is the
corresponding author. Maria Rosaria Mattei and Luigi Frunzo participated
in the development of the conceptual model framework, in the
implementation of the numerical code, in the interpretation of the
simulation results and thoroughly revised the manuscript. Giovanni
Esposito and Eric D. van Hullebusch helped in the identification of the
variables and processes involved, in the definition of the modelling
scenarios, in the results interpretation and thoroughly revised the
manuscript.
Publication III Bikash C Maharaj carried out the mathematical model formulation,
manuscript.
Publication IV Bikash C Maharaj carried out the mathematical model formulation,

manuscript.
ix
ACKNOWLEDGEMENT
I would like to convey my greatest regard and sincere gratitude to the wonderful time and people
I came across during the course of my doctoral studies. The challenges of working in an emerging
field of environmental engineering, for a Joint PhD program, provided me an opportunity to grow
scientifically and personally. First and foremost, I thank the Marie Curie European Joint Doctoral
program ABWET (Advanced Biological Waste-to-Energy Technologies) committee for providing
me with an opportunity to be a part of a challenging, engaging and stimulating research program.
I exclusively thank my PhD thesis promoter Prof. Giovanni Esposito, Prof. Eric D. van Hullebusch,
Prof. Luigi Frunzo for their direction and consultation during the course of the study. Their
continuous encouragement fueled my passion to stay glued on the scientific problems which
eventually taught me the basic tenets of analysis and exploration. My heartfelt thanks to my PhD
thesis co-supervisor, Dr. Maria Rosaria Mattei, for being the main collaborator during the study. I
am obliged forever to Maria for her boundless and infinite support and discussions even on the
weekends and summer holidays. She could reach me from anywhere on the globe. She literally
travelled with me. Thank you Maria for being my “Guru”. I thank Prof. Fernando G Fermoso of
Instituto de la Grasa (C.S.I.C) for giving me the opportunity to participate as a student in the EU
COST Action on ‘The ecological roles of trace elements in anaerobic biotechnologies’. I thank all
my colleagues and friends for making the period of my stay at Cassino, Naples and Delft hassle
free and entertaining. I specifically thank Gilbert and Andreina for their encouragements. I also
thank Dr.Vincenzo Luongo and Dr. Ludovico Puntoni from University of Naples Federico II. I
thank Grafa, Friarielli and Alici for their existence in the human food chain. I am grateful to
Priyanka as well. I seek blessings from Bou, Bapa and Nana. Finally, I thank Archana for her
unconditional love and support during all these years and days to come.
x
Table of Contents Page no
Abstract iv
Preface viii
Author’s contribution ix
Acknowledgements x
Abbreviations 1
1 Introduction 3
1.1 Thesis overview 5
2 Background 8
2.1 Anaerobic digestion and TEs 8
2.1.1 Anaerobic digestion 8
2.1.2 Trace elements 9
2.1.3 Importance of TEs in AD 9
2.1.4 Requirement, inhibition, stimulation and supplementation of TEs in AD 11
2.1.5 Major processes affecting TE dynamics 16
2.1.4.1 TE precipitation/dissolution 16
2.1.4.2 TE complexation 17
2.1.4.3 TE adsorption 20
2.1.4.3 TE biouptake 20
2.1.6 Methods to assess TE speciation in AD 21
2.2 Mathematical modeling of AD 22
2.2.1 ADM1 22
2.2.2 Extensions of ADM1 23
2.3 Modeling TE associated physicochemical processes in AD 24
2.3.1 Modeling TE precipitation/dissolution 24
2.3.1.1 Equilibrium approach 24
2.3.1.2 Equilibrium + kinetic approach 25
2.3.1.3 Kinetic approach 26
2.3.1.4 Precipitation modeling within ADM1 framework 26
2.3.2 Modeling TE complexation process 27
2.3.3 Modeling TE adsorption 28
2.3.4 Modeling TE biouptake 29
3 Model development 32
3.1 ADM1 based mathematical model for TE precipitation/dissolution in anaerobic digestion processes 33
3.1.1 Model implementation and simulations 37
3.2 ADM1 based mathematical model for TE complexation in anaerobic digestion processes 38
3.3 ADM1 based mathematical model for TE speciation in anaerobic digestion processes 42
4 Discussion
4.1 ADM1 based mathematical model for TE precipitation/dissolution in anaerobic digestion processes 47
4.2 ADM1 based mathematical model for TE complexation in anaerobic digestion processes 48
4.3 ADM1 based mathematical model for TE speciation in anaerobic digestion processes 51
4.4 Sources of gaps and potential error in modelling TE speciation in AD 54
5 Conclusion and outlook 57
6 Bibliography 60
7 Publication I
Publication II
Publication III
Publication IV
xi
1
Abbreviations
AD Anaerobic Digestion
ADM1 Anaerobic Digestion Model # 1
BCR Breakpoint Cluster Region
BLM Biotic Ligand Model
BOD Biochemical Oxygen Demand
CH4 Methane
CO2 Carbon Dioxide
COD Chemical Oxygen Demand
EPR Extended Producer Responsibility
EXAFS Extended X-Ray Absorption Fine Structure
FIAM Free Ion Activity Model
H2 Hydrogen
H2 O Water
H2 S Hydrogen Sulfide (H2s)
IWA International Water Association
LCFA Long Chain Fatty Acids
MATLAB MATRIX LABORATORY
ODE Ordinary Differential Equation
pH Potential of Hydrogen
STXM Scanning Transmission X-Ray Microscopy
T Temperature
TE Trace Element
TEM/SEM-EDX Transmission Electron Microscope/ Scanning Electron Microscope- Energy Dispersive
X-Ray Analyzer
UASB Upflow Anaerobic Sludge Blanket
VFA Volatile Fatty Acid
VFA-TE Volatile Fatty Acid bound Trace Elements
XANES X-Ray Absorption Near Edge Structure
XPS X-Ray Photoelectron Spectroscopy
2
1
List of chemical components and species
Ca Calcium
Co Cobalt
CO2 Carbon Dioxide
Co3(PO4)2 Cobalt (2+); diphosphate
CoCO3 Cobalt (2+); carbonate
CoS Sulfanylidenecobalt
EDDS 2-[2-(1,2-dicarboxyethylamino)ethylamino]butanedioic acid
EDTA 2-[2-[bis(carboxymethyl)amino]ethyl-(carboxymethyl)amino]acetic acid
Fe Iron
Fe3(PO4)2 iron(2+);diphosphate
FeCO3 iron(2+);carbonate
FeS sulfanylideneiron
H2 O Water
Mg Magnesium
Mn Manganese
Mo Molybdenum
N Nitrogen
Na Sodium
NH3 Ammonia
Ni Nickel
Ni3(PO4)2 Nickel(2+) diphosphate
NiCO3 Nickel(2+) carbonate
NiS sulfanylidenenickel
NTA N,N-di(carboxymethyl)-glycine
P Phosphorus
S Sulphur
Se Selenium
TE-EDTA Trace element 2-[2-[bis(carboxymethyl)amino]ethyl-
(carboxymethyl)amino]acetic acid
W Tungsten
Zn Zinc
9
10
11
12
13
2
14 CHAPTER 1. INTRODUCTION
15 Anaerobic digestion (AD) is a sequential and multistep collective anaerobic process in which a
16 consortium of microorganisms degrades organic material to produce biogas (CH4, CO2, H2 and
17 H2S) and digestate (Appels et al., 2011; Dai et al., 2017). AD is being recognized as a crucial
18 sustainable technology to harness energy from biomass and to mitigate current CO 2 emissions
19 (Appels et al., 2008). AD reactors have been generally applied to the treatment of sewage sludge
20 as well as to the degradation of the organic fraction of solid waste (Esposito et al., 2012; Khalid et
21 al., 2011). The first step in AD involves the hydrolysis of complex carbohydrates, lipids and
22 proteins to form simpler metabolites such as sugars, amino acids and long chain fatty acids (Appels
23 et al., 2011; Mata-álvarez, 2015; O’Flaherty et al., 2006; Thauer, 1998). Further, these smaller
24 metabolites are taken up by various microbes to produce acids and alcohols during the
25 acidogenesis step. Subsequently, during acetogenesis, acids and alcohols are further degraded to
26 form acetate as major product. In the final step of AD, methanogenic archaea convert acetate into
27 methane. The competition and interdependence for intermediary metabolites of the various
28 microbial groups involved in AD creates fragile environment which can be easily affected by any
29 external perturbation (Feng et al., 2010; Jiang, 2006). Such a sensitivity for external disturbances
30 could lead to AD failure. Among many routinely discussed external disturbances, during the recent
31 years there has been increased focus on the role of inorganic components (Trace Elements (TEs),
32 sulfur (S) and phosphorus (P)) and abiotic processes (precipitation, complexation and adsorption)
33 (Choong et al., 2016; Federovich et al., 2003; Fermoso et al., 2015, 2008a; Hao et al., 2014; Thanh
34 et al., 2015; Zandvoort et al., 2006). Recently, many attempts have been made to predict and link
35 the biotic biochemical degradation reactions with the abiotic physicochemistry of the AD process
36 and vice versa (Fermoso et al., 2019; Roussel et al., 2018; Thanh et al., 2017a; van Hullebusch et
37 al., 2016). A major line of the focus is on the bioavailability of TEs, which are essential
38 micronutrients in the AD processes.
39 Elements which are found in ‘trace amount’ in a system are generally termed as TEs. Iron (Fe),
40 molybdenum (Mo), selenium (Se), cobalt (Co), nickel (Ni), copper (Cu), manganese (Mn), zinc
41 (Zn) and boron (B) fall in this category. Fe, Ni and Co are in the center of studies in optimizing
42 AD processes (Fermoso et al., 2008a, 2008c; Hu et al., 2008; Jansen et al., 2007; Thanh et al.,
43 2017b, 2015; Van Hullebusch et al., 2006; Yekta et al., 2014a). In elemental classification, Fe, Ni
44 and Co belong to the d-block of the modern periodic table and are also termed as d-block transition
45 metals. These TEs characteristically engage in various physicochemical processes in AD systems.
46 TEs are essential for the structural integrity of the catalytic centers of the enzymes involved in the
47 biochemical pathways typical of the anaerobic biodegradation of wastes (Glass and Orphan, 2012).
48 As geochemical entities in AD systems they exhibit and undergo different biogeochemical
49 processes such as precipitation/dissolution, complexation, adsorption (Fermoso et al., 2015; Yekta
50 et al., 2016; Zandvoort et al., 2006). These biogeochemical processes are generally in equilibrium
51 to each other and thus affect the dynamics of the various TEs with respect to the different phases
52 in the anaerobic reactor (Roussel, 2012). For example, an increased precipitation of sulfide ions in
3
53 the form of FeS leads to a shortage of Fe for the synthesis and proper function of various enzymes.
54 This might result in reactor instability and at times failure. Experimental studies have been
55 conducted to study the effect of a TEs dosing strategy based on the operating conditions (Osuna et
56 al., 2003; Roussel et al., 2018; Thanh et al., 2017a; Zandvoort et al., 2006; Zhang et al., 2011). It
57 has been argued that once TEs are dosed into the reactor they may go to any available reaction
58 phase. They may form sulfide, carbonate and phosphate precipitates or they may end up
59 complexing with available organic ligands such as VFAs and any synthetic chelating agents such
60 as EDTA. Further, a fraction of TEs may also get adsorbed on the surfaces of inert, biomasses and
61 mineral precipitates (Fermoso et al., 2015). These reactions take place at a higher velocity as
62 compared to the biouptake of TEs by the microbial community (Batstone et al., 2012). The source
63 stream of TEs in AD systems is largely dosage, degradation of TE containing substrates during
64 AD and human activities. TEs exist in AD reactors in different forms, namely carbonate fraction,
65 sulfide fraction, phosphate fraction, complexed fraction, adsorbed fraction and free TE fraction.
66 The amount of TEs that can be biouptaken by microorganisms is the “bioavailable fraction”. Such
67 fractionation represents the “speciation” pattern of a particular TE. Here the term speciation is
68 used in a geochemistry sense, denoting the presence of TEs in various chemical forms based on
69 the reactor operation conditions. Decrypting the speciation pattern of TEs will help in
70 understanding the dynamics of TEs during various phases of reactor operation and hence equip the
71 research engineer in designing a defined TE supplementation strategy. Such a dosing strategy can
72 be streamlined and made precise only when the amount of TEs in AD systems is effectively
73 quantified so as to allow minimum release and exposure of TEs into the environment. Most of
74 experimental studies employ chemical extraction methods to predict TEs speciation. Combination
75 of BCR sequential extraction, Tessier sequential extraction, Simultaneously Extracted
76 Metals/Acid-Volatile sulfide, X-ray absorption spectroscopy are widely recruited chemical
77 quantification methods (Thanh et al., 2015; van Hullebusch et al., 2016). These methods are easier
78 to implement but the maintenance of the anaerobic environment during sampling, the
79 unavailability of routine protocols and the low detection limits (resolution) are major setbacks in
80 quantifying TEs (van Hullebusch et al., 2016).
81 Over the last years, there have been attempts to implement important physicochemical processes
82 in AD and waste water treatment process models. ADM1 based models for sulphur reduction have
83 been defined and validated (Barrera et al., 2015; Federovich et al., 2003). Non-idealities arising
84 due to ionic strength of the system has been incorporated in ADM1 (Solon et al., 2015).
85 Interactions of phosphorous, sulphur and iron have been also taken into considered (Flores-Alsina
86 et al., 2016). Multiple mineral precipitation have also been estimated by saturation index
87 calculation which is based on ion activity correction (Mbamba et al., 2015). Precipitation of
88 calcium magnesium, inorganic phosphorous and nitrogen has been modelled as well. (Zhang et al.,
89 2015). Ion association-dissociation, precipitation-dissolution and liquid gas transfers also have
90 been described to establish a framework for incorporating physicochemical processes in waste
91 water treatment process models (Lizarralde et al., 2015). Although some of these studies included
92 Fe in the physicochemical model description, there has been a gap in modelling the effect of TE
4
93 on microbial kinetics, both in terms of inhibition and stimulation as well as the effect of
94 multimineral precipitation on TE bioavailability.
95 Hence, a mathematical model able to consider the intricate dynamics of TEs in AD systems can
96 be recruited to support experimental studies. The mathematical model can be specifically targeted
97 to study a particular process which affects TE bioavailability. The modeling results may confirm
98 hypotheses or ask new questions which close the loop in a symbiotic relationship between
99 experimentalists and mathematicians. Indeed, the more we learn about TE speciation from
100 laboratory studies the more components we can incorporate into our models. The mathematical
101 modeling of AD has been widely performed during the last decades (Batstone et al., 2015, 2006;
102 Gavala et al., 2003; Lauwers et al., 2013; Steyer et al., 1999). In this regard, the IWA benchmark
103 model, anaerobic digestion model # 1 (ADM1) is a widely accepted tool, which can be recruited
104 to study the underlying mechanistic principles of AD, to estimate system behavior in various
105 operating scenarios, to study the interaction between operational parameters and thus improve the
106 overall process performance. However, the ADM1 neglects a detailed description of the
107 physicochemical processes which would be essential to model the dynamics of TEs. Moreover, it
108 lacks a proper link between the TEs speciation and their effect on the biodegradation processes. A
109 model able to take into account all these factors could further be employed in the development of
110 a model-based predictive control system for TE dosing.
111 The main research question of this thesis was to model the role of TEs in AD reactors. As specified
112 above, TEs undergo multiple processes and the fate of TEs depends on the extent of these
113 biogeochemical reactions. This research was aimed at answering the following questions, among
114 others: How much do TEs contribute to the anaerobic digestor performance? Which speciation
115 pattern of TEs dominate the reactor at a given set of operating conditions? Is it possible for a TE
116 dosing to out-compete EDTA dosing and vice versa in terms of methane stimulation potential?
117 How influential is the adsorption of TEs and the binding site density of on the TE speciation
118 pattern?
119 More specifically, the purpose of this research was to study the dynamics of TEs in an AD system.
120 A mathematical model was developed in order to quantitatively relate the processes affecting TE
121 dynamics, assess the relationship between TE speciation and bioavailability and quantify TEs
122 effects on microbial growth and biogas production rate. The mathematical model was based on the
123 approach used in the ADM1. The model incorporated precipitation/dissolution, organic
124 complexation and adsorption/desorption in the form of three distinct sub-models. TE bio-uptake
125 and release from complex organic matter were considered as well. The extended models
126 maintained the structure of ADM1 for the biochemical reactions and considered the anaerobic
127 digester as a continuously stirred batch system with three phases, viz. solid, liquid and gas. Decay
128 of microorganism was considered as well. The differential mass balance equations governing the
129 TEs, substrates, products and bacterial groups dynamics involved in the AD processes were
130 synthesized and integrated using an original software developed on MATLAB platform. The
131 investigations presented in this thesis were based on a dynamic mathematical model of TE
5
132 dynamics in a batch anaerobic reactor. The model was mathematically analyzed based on various
133 numerical simulation scenarios.
134 1.1 Thesis Overview
135 Due to interdisciplinary nature of this thesis, the Background has been provided in Chapter 2, the
136 developed mathematical model framework in Chapter 3, and the Discussion in Chapter 4. Chapter
137 2 is intended as a summary of the research field and a necessary context for the specific problems
138 addressed in this research. Chapter 5 is dedicated to conclusion and future work. The scientific
139 contributions of this thesis are in the four publications (three accepted and one to be submitted)
140 that follow in the last part of the thesis. An overview of the four publications is reported below.
141 The first publication is in the form of a book chapter which provides a general background for the
142 mathematical modeling of TEs dynamics in AD. The second publication presents the precipitation
143 model. The third publication introduces a novel model which traces the TEs dynamics including
144 precipitation and complexation reactions. The fourth publication demonstrates a model for TE
145 speciation as it involves precipitation, complexation and adsorption processes in AD systems. All
146 the three models mentioned above are based on the ADM1 framework.
147 Publication I- Mathematical modelling of trace element dynamics in anaerobic digestion

148 environments (Book Chapter).
149 In Paper I the interdisciplinary background of the research field has been reviewed. This chapter
150 revisited the model characteristics describing the main processes that affect TE dynamics in AD
151 environment. The knowledge from the literature indicates that the effect and fate of TEs in AD
152 system can be modelled to predict the dynamic behaviour of TEs in AD. However, there is a need
153 to improve the currently available AD models to account for precipitation, adsorption, aquatic
154 complexation and bio-uptake of TE, to quantify their fate and effect. Consequently, considering
155 the multiplicity of processes that affects TE dynamics in AD systems as well as the importance of
156 these processes in defining the biology and physicochemistry of the AD system, the need for a TE
157 speciation module embedded within the ADM1 model has been recognized. This sets the priorities
158 in the efforts towards ‘mathematical model-based controlled TE dosing’ in AD systems.
159 Publication II- ADM1 based mathematical model of trace element precipitation/dissolution in
160 anaerobic digestion processes (Original research article).
161 In Paper II, the precipitation model integrates the existing ADM1 with liquid-solid precipitation
162 reactions. The dependence of biochemical processes on physicochemical reactions has been
163 improved by adding new inorganic components as a part of precipitation reactions taking place in
164 the AD. These inorganic components influence the biochemical processes by affecting the pH of
165 the system. Although total free heavy metal concentration is not directly related to the pH,
166 precipitation affects the total free trace elements present in an anaerobic digester. Taking these into
167 account, new chemical equilibrium association/dissociation, ion pairing, and precipitation
168 reactions have been added to ADM1. In particular, the developed sub-model takes into
6
169 consideration inorganic carbonate, inorganic phosphate and inorganic sulphate components in
170 liquid phase. The components of the three chemical systems (carbonate, phosphate and sulphate)
171 react, in the liquid phase, to form precipitates in solid phase. The solid phase processes are
172 governed by the Ksp values. Overall, the model tracks the dynamics of 62 state variables which
173 constitute the differential mass balance equations governing the trace elements, substrates,
174 products and bacterial groups involved in the anaerobic digestion of solid waste. Finally, in-silico
175 experiments were performed to demonstrate the model prediction capability. The modelling and
176 simulation results appeared in the form of a scientific publication dedicated to TE
177 precipitation/dissolution reaction in AD.
178 Publication III- ADM1 based mathematical model of trace element complexation in anaerobic
179 digestion processes. (Original research article)
180 In Paper III a complexation model focuses on TE aqueous complexation reactions taking place in
181 AD. In this model, AD model has been further modified in order to simulate the TE complexation,
182 precipitation and its effect on anaerobic digestion in terms of total methane production. In this
183 regard, the biodegradation processes, including hydrolysis, acidogenesis, acetogenesis,
184 methanogenesis, and the acid-base equilibrium reactions have been coupled to the TE
185 complexation, precipitation and redissolution processes and the effects on total methane
186 production have been evaluated. The incorporation of the precipitation and complexation reactions
187 in the physicochemical module led to the definition of new inorganic components (in addition to
188 precipitation sub-model) in the ADM1 framework. Subsequently, the charge balance has been
189 modified in order to consider the effects of additional components. The new model can predict:
190 complexation of TEs with EDTA and VFAs; and effect of change in initial calcium and magnesium
191 concentration in an in-silico anaerobic batch system. The results for the complexation sub-module
192 have been published in a peer-review scientific journal.
193 Publication IV- ADMI based mathematical model trace element speciation in anaerobic
194 digestion processes. (Original research article)
195 In Paper IV a sorption model has been incorporated in the already extended ADM1 framework to
196 simulate TEs speciation and its effect on the methane production. TEs sorption-desorption
197 reactions with biomass, inert and mineral precipitates have been incorporated in an extended
198 ADM1 model which considers TE precipitation/dissolution, complexation reactions as well as
199 biodegradation processes. The proposed model follows the structure of the original ADM1. The
200 incorporation of the sorption-desorption reactions required the definition of new adsorption-
201 desorption process rates in the ADM1 framework. The charge balance has been modified
202 accordingly to consider the effects of the new components. Further, in-silico experiments have
203 been performed to showcase the model prediction capability in terms of methane production and
204 TEs fractionation. The manuscript containing the results for the complete model considering
205 precipitation, complexation and sorption is to be submitted.
206
7
207 CHPATER 2. BACKGROUND
208 2.1 Anaerobic digestion and TEs
209 2.1.1 Anaerobic digestion
210 Anaerobic digestion (AD) is a multistage process in which a consortium of microorganisms acts
211 upon organic matter to produce biogas mainly consisting of carbon dioxide and methane (Appels
212 et al., 2011). AD has been used extensively in the recent past to treat the organic fraction of solid
213 wastes arising from a variety of waste streams (i.e municipal and industrial). Indeed, it is
214 noteworthy to mention that since the work of Young and McCarty (1969), AD has gained much
215 attention and has evolved as a reliable and cost efficient technology (Jimenez et al., 2015). The
216 rapid implementation of AD technologies around the globe is a proof of the advances in this field
217 and the transition towards a green and circular economy. The anaerobic production of methane is
218 a successful technology in terms of net energy yield, risk to environment, conversion efficiency
219 and capital investment.
220 AD is a collection of processes by which microorganisms, associated to each other by metabolic

221 reactions, break down biodegradable material in the absence of oxygen. The degradation process
222 engages direct and indirect relationship among microbial consortia, principally acetogenic bacteria
223 and methanogenic archaea, to degrade complex organic matter. The balance between microbial
224 communities and substrate concentrations is maintained by the cascading nature of the various
225 biochemical reactions involved in the biodegradation. The conversion of organic matter into
226 methane occurs in five main biochemical steps (Batstone et al., 2002; Gavala et al., 2003).
227 Disintegration of composites and their subsequent hydrolysis to soluble monomers are
228 extracellular steps. Hydrolysis involves the transformation of high molecular mass complex
229 organic compounds into lower molecular mass compounds opposite as carbon and energy source.
230 The next step named acidogenesis, involves the microbial conversion of the compounds resulting
231 from the hydrolysis step into identifiable lower-molecular-mass intermediate compounds. Volatile
232 fatty acids produced during acidogenesis are further utilized by a group of bacteria (acetogens) to
233 produce acetate in a sub-step known as acetogenesis. The last step called methanogenesis involves
234 the bacterial conversion of the intermediate compounds into simpler end-products such as methane
235 and carbon dioxide. Although several nomenclatures have been proposed for these steps, Speece
236 and McCarty (1962) named the hydrolysis, acidogenesis and acetogenesis as constant BOD phase
237 and the methanogenesis as reducing BOD phase, due to the reduction of BOD or COD related to
238 the methane formation. High organic loading rates and low sludge production are two of the main
239 advantages AD exhibits over the other biological treatment processes. But the feature emerging as
240 a major driver for the increased application of AD is the energy production. Indeed, this technology
241 has a net energy production and the biogas production can also replace fossil fuel sources and
242 therefore has a positive effect on greenhouse gas reduction. To summarize the main advantages of
243 using AD processes:
244  it is a simple and sustainable method;
8
245  apart from the primary function of energy production, it produces other value-added
246 products such as fertilizers, hydrogen and SCFA;
247  the mixed microbial community requires lower amount of nutrients due to the low growth
248 yield and the synchronizing nature;
249  The sludge production is lower as compared to aerobic treatments.
250 However, there are some aspects of AD technology, such as start-up time (Filbert and W, 2011;
251 Yang et al., 2019), process stability and installation costs (Batstone et al., 2012; Nguyen et al.,
252 2015; Paulo et al., 2015; Xu et al., 2015), that require considerable improvements in current
253 practice, which have been the key idea of this thesis. The major drawbacks deriving from the
254 operation of AD processes are usually overcome by employing latest process development and
255 optimization techniques, and modern reactor designs (Abdelgadir et al., 2014; Ahring et al., 1995;
256 Fabiyi et al., 2015; Fricke et al., 2007; Guendouz et al., 2010; Pind et al., 2003). Moreover,
257 mathematical models have been often recruited in process monitoring and control as they predict
258 the AD behavior and provide key insights into the complex network of biochemical and
259 physicochemical processes involved in AD systems (Fezzani and Ben Cheikh, 2009; Silva et al.,
260 2009; Tugtas et al., 2006; Xue et al., 2015).
261
262 2.1.2 Trace elements
263 IUPAC definition: Any element having an average concentration of less than about 100 parts per
264 million atoms (ppma) or less than 100 µg/g in a system is called a Trace Element (TE). In AD
265 systems, iron (Fe), cobalt (Co), nickel (Ni), selenium (Se), zinc (Zn), copper (Cu), molybdenum
266 (Mo), tungsten (W) have been reported as TEs (Glass and Orphan, 2012; Oleszkiewicz and
267 Sharma, 1990; Thanh et al., 2015; Uemura, 2010).
268
269 2.1.3 Importance of TEs in AD
270 In addition to macronutrients (carbon, nitrogen and phosphorus), TEs are essential micronutrients
271 which provide structural, physicochemical and synchronistic support during AD reactions (Figure
272 1) (Glass and Orphan, 2012). Microorganisms draw upon a constant supply of nutrients
273 (macro/micro) to concurrently carry out biodegradation of organic matter and maintain cell
274 homeostasis. Even so, ideal reactor conditions (pH, temeperature and redox potential) are
275 necessary (Batstone et al., 2012; Fermoso et al., 2015; Kumar et al., 2010; Lokshina et al., 2001;
276 McCartney and Oleszkiewicz, 1991; Thanh et al., 2015). Any disturbance to the availability of
277 these factors, including TEs, can affect the overall methane production. At the structural level, TEs
278 form the core of many enzymes involved in the conversion of organic biomass into methane. The
279 effect of TEs on AD performance has been extensively reported (Fermoso et al., 2008b; Gonzalez-
280 gil and Kleerebezem, 1999; Jiang, 2006; Mudhoo and Kumar, 2013; Roussel, 2012; Thanh et al.,
281 2015; Zandvoort et al., 2006). Co forms the core of the vitamin B12 which acts a cofactor for the
282 enzyme methylase involved in methane production (Stupperich and Kräutler, 1988; Mazumder,
283 Nishio, et al., 1987; Banerjee and Ragsdale, 2003). The coenzyme F430 which contains Ni is
9
284 essential for the functioning of methylcoenzyme M reductase, which is involved in the reduction
285 of coenzyme M to methane in methanogens (Finazzo et al., 2003; Thauer, 1998). Fe is involved in
286 the transport system of the methanogenic archaea for the conversion of CO2 to CH4, and functions
287 both as an electron acceptor and donor (Thanh et al., 2015). Similarly, Mn
288
289
290 Figure 1. Occurrence of TEs in methanogenesis step of AD process (adapted and modified from
291 (Glass and Orphan, 2012)).
292
10
293 acts as an electron acceptor in anaerobic processes. Zn, Cu, and Ni have all been found in
294 hydrogenase. TEs such as W and Mo are also found in enzymes such as formate dehydrogenase
295 involved in formate formation from propionates by propionate oxidizers. W and Mo can also be
296 considered as essential but their low concentrations in the sludge in comparison with the other
297 metals reduce their interests in this study (Fermoso et al., 2009; Oleszkiewicz and Sharma, 1990).
298 Table 1 enlists the TEs involved in AD and the related stimulating concentrations (Glass and
299 Orphan, 2012). At physicochemical level, TEs form multiple organic and inorganic species in an
300 anaerobic digestor. TEs have been widely reported to be precipitated with carbonate, sulphate and
301 phosphate groups present or produced during AD (Musvoto et al., 2000b). Inorganic and organic
302 complexation also constitute an important set of chemical reactions. In addition to precipitation
303 and complexation, the presence of large amount of sorbents (biomass, precipitates and inert) with
304 multiple binding sites (hydroxyl, carboxyl and phsophate) also drives TE adsorption onto their
305 surfaces (Fermoso et al., 2015; Roussel, 2012; Schwarz and Rittmann, 2007; Van Hullebusch et
306 al., 2006). One significant outcome of such physicochemical reactions, which has been widely
307 reported, is the formation of iron sulfide (FeS) precipitates which in turn reduce the hydrogen
308 sulfide content, thereby lifting the hydrogen sulfide inhibition of the AD process (Gustavsson,
309 2012; Hille et al., 2004; Kong et al., 2016; Oude Elferink et al., 1994; Shakeri et al., 2012).
310 Complexation of TEs with synthetic organic chelators (NTA/EDTA) has been reported as well
311 (Aquino and Stuckey, 2008; Hu et al., 2008; Schiavon Maia et al., 2017; Vintiloiu et al., 2013).
312 Although there have been many studies on adsorption of TEs in AD system (Fermoso et al., 2015),
313 the degree of effect of TEs adsorption in AD systems is still not clear. At microbial community
314 level, unavailability of TEs may break the synchrony between microbial species involved in AD
315 (Feng et al., 2010; Glass and Orphan, 2012; Worms et al., 2006; Yekta et al., 2017). The
316 methanogenesis step is the major point of dysfunction. Lower diversity and lower functional
317 redundancy of methanogenic groups than other microbial groups have been cited as the primary
318 reason (Demirel and Scherer, 2011; Wintsche et al., 2018). Consequently, TEs induce
319 predominance among methanogenic activity in AD systems. Due to such crucial physicochemical
320 effect and structural function of TEs, the analysis of limitation, stimulation, requirement and
321 supplementation of TEs in AD becomes crucial (Oleszkiewicz and Sharma, 1990; Thanh et al.,
322 2015; Zandvoort et al., 2006).
323
324
325 2.1.4 Requirement, inhibition, stimulation and supplementation strategy of TEs in AD.
326 Requirement:
327 The optimum TEs requirement depend on: 1) reactor configuration; 2) type of substrate used; 3)
328 temperature - mesophilic or thermophilic; 4) type of biomass: monoculture; co-culture, attached
329 or suspended (Thanh et al., 2015).
330 1. Starting with reactor configuration, the optimum TE requirement varies greatly with difference
331 in reactor configuration and mode of operation (Guendouz et al., 2010; Thanh et al., 2017a;
332 Wijekoon et al., 2011). This is primarily because of mass transfer limitations which vary in
11
333 different bioreactor configuration (Stuckey, 2012). For example, the amount of TE required in
334 a continuously stirred tank reactor is different from a reactor which is operating as an upflow
335 anaerobic sludge blanket (UASB) reactor (Gonzalez-Gil et al., 2003; Gonzalez-gil and
336 Kleerebezem, 1999; Zandvoort et al., 2006). In a UASB nutrient limitation is a major
337 denominator for stable production of methane. Similarly, submerged anaerobic membrane
338 reactors are different as compared to the previous two configurations due to the possibility of
339 decoupling hydraulic retention time and solid retention time (Thanh et al., 2017a).
340 2. Likewise, the use of different organic matrices reflects on TEs requirement (Khalid et al., 2011)
341 as they have different metal content. For an instance, the higher amount of sulfur and the TE
342 deficiency in wheat stillage render it unproductive for methane production (Schmidt et al., 2014;
343 Yekta et al., 2014a). Additionally, anthropogenic factors also lead to difference in TE content
344 of substrates. To exemplify, the organic fraction of municipal solid waste differ greatly, in terms
345 of TEs content, with the region of origin (He et al., 1992). TE content of organic fraction from
346 a region using majority of tomato in food will have different TE content as compared to a region
347 using majority of mozzarella in food.
348 3. Similarly, temperature is also a crucial factor because AD depends largely on enzyme produced
349 by microbial species. The methanogenic activity of mesophilic methanogens greatly vary from
350 the methanogenic activity of the thermophilic counterparts (Kim et al., 2003; Sh, 2010; Uemura,
351 2010. Consequently, the amount of TEs needed is different for each type of cultures. It has been
352 reported that mesophilic methanogens require higher amount of TEs as compared to
353 thermophilic methanogens (Uemura, 2010). Certainly, this is due to the increased metabolic
354 rate which in turn is associated to a higher TEs uptake. Moreover, by influencing the
355 precipitation reactions, temperature also affects the physicochemistry of the AD system
356 (Musvoto et al., 2000b).
357 4. In the same lines, TEs requirement for various types of biomass/culture are different. For
358 example, the methanogenic pathways operating in a single pure culture methanogen is different
359 from the methanogenic pathway operating in a biofilm (Gonzalez-gil and Kleerebezem, 1999).
360 In pure culture there is absence or no need of synchrony between different microbial groups
361 (Oleszkiewicz and Sharma, 1990; Thanh et al., 2015; Zeikus et al., 1975). Whereas, in mixed
362 culture systems, the extent of TEs requirement is different as the microbes operate in tandem
363 (Oleszkiewicz and Sharma, 1990). It has been reported that for CO2 reduction acetoclastic and
364 methylotropic pathways, the requirements of Fe, Ni and Zn are more or less similar (Glass and
365 Orphan, 2012).
366 Due to such a large amount of variability in the requirement of TEs, it is difficult to define a TE
367 dosing strategy which can enhance methane production. Nevertheless, the amount of TE required
368 for the maximum production of methane during anaerobic digestion is generally detected by one-
369 factor-one-time strategy from case to case basis. In such a strategy, the reactor is fed with only one
370 type of TEs and the amount of total methane produced is compared with methane produced when
371 no TEs are added. The amount of TE required for optimum production of methane has been widely
12
372 reported. It should be noted that the optimum concentration of TEs are generally depicted in terms
373 of total TEs concentration.
374
375 Figure 2. Deficiency, stimulation, optimal, inhibition and toxicity of TE concentration.
376
377 Inhibition:
378 The inhibition of methane production (Figure 2) due to the decrease in amount of TEs in AD
379 systems is evident and widely reported in literature (Chen et al., 2008; Fermoso et al., 2019, 2015;
380 Roussel et al., 2018; Thanh et al., 2015). A deficiency of TEs may lead to a shortage and or
381 improper synthesis of enzyme and cofactors associated with the cell maintenance as well as the
382 methane production pathway. Consequently, the methanogenic activity is hugely impaired which
383 further results in process instability and failure. When TEs are not present in the seed culture or
384 sludge, the TEs limitation can be easily induced by not feeding any TEs into the system for a
385 prolonged period of time. However, when seed culture contains some amount of TEs, it takes
386 longer time to induce TE limitation in the system (Zandvoort et al., 2006). Moreover, it is a
387 necessary prerequisite to ascertain the time period when TEs limitation starts in the system before
388 committing to any TE dosing strategy. Similarly, it was shown that Zn and Ni can inhibit methane
389 production and induce the failure of the anaerobic system (Fermoso et al., 2008c). In particular,
390 the decrease in methanogenic activity follows the decrease in Ni and Co concentrations in the
391 liquid phase after 110 days, which is accompanied by an accumulation of methanol (substrate) and
392 acetate. This led to the pH drop and subsequently the reactor failure. When the TEs are again made
393 available in the system, the reactor recovered with a 99% removal of methanol. Similarly, the AD
394 of wheat stillage resulted in the accumulation of VFAs due to a deficiency of Fe and Ni in the
395 system. To sum up, the basic premise for explaining TE limitation and associated reactor behavior
13
396 is based on accumulation of VFAs and the sensitivity of methanogens to TEs deficiency (Osuna
397 et al., 2003; Zhang et al., 2011; Zhang and Jahng, 2012). TE deficiency might also be due to various
398 physicochemical processes (precipitation, complexation and adsorption) undergoing inside an
399 anaerobic digester (Fermoso et al., 2019, 2015; Roussel, 2012; van Hullebusch et al., 2016; Yekta,
400 2014). Another line of reasoning might be the link between overall nutrient deficiency due to
401 organic shock loading and TEs deficiency. To reverse such limiting effects, TEs are generally
402 supplemented in the reactor.
403
404 Stimulation:
405 In order to overcome and or avoid TE inhibition effect (accumulation of VFAs), TEs are
406 supplemented in the AD system (Choong et al., 2016; Hickey et al., 1989; Mancini et al., 2018;
407 Osuna et al., 2003). Overall, TE supplementation is one of the methods to stimulate methanogenic
408 activity by increasing VFA utilization and thereby decreasing substrate accumulation (Zandvoort
409 et al., 2006). Generally, TE supplementation is carried out keeping in mind the stimulatory
410 concentration range of TEs for a particular set of factors as mentioned above. Table 1 shows the
411 concentration of TEs required to stimulate pure culture of methanogens. Apart from operational
412 parameters, TE stimulation concentration also depends on the TEs cell content (Jansen et al.,
413 2007). The content of TEs in cells is in the following order: Fe>Zn>Ni>Co>Mo>Cu>Mn (Scherer
414 and Sahm, 1981). Fe has been found to be the most abundant TE followed by Ni, Co and Mo. This
415 is due to the fact that methanogenesis is one of the highly TE rich enzymatic pathways known in
416 biology (Glass and Orphan, 2012). Stimulation of methanogenic activities is also subject to the
417 bioavailability of TEs in AD systems. The TEs bioavailability is a function of the bio-uptake and
418 physicochemical processes involved in AD. Some literature reviews have been reported on the
419 stimulatory effect of TE in AD system (Choong et al., 2016; Oleszkiewicz and Sharma, 1990;
420 Thanh et al., 2015; Zandvoort et al., 2006). The optimum stimulatory range of TEs is generally
421 reported in terms of total TE (van Hullebusch et al., 2016), as the quantification of free TE is
422 difficult, especially in AD systems. However, there have been some attempts to quantify TE
423 stimulation concentration range on the basis of free TEs (Jansen et al., 2007; Zhao et al., 2016).
424
14
425 Table 1. TEs involved in AD and the related stimulating concentrations. (Adapted from (Glass
426 and Orphan, 2012)).
Concentration
Metal Substrate Species
(µ M)
Fe H2/CO2 300-500 Methanospirillum hungatei
H2/CO2 >15 Methanococcus voltae
Acetate 100 Methanothrix soeggenii
Methanol 50 Methanosarcina bakeri
Ni H2/CO2 1 Methanobacterium thermoautotrophicum
H2/CO2 0.2 Methanococcus voltae
Acetate 2 Methanothrix soehngenii
Methanol 0.1 Methanosarcina barkeri
Co Acetate 2 Methanothrix soehngenii
Methanol 1 Methanosarcina barkeri
Mo Acetate 2 Methanothrix soehngenii
W H2/CO2 1 Methanocorpusculum parvum
Formate 0.5 Methanocorpusculum parvum
427
428 Supplementation strategy:
429 Which TEs are essential?
430 The TE content depends on the waste streams (Wood and Tchobanoglous, 1975). So, the first step
431 in a TE supplementation strategy would be to individuate the deficient TE as well as the rich TE.
432 For example, wheat stillage are rich sulfur compounds and deficient in most all TEs (Gustavsson
433 et al., 2013). Based on the deficiency of TEs a dose of single TEs can be used or a cocktail of TEs.
434 The primary factor to keep in mind is that the TE salt should rapidly solubilized to produce free
435 TEs which will be easily uptaken by the microbial groups and exhibit no toxic effect.
436 How should the metals be supplied?
437 Traditionally TEs have been supplemented in AD systems as either inorganic or organic TE
438 compounds (Choong et al., 2016; Fermoso et al., 2008b; Roussel, 2012; Vintiloiu et al., 2013).
439 Usually, TEs bound to chlorine (MeCl2) are used as inorganic compounds and TEs bound to
440 EDTA, NTA and citrate are used as organic compounds (Fermoso et al., 2008a). Chloride salts of
441 TEs are chosen due to the high solubility. Further, the increased chloride concentration does not
442 have any negative effects on the metabolic activity of the microbial groups. One disadvantage of
443 using chloride salt relies on the formation of TEs precipitates, particularly sulfide. TE from
444 chloride salts added to anaerobic digester immediately dissociate to react with sulfides and form
445 precipitates. As a remedy, organic bound TEs have been adopted. In case of organic bound TEs
446 such as EDTA, NTA and citrate, the TEs remain in the liquid phase for a long time as compared
15
447 to chloride. It has also been found that increasing chloride concentration in the AD system do not
448 necessarily affect the metabolic activity of microorganisms. Fermoso et al (2008) compared the
449 two forms of chemical dosing by supplementing CoCl2 and Co-EDTA in a UASB reactor. EDTA
450 bound Co remained in the solution for longer time while chloride bound Co precipitated (Fermoso
451 et al., 2008a). Such a behavior was also observed in another study (Bartacek et al., 2008). In the
452 study 90% of Co stayed in the solution when it was added in the form of Co-EDTA and a lower
453 amount of Co (30%) stayed in the solution when supplemented in the chloride form (CoCl2).
454 When should TE be dosed?
455 The timing of TE dosing is a necessary consideration in a TE supplementation strategy. This is to
456 avoid any unnecessary dosing of TEs which may lead to accumulation of TEs in the anaerobic
457 digester, eventually leading to a toxic effect on methanogens and hence reduced methanogenic
458 activity. The mode of dosing is also an important factor due to the variability in the AD operation
459 mode. It has been reported that, the effect of TEs supplementation is better studied when TEs are
460 dosed in continuous mode of reactor operation (Gonzalez-gil and Kleerebezem, 1999; Zandvoort
461 et al., 2006).
462 Further, any prior knowledge of the TEs bioavailability is helpful in designing a proper dosing
463 strategy, as the TEs dynamics changes with the course of the reactor operation (Roussel et al.,
464 2018; Zandvoort et al., 2006). For an instance, with the course of AD the amount of sulfide
465 precipitates increases (Fermoso et al., 2015; Flores-Alsina et al., 2016a; McCartney and
466 Oleszkiewicz, 1991). So, efforts should be made to limit the sulfide precipitation of TEs by adding
467 necessary chemicals or changing the reactor operational parameters. Furthermore, the dissolution
468 of precipitates may add to the pool of bioavailable TE, thereby increasing it beyond the threshold
469 limit of stimulation. In conclusion, a better grasp on the dynamics of TEs in AD systems will help
470 in determining the fate of TEs and hence an efficient and robust TE dosing strategy which includes
471 the timing of TE dosing.
472
473 2.1.5 Major processes affecting TEs dynamics in AD
474 The dynamics of TEs in AD systems is a function of the biochemical, physicochemical and
475 biouptake processes involved. The physicochemical processes are the dominant ones.
476 Physicochemical processes do not directly involve enzymes as catalyst (Batstone et al., 2012).
477 Precipitation-dissolution, aqueous complexation, acid-base and adsorption are the major groups of
478 physicochemical processes (Worms et al., 2006)
479
480 2.1.5.1 TE Precipitation and dissolution:
481 Mineral precipitation processes in AD systems are a result of the interactions among ions in the
482 liquid phase (Stumm and Morgan, 1996). Dissolution processes are the opposite of precipitation
483 processes in which precipitates in the solid phase break down into ions in the liquid phase
484 (Lizarralde et al., 2015a; K. N. Ohlinger et al., 1998). AD systems have multiple
485 precipitation/dissolution processes taking place continuously and concurrently with biological
486 reactions. This geochemical process (precipitation and dissolution) imparts a change in the
16
487 chemical composition of the liquid phase of the system (Callander and Barford, 1983a). The
488 change in the physicochemistry due to precipitation reactions is rapid because
489 precipitation/dissolution reactions are faster than biological reactions (Batstone et al., 2012). The
490 extent of precipitation/dissolution can be estimated by predicting saturation (Mbamba et al., 2015;
491 Stumm and Morgan, 1996). Saturation is a point in the concentration range of ionic constituents
492 of a mineral above which there can be no more dissolution of the mineral. Any further addition of
493 the ionic constituents will precipitate out. Such a point of saturation primarily depends on the
494 chemical nature of the constituent ions, temperature, pressure and concentration. The saturation
495 point is quantitatively calculated using the solubility product (Table 2) (Callander and Barford,
496 1983a; Stumm and Morgan, 1996).
497 It has been generally accepted that carbonate, sulfide and phosphate chemical systems dominate
498 the physicochemistry of AD reactors (Callander and Barford, 1983a; Fermoso et al., 2015; Flores-
499 Alsina et al., 2016b). For an instance, the thermodynamic equilibria associated with the formation
500 of TE sulfide precipitates are as follows:
501
𝐾𝑝1 [𝐻𝑆 − ][𝐻+ ]
𝐻2 𝑆 ↔ 𝐻𝑆 − + 𝐻 + 𝐾𝑝1 = 𝑝𝐾1 = 6.99 (2.1)
𝐻2 𝑆
𝐾𝑝2 [𝑆 ][𝐻+ ]
2−
𝐻𝑆 − ↔ 𝑆 2− + 𝐻 + 𝐾𝑝2 = 𝑝𝐾2 = 17.4 (2.2)
[𝐻𝑆 − ]
𝑇𝐸 2+ + 𝑆 2− ↔ 𝑇𝐸𝑆(𝑆) (2.3)
𝑇𝐸 2+ + 𝐻𝑆 − ↔ 𝑇𝐸𝑆(𝑆) + 𝐻+ (2.4)
502
503 Table 2. Solubility values (pKsp) for the three TEs under study.
TE Mineral pKsp Reference

CoCO3 10 (Callander and Barford, 1983a)
Co Co3(PO4)2 34.7 (Callander and Barford, 1983a)
CoS 25.5 (Ball, 1991)
FeCO3 10.7 (Ball, 1991)
18.4 Hogfeldt (1982)
Fe FeS
22.4 Jong and Parry (2009)
Pyrite- FeS2 18.5 (Ball, 1991)
NiCO3 6.9 (Ball, 1991)
Ni Ni3(PO4)2 31.3 (Ball, 1991)
NiS 20.7 (Callander and Barford, 1983a)
504
505 2.1.5.2 TE Complexation:
506 Aqueous complexation processes are those processes in which an organic ligand (natural or
507 synthetic) reacts with TEs to form a coordination complex or TE-chelate complex (Aquino and
17
508 Stuckey, 2007; Callander and Barford, 1983b). AD is a process which is characterized by high
509 organic matter content. Biomass, humic substances and degradation intermediates (such as VFAs
510 and alcohols) constitute the organic matter and form the pool of organic ligand (Aquino and
511 Stuckey, 2004). Moreover, synthetic organic ligands such as EDTA (Figure 3) and NTA are
512 occasionally added in order to improve biogas production (Aquino and Stuckey, 2007; Fermoso et
513 al., 2008a; Hu et al., 2008; Vintiloiu et al., 2013). Positively charged TEs ions have affinity for
514 these organic ligands due to the presence of specific functional groups (Figure 4) and their position
515 on the ligand molecule (Stumm and Morgan, 1996). The interaction of organic ligands with TEs
516 can be explained by the virtue of crystal field theory and molecular orbital theory (Stumm and
517 Morgan, 1996).
518
519 Table 3. Stability constants values (pKsp) for synthetic and organic chelators (adapted from
520 (Callander and Barford, 1983a))
Log Stability Constants

3+ 2+
Chelator Fe Cu Zn2+ Co2+ Fe2+ Mn2+ Ca2+ Mg2+
EDDHA 33.9 15 9.3 - 14.3 - 7.2 2.9
CDTA 27.5 21.3 18.5 18.9 16.3 14.7 12.5 10.3
EDTA 25.1 18.3 16.3 16.2 14.3 13.6 10.7 8.7
NTA 15.9 12.3 10.5 10.6 8.8 7.4 6.4 7.0
Histidine - 18.3 12.9 13.9 9.3 7.7 - -
Glycine - 15.2 9.5 8.9 7.8 4.7 -
Citric acid 11.4 5.9 5.0 5.0 4.4 3.7 3.6 3.3
Polyphosphate 6.5 5.5 6.0 3.0 3.0 5.5 3.0 3.2
521
522 Carboxyl, amino, sulphur and phenolic groups are found in common chelating organic ligands.
523 The potential to form complexes depends to a large extent on the capacity of these functional
524 groups to hold the TE ion (Aquino and Stuckey, 2007; Callander and Barford, 1983a; Stumm and
525 Morgan, 1996; Vintiloiu et al., 2013). Such potential is quantitatively expressed as the stability
526 constant (Table 3) of a particular complex formed. For example, organic ligands such as citric
527 acid and glycine form less stable complexes (Log stability constant: ~ 4-8) with TEs as compare
528 to Histidine, EDTA and NTA (Log stability constant: ~15-20). The major complexes formed
529 during AD processes are VFA-TE complexes, as VFAs are the most abundant organic ligands and
530 contain a suitable functional group (carboxyl) for complex formation (Lizarralde et al., 2015b). If
531 EDTA/NTA are added to the anaerobic digester, EDTA/NTA-TE complexes would be the most
532 stable organic ligand complexes present in the system (Aquino and Stuckey, 2007; Fermoso et al.,
533 2008a; Vintiloiu et al., 2013).
534
18
2+
Me
Ethelyenediaminetetracetic acid (EDTA) EDTA-Metal Complex

EDTA addition improves the bioavailability of TEs: Penetrate granules easily
EDTA supplement inhibit methanogenic activity: Penetrates granules easily but strong binding of EDTA
to TEs
535
536 Figure 3. Chelation reaction of TE with Ethylenediaminetetraacetic acid (EDTA) to form EDTA-
537 TE complex.
538
539
Three organic ligands constitute:

1) Major degradation products in AD: VFAs
2) Humic substances: Fulvic acid & humic acid
(Acetate) 3) Synthetic chelators: EDTA & NTA
Carboxylic, phenol and thiol group are crucial in
complexation
(Propionate) The extent of Complex formation is strongly pH

dependent. Two reasons:
The chemical reactions are pH dependent:
2𝑅 − 𝐶𝑂𝑂𝐻 + 𝑇𝐸 2+ ↔ (𝑅𝑂𝑂)2 𝑇𝐸 + 2𝐻+ , i.e.,
H+ ions are “exchanged” in the reaction
(Butyrate) The electrostatic effect:
For organic substances, as in case of other surfaces, a
change of the opposite sign enhances complex formation
through electrostatic attraction
(Sulphur containing amino acid)

540
541
542 Figure 4. Major organic ligands in anaerobic system with functional groups.
543
544
545
19
546 2.1.5.3 TE Adsorption
547
548 TE adsorption processes refer to the processes in which TEs from the liquid phase interact with
549 sorbing components carrying functional groups capable of interacting with TE ions (Stumm and
550 Morgan, 1996). Carboxyl, phosphate and hydroxyl groups constitute the majority of such
551 functional groups. The extent of adsorption processes in AD systems is due to two major reasons:
552 1) the availability of large amount of surface molecules which provide necessary functional
553 groups, and 2) the need of the microorganisms to uptake TEs from the local aqueous environment
554 (Konhauser, 2007). Microorganism growth depends on the uptake and the transport of essential
555 TEs from their local milieu. As a result, there is a necessity to maintain a hydrophilic interface
556 (Konhauser, 2007). The latter is achieved by having a large outer surface area and projecting
557 anionic organic ligands on the cell surface. These anionic organic ligands become point of
558 adsorption-desorption for a facilitated diffusion of TEs (i.e. Fe, Co, and Ni) and further TEs
559 internalization (Worms et al., 2006). It has been reported that adsorption-desorption processes of
560 TEs on the cell surface and the available particulate matter (i.e. mineral precipitates and inert)
561 strongly affect the availability of TEs in an AD system (Fermoso et al., 2015; Hullebusch et al.,
562 2016).
563 The sorption/desorption of TEs with mineral precipitate, biomass and inert material are considered
564 kinetically controlled and are expressed as follows:
𝑘1
𝑇𝐸+ ≡ 𝑋 → (≡ 𝑋 − 𝑇𝐸) (2.5)
𝑘−1
(≡ 𝑋 − 𝑇𝐸 ) → 𝑇𝐸 + ≡ 𝑋 (2.6)
565 where, 𝑘1 is the (≡ 𝑋 − 𝑇𝐸 ) formation rate constant; 𝑘−1 is the dissociation constant.
566
567
568 2.1.5.4 TE Biouptake
569 Biouptake of TE by different microorganisms in AD processes depends on the availability of TE
570 in the desired form or species and the prevailing reactor conditions (Jansen et al., 2007; Worms et
571 al., 2006; Yekta et al., 2016). More importantly, physicochemical processes, discussed above,
572 determine in which form the TE will exist in the reactor at a particular given time. TEs are
573 transported into the microbial cells by different mechanisms (Waldron and Robinson, 2009).
574 Majority of TEs are transported into the microbial cell as free ions by passive diffusion. At the
575 same time, active transporters have been also reported for transport of TEs (Worms et al., 2006).
576 For example, P-type ATPases, are responsible for detoxification and internalization of TEs. Once
577 internalized the biological response to TEs will affect the physicochemical composition of the
578 liquid phase of the anaerobic digester (such as pH change and change in the overall charge of the
579 system including surface charge), which in return affects the internalization flux. Thus, a large
580 number of parameters will affect the biouptake of TEs and hence the stimulation or inhibition of
581 methanogenic activity. In general, the biouptake of TEs is explained by simple models such as
582 Free Ion Activity Model (FIAM) and Biotic Ligand Model (BLM) (Fermoso et al., 2015;
20
583 Hullebusch et al., 2016; Jansen, 2004; Worms et al., 2006). FIAM describes the biological effect
584 of TEs as a factor of free ion activity, whereas, BLM connects biological effects as a function of
585 the equilibrium amount of TE adsorbed on to as well as uptaken into the cell. These models are
586 successful at predicting bioavailability in a steady state condition (Worms et al., 2006). This is
587 because dynamics of the TE species is not an issue when equilibrium is maintained throughout the
588 system. Under such pseudo-equilibrium conditions, the internalization flux of TEs can be related
589 to the metal concentration in the solution (Worms et al., 2006). However, these models are grossly
590 simple as they consider the system to be at equilibrium, which is generally not the case.
591
592 2.1.6 Methods to assess speciation of TEs in AD
593 In AD systems, TEs are spread across the solid and liquid phase. TEs can be found as free ions,
594 precipitates, bound to organic complexes and adsorbed on biomass, inert and precipitates.
595 Consequently, the methods employed to assess the speciation of TEs vary depending on the type
596 of sample, TE being analyzed and the suitability of the outcome to the desired objective. The
597 methods used to assess TE speciation in AD systems can be broadly grouped in three different
598 categories: 1) Direct methods (spectroscopic); 2) Indirect methods (chemical/ voltammetry); and
599 3) mathematical modeling.
600 Direct determination of TE speciation in samples from an anaerobic digester can be achieved by
601 various instrumental methods, such as, TEM/SEM-EDX (Hullebusch et al., 2016) , XANES
602 (Shakeri et al., 2012), EXAFS, STXM, XPS and EPR (Hullebusch et al., 2016). The primary
603 principle behind these spectroscopic methods is that each element has unique atomic structure
604 which exhibits unique set of peaks in the emission spectrum. Indirect determination of TE
605 speciation is based on the sequential extraction methods and voltammetry based methods (Dong
606 et al., 2013; Hullebusch et al., 2016; Thanh et al., 2015). Tessier sequential extraction (Tessier et
607 al., 1979), modified Tessier sequential extraction (Ortner et al., 2015), Accelerated BCR extraction
608 scheme are widely used chemical extraction methods used to quantify the total concentration of
609 TEs in a given sample (Filgueiras et al., 2002). These methods segregate the TE content of the
610 given sample into different portions which are soluble in a particular reagent under specific
611 experimental condition. The chemical reagents selected for the extraction method always target an
612 already supposed mineral phase.
613 Mathematical models based on the thermodynamic approach have been employed in few studies
614 to capture the TE speciation and bioavailability in AD systems. In one of the earliest attempt to
615 predict TE speciation in mathematical terms (Callander and Barford, 1983a) predicted the
616 dominance of sulfide minerals in the AD system. Later a chemical speciation model was used to
617 evaluate the effects of biomass and microbial chelators on TEs speciation which was coupled with
618 sequential extraction measurements (Aquino and Stuckey, 2007). Furthermore, TE biouptake was
619 linked to TE speciation modeling using pure culture in a controlled experimental condition. The
620 model could predict the biological effect of excess and deficiency of TEs in the given reactor
621 conditions. Recently, chemical equilibrium modeling tools such as Visual MINTEQ have been
622 successfully used to establish the relationship between speciation of sulfur and TE (Yekta et al.,
21
623 2014b) which concluded a strong interrelationship between S and Fe concentrations based on the
624 comparison of model predicted values and sequential extraction methods. To predict interactions
625 and estimate the overall speciation pattern of TEs across all the thermodynamic phases is possible
626 in case of mathematical models, which is difficult to achieve with experimental methods.
627
628 2.2 Mathematical modeling of anaerobic digestion process
629 2.2.1 Anaerobic digestion model No 1
630 ADM1 is an integrated basis for modelling AD process (Batstone et al., 2006, 2002). It was
631 formulated by the IWA Task group on mathematical modelling of AD to strengthen and merge
632 varied arrays of standalone models available (Gavala et al., 2003). The key assumptions of this
633 well-accepted model are highlighted below. The principal conversion processes taken into
634 consideration can be divided into two types, viz, biochemical processes and physicochemical
635 processes. The carbon flow is depicted in Figure 5. Biochemical processes are normally carried
636 out by intra-or extracellular enzymes acting on a pool of accessible and degradable organic matter
637 or waste.
638
639 Figure 5: Biochemical conversions in ADM1.
22
640 In ADM1, the particulate organic matter 𝑋𝑐 undergoes the disintegration process which produces
641 carbohydrates 𝑋𝑐ℎ , proteins 𝑋𝑝𝑟 , lipids 𝑋𝑙𝑖 and inerts 𝑋𝐼 . Inerts are modeled as a biochemically
642 inactive component. In the next step, the large non-soluble particulates, carbohydrates, proteins,
643 and lipids undergo hydrolysis which results in the release of monosaccharides 𝑆𝑠𝑢 , aminoacids 𝑆𝑎𝑎
644 and long chain fatty acids 𝑆𝑓𝑎 respectively. Subsequently, these soluble monomers are utilized by
645 the respective degrading microorganisms, sugar degraders 𝑋𝑠𝑢 , amino acid degraders 𝑋𝑎𝑎 and
646 LCFA degraders 𝑋𝑓𝑎 in the acidogenesis step to produce valerate 𝑆𝑣𝑎 . butyrate 𝑆𝑏𝑢 and propionate
647 𝑆𝑝𝑟𝑜 . Valerate, butyrate and propionate are further degraded by valerate butyrate degraders 𝑋𝑐4
648 and propionate degraders 𝑋𝑝𝑟𝑜 to produce acetate 𝑆𝑎𝑐 . Hydrogen is a by-product of this step. The
649 final step of methanogenesis occurs via two pathways: 1) cleavage of acetate by acetoclastic
650 bacteria 𝑋𝑎𝑐 ; and 2) reduction of CO2 and H2 by hydrogentrophic methanogens 𝑋ℎ2 . Apart from
651 biochemical transformations, physico-chemical processes such as liquid-liquid transformations
652 (ion association/dissociation) and gas-liquid transfers have been considered in ADM1. Six acid-
653 base pairs have been included along with CO2, CH4 and H2 gas-liquid transfer processes.
654
655 2.2.2 Extensions of ADM1
656 Ever since the ADM1 has been released, there has been a growing demand to include other
657 necessary processes in the ADM1 framework. This has been acknowledged by the Task Group,
658 which segregate these demands in three important heads: 1) phosphorus metabolism; 2) sulfur
659 metabolism; and 3) precipitation/dissolution of minerals. Additionally, there have been attempts
660 to include other important biochemical and physicochemical processes in the ADM1 framework.
661 A list of such model extensions and their feature has been reviewed and depicted in Table 4. This
662 list does not include ADM1 extensions on mineral precipitation, which have been described in the
663 appropriate section.
664 Table 4. Various important extensions of ADM1 (adapted and modified from (Lauwers et al.,
665 2013))
Author Contributions
Batstone et al., 2003 First extention to consider precipitation of mineral CaCO3.

Batstone et al., 2003 Acetogenesis of straight and branched chain VFAs. A common kinetic
parameter was used for branched chain VFA degradation.
Fedorovich et al. 2003 First significant model extension in the biochemical framework of ADM1
relating to sulphate reduction in AD environment.
Wett et al. 2006 Inclusion of nitrogen content of inerts.
Parker et al. 2006 Production of sulfur related odourants in AD systems.
Tugtas et al. 2006 Oxidation of inorganic nitrogen species.
Rodriguez et al. 2006 Effect of pH and hydrogen on carbohydrate breakdown.
Peiris et al. 2006 Alternate biochemical pathway for Lactate and ethanol metabolism
Fountoulakis et al., 2006 Introduction of sorption and desorption kinetics in ADM1
23
Kleerebezem et al. 2006 Improved power expression for hydrogen inhibition.
Zaher et al. 2009 Cyanide hydrolysis, inhibition and uptake.
Shimada et al., 2007 Storage of excess compounds by microorganism
Myint et al., 2007 Surface based hydrolysis kinetics of cellulose copmounds
Ramirez et al. 2009 Contois kinetics for the first two steps of AD as a substitute of first order
kinetics, Hill function for NH3 inhibition.
Yasui et al. 2008 Classifying substrate based on biodegradation potential
Penumathsa et al. 2008 Improved stoichiometry for lactate biodegradation
Fezzani et al., 2009 Inhibition of aceticlastic pathway by total butyrate, propionate and
valerate
Fezzani et al., 2009 Phenol biodegradation by modifying hydrolysis reactions.
Palatsi et al. 2010 Long chain fatty acid effect on acetogenesis
Bollon et al. 2011 Lumping of disintegration, hydrolysis and acidogenesis
Mairet et al. 2011 Contois based hydrolysis model for improved prediction.
Esposito et al. 2011 Improved surface related kinetic model for hydrolysis
Hierholtzer et al. 2012 Na as a new variable in the ADM1 framework with limitations on
methanogenesis.
Ernesto et al. 2015 Sulfate reduction for high sulphur content wastewater
Garcia et al., 2013 Microbial population dynamics. ADM1 used with microorganisms
factorial designs.
Solon et al. 2015 Activity correction to estimate ionic species in AD systems.
Santiago et al. 2015 Anaerobic co-digestion of fermentable substrates
Zhang et al. 2015 Improved physicochemical processes to include precipitation of minerals
Xavier et al. 2016 Phosphorous, sulfur and iron interaction for dynamic simulation of AD.
666
667 2.3 Modeling TE-associated physiochemical reactions in AD
668 2.3.1 Modeling precipitation processes
669 There are very few studies which focus on modeling TE precipitation processes in AD systems.
670 However, in wastewater treatment field, two broad modeling techniques have been widely used to
671 model the precipitation process (Kazadi Mbamba et al., 2015b). These modeling strategies are
672 restricted either to thermodynamic equilibrium approach or to a kinetic approach, while some of
673 the studies adopted a combination of the two kinetic approaches (Barat et al., 2011).
674
675 Equilibrium composition method
676 The thermodynamic equilibrium method for estimating precipitation is based on a well-established
677 multicomponent approach (Allison et al., 1991; Morel and Morgan, 1972; Rittmann et al., 2002).
678 In this approach, the system is described as a collection of components and species. The
679 components and species constitute an acid-base system for which the total proton condition is
680 calculated. The acid-base reactions are assumed to be in equilibrium. The system of equations is
681 solved iteratively by the Newton-Raphson algorithm and making use of the equilibrium constants
682 and the law of mass action (Kazadi Mbamba et al., 2015b; Solon et al., 2015). This method has
24
683 been extensively used in predicting the physicochemistry of wastewater systems. Indeed, there
684 have been multiple efforts to model precipitation processes using the equilibrium constant
685 approach with a variation in the number and types of components and species as required.
686 Moreover, most of geochemical modelling softwares rely heavily on this method, which is
687 essentially based on chemical equilibrium calculations and thermodynamic databases. For
688 instance, the chemical equilibrium composition has been used to quantify the amount of precipitate
689 formed when arbitrary amount of N, Mg and P were added to the wastewater system.
690 Concentration ratios were denoted in equilibrium reactions which is followed by a quadratic
691 equation-based approximation coefficient (Scott et al., 1991; Wrigley et al., 1992). Further,
692 phosphate estimation was carried out using the charge balance method which was extended to the
693 remaining components of the system. In (K.N. Ohlinger et al., 1998) the incorporation of ionic
694 strength, ionic activities and complexation effects on ion speciation was studied. Here precipitation
695 reactions were modeled along with complexation reactions. Further, a chemical equilibrium
696 calculation tool, MINEQL+ was employed to calculate the theoretical equilibrium concentration
697 for total Mg and NH3. An objective function was obtained based on MINEQL+, the measured
698 theoretical concentrations and variations from lab tests. The minimization of the objective function
699 was carried out to arrive at a solubility constant of 𝑝𝐾𝑆𝑃 = 13.26 for struvite.
700
701 The thermodynamic approach to model precipitation generally does not provide information about
702 the dynamics of the species or components taking part in the precipitation reaction. Additionally,
703 studies based on the thermodynamic approach do not take into account the biological processes
704 taking place in the system. Hence, the influence of the biological processes, in terms of pH, on
705 precipitation and vice-versa has been side-lined in this category of models.
706
707 Equilibrium composition + Kinetic method
708 In this method, some of the physicochemical reactions are solved by adopting the equilibrium
709 approach whereas other reactions are kinetically solved. Such a choice of assigning a specific
710 approach to a particular set of reactions largely depends on the process being modelled and the
711 precipitation reaction under study. As a rule of thumb, protonation/deprotonation reactions are
712 modelled as equilibrium based while slower precipitation reactions are kinetically modelled. For
713 example, a calcium phosphate precipitation model was incorporated into the activated sludge
714 model # 2 (Barat et al., 2011). The model considered two types of reactions for precipitation of
715 calcium phosphate. Equilibrium approach was employed to model the aqueous phase reactions and
716 ion pairing reactions which are generally fast. Slower aqueous to solid phase reactions were
717 kinetically modeled. A surface based kinetic was considered for precipitation reactions (solid
718 phase). Using both equilibrium and kinetic approach reduces the number of differential equations
719 needed to describe the chemical system, thereby increasing the computational efficiency.
720
721
722
25
723 Kinetic method
724 In the kinetic method, all the physicochemical reactions in the system are modeled as kinetic
725 reactions in a dynamic state differential equation set. These set of differential equations are solved
726 over time to predict the dynamic behavior of the components. Precipitation models based on
727 kinetic methods have been reported widely. For example, based on the predictions of a kinetic
728 model and experimental results, it was demonstrated that a reversible precipitation process exists
729 in activated sludge containing high amounts of dissolved calcium and phosphorous (Maurer and
730 Boller, 1999). In a significant piece of contribution, (Musvoto et al., 2000) modeled a three phases
731 carbonate system including precipitation/dissolution and gas stripping processes. The focus of the
732 study was on the formation of struvite, newberyite, amorphous calcium and magnesium carbonate.
733 In a noteworthy attempt, calcite precipitation was modelled, and the model simulations were
734 validated with titration and constant composition experiments (Mbamba et al., 2015a). Davies
735 equation with temperature correction was used to obtain the activity coefficients. pH was
736 calculated using the proton balance method. Precipitation and dissolution reactions were
737 considered based on the saturation index 𝑆𝐼 approach. Recently, (Lizarralde et al., 2015a)
738 developed a systematic general methodology to include physicochemical processes in multiphase
739 wastewater treatment models under a plant wide modelling framework. The prescribed
740 methodology included: a) the definition of the model which enlists various components and
741 transformations; b) the mass transport definition for a multiphase model; and c) the numerical
742 solution procedure. The model used precipitation and dissolution reactions to describe solid phase
743 physicochemical transformations. The aqueous reactions were based on the ion activity product.
744 The major precipitates considered were 𝐶𝑎𝐶𝑂3 , 𝑀𝑔𝐶𝑂3 , 𝐶𝑎3 (𝑃𝑂4 )2 , struvite, k-struvite and
745 newberyite.
746
747 Precipitation modeling within ADM1 framework:
748 The original ADM1, soon after its release, received criticism and modifications for errors and
749 omissions (Batstone et al., 2006; Kleerebezem and van Loosdrecht, 2006). Apart from the
750 complexity of the model structure and the uncertainty of parameter values, requests were received
751 to incorporate other anaerobic processes such as sulfate reduction, phosphorous metabolism and
752 physicochemical processes (Batstone et al., 2006; Kleerebezem and van Loosdrecht, 2006).
753 Mineral precipitation was the center of these arguments. Subsequently, the authors admitted the
754 omissions and a “Generalized Physicochemical Modelling Framework” was recently developed
755 in (Mbamba et al., 2015a). Multiple studies have been carried out to incorporate new biochemical
756 and physicochemical reactions into the framework of ADM1 (Federovich et al., 2003; Galí et al.,
757 2009; Parker and Wu, 2006; Peiris et al., 2006)(Batstone and Keller, 2003; Flores-Alsina et al.,
758 2016; Zhang et al., 2015). The most immediate modification to the ADM1, soon after its release,
759 was carried to include Calcium carbonate 𝐶𝑎𝐶𝑂3 precipitation (Batstone and Keller, 2003). Along
760 with simple first order kinetic rate equations for 𝐻𝐶𝑂3− /𝐶𝑂32− acid base reactions, three additional
761 state variables 𝑆𝐶𝑂32− , 𝑆𝐶𝑎 and 𝑆𝐶𝑎𝐶𝑂3 were introduced in ADM1 to predict 𝐶𝑎𝐶𝑂3 precipitation. A
762 more detailed physicochemical framework was added to ADM1 by (Zhang et al., 2015) to simulate
26
763 the dynamics of calcium carbonate 𝐶𝑎𝐶𝑂3 , magnesium carbonate 𝑀𝑔𝐶𝑂3, Struvite 𝑀𝑔𝑁𝐻4 𝑃𝑂3 ,
764 magnesium phosphate 𝑀𝑔𝐻𝑃𝑂4 and tricalcium phosphate 𝐶𝑎3 (𝑃𝑂4 )2 . The model was validated
765 on data available in literature. Recently, an extended ADM1 was developed to take into account
766 the interactions between phosphorous, iron and sulfur, as well as the related precipitation
767 processes. The model extensions were based on the aqueous chemistry model introduced in (Solon
768 et al., 2015) and the multiple mineral precipitation model of (Kazadi Mbamba et al., 2015a;
769 Mbamba et al., 2015). The aqueous chemistry model introduced ionic strength correction for
770 various physicochemical reactions taking place in the system. The mineral precipitation model
771 focused on implementing a saturation index based reversible mechanism for precipitation. The
772 model primarily focused on phosphorous mineral precipitation in addition to carbonate and sulfide
773 mineral precipitation.
774
775 2.3.2 Modeling TE complexation processes in AD
776 AD of complex organic matter produces soluble and particulate intermediates. In addition to the
777 intermediates, microbial biomass and humic acid related substances are present in the AD system.
778 These intermediates are excellent carriers of various chemical functional groups. For example,
779 VFAs are carrier of carboxylic acid groups and particulate cysteine containing particulate proteins
780 harbor thiol groups (Yekta et al., 2014a, 2014b). The soluble intermediates such as VFAs react
781 interacts with positively charged TE ions and form complexation species. On other hand, in the
782 course of AD operation various chemical agents, apart from TEs are dosed into the reactor. These
783 chemical agents are generally added to improve the bioavailability of TEs. EDTA, NTA and EDDS
784 are one of those groups of chemical agents added. These synthetic chelators also take part in
785 complexation reactions with TEs. Overall, there are three possible types of complexation agents
786 in the AD system in the liquid phase: a) soluble organic intermediates of AD (primarily VFA); b)
787 Synthetic agents for TE chelation (EDTA,NTA, and EDDS) and c) soil organic acids. Modeling
788 complexation reactions in the ADM1 framework is rare. However, inspiration can be taken from
789 standalone biodegradation models (Willet and Rittmann, 2003).
790 A kinetic approach can be recruited to model the association and dissociation of TE complexes.
791 The negatively charged organic ligands react with positively charged TEs to form complexes. New
792 components and species and model equations required to describe complexation reactions in
793 ADM1 can be added in conjugation to precipitation/dissolution processes can be added to the
794 ADM1 framework. All the components and the species are formed in the aqueous phase. State
795 variables for acid/base system of the organic ligand and complexes should be incorporated as well.
796 The mathematical formulations of the association and dissociation of TE complexes can be defined
797 as follows:
𝑘1 𝑥+𝑦
𝑀𝑒 𝑥 + 𝐿𝑦 → [𝑀𝑒𝐿] (2.7)
𝑘−1
[𝑀𝑒𝐿]𝑥+𝑦 → 𝑀𝑒 𝑥 + 𝐿𝑦 (2.8)
27
798 where k1 is the [𝑀𝑒𝐿] 𝑥+𝑦 formation rate constant in [M-1h-1]; k-1 is the dissociation constant in [h-
1
799 ]. The individual association (7) and dissociation (8) reactions may be worked out as:
𝐾[𝑀𝑒𝐿]𝑥+𝑦
𝑀𝑒 𝑥 + 𝐿𝑦 ↔ [𝑀𝑒𝐿]𝑥+𝑦 (2.9)
800 where 𝐾[𝑀𝑒𝐿]𝑥+𝑦 is the equilibrium constant for [𝑀𝑒𝐿] 𝑥+𝑦 in M-1. The rate of change for the
801 complex may be written as:
𝑑[𝑀𝑒𝐿] 𝑥+𝑦
= 𝑘1 [𝑀𝑒 𝑥 ][𝐿𝑦 ] − 𝑘−1 [𝑀𝑒𝐿] 𝑥+𝑦 (2.10)
𝑑𝑡
802 where 𝑀𝑒 𝑥 , 𝐿𝑦 and [𝑀𝑒𝐿] 𝑥+𝑦 are the dynamic state variables for the free TE concentration,
803 organic ligand and TE-organic complex respectively. The formation rate constant, 𝑘1 , for all the
804 species and the dissociation rate constant, 𝑘−1 , can be easily quantified from stability constant as
805 follows:
𝑘1
𝐾[𝑀𝑒𝐿]𝑥+𝑦 = (2.11)
𝑘−1
806 The values for 𝑘1 and 𝐾 are generally derived from data available in standard literature, which are
807 based on the basic premise of electrostatic interaction involving cations and water molecules. The
808 loss of water from the outer sphere during complexation reaction is directly proportional to the
809 electrostatic interactions between the cations and water molecules. The rate of loss of water from
810 such a complex decreases with increase in charge on cation. It has been calculated for many metal
811 cations. For an instance, the overall rate formation constant (𝑘1 ) for Fe-EDTA species falls in the
812 order of 1010 M -1 h-1 (Willet and Rittmann, 2003). The value of the reverse rate constant (𝑘−1 )
813 can be calculated from the stability constant (𝐾) for the particular species.
814 Several studies have reported the essential role of TE-EDTA complexes in the recovery, stability
815 and efficiency of AD process, specifically when the primary reason of process inhibition is mineral
816 precipitation. Although, fate of EDTA in the environment is under scrutiny and new green
817 chelating agents (EDDS) are being tested in AD systems, it should not restrict the AD
818 mathematical modeling community to develop mechanistic models. Because across different
819 organic/synthetic chelating agents the underlying mechanisms of complexation reactions are
820 uniform.
821 2.3.3 Modeling TE Adsorption processes in AD
822 TEs undergo adsorption/desorption processes on the solid phase of AD. There are numerous
823 surfaces in AD which harbor active sites (physically/chemically) for adsorption/desorption of TEs.
824 TE undergo physical adsorption with the physically active site where as the adsorption/desorption
825 taking place with chemically active sites is called chemisorption. Here we will focus on the
826 physical adsorption. The principle of physical adsorption is the difference in net charge
827 accumulation at the surface of the solid species and the ionic TEs. There are primarily three types
28
828 of surfaces available in AD which facilitate physical adsorption. They are: a) microbial biomass
829 surface; b) inorganic precipitate surfaces; and c) inert materials produced during the disintegration
830 step of AD. It is widely accepted that biomass surfaces carry a hydrophilic character in order to
831 acquire nutrients through diffusion. So, positively charged TEs are a are attracted towards the cell
832 surface of microbes in AD. Similarly, the mineral precipitates formed during AD has been reported
833 to undergo adsorption desorption with TEs. For example, Ni and Co has been reported to be
834 adsorbed on the surface of mineral precipitates, FeS. The case of inerts is interesting as it harbors
835 a maximum amount of unknown chemically active species which are often inorganic/organic in
836 nature (humic acid, fulvic acid and silicate minerals) It should be noted that the term “inert” does
837 not refer to the absence of any physicochemical activity. The term is derived from the ADM1
838 framework, in which, soluble and particulate inerts are formed in the disintegration step of AD.
839 There after, these entities do not take part in the biodegradation process. Inerts constitute around
840 10% of COD in AD systems. Moreover, the task force on the ADM1, have discussed the role of
841 “inerts” as a possible window in modelling inorganic/organic reactions which have not been taken
842 into consideration in the original ADM1.
843
844 In general, adsorption/desorption processes have been reviewed widely in geomicrobial systems.
845 Most of these reviews focus on the molecular level studies rather on chemical process level. Also,
846 the literature on the adsorption of trace elements in the context of AD is limited. There have been
847 limited attempts to model the adsorption of inorganic components in AD. For example, in order to
848 model the emission of odorant from AD processes, heavy metal adsorption to sulfur compounds
849 have been modeled. However, such studies do not consider the over all effect of heavy metal on
850 the microbial activity and methane production. Moreover, such studies focused more on the Fe
851 while neglecting Ni and Co which are critical to methanogenesis step. Finally, the concept of
852 binding site density, which is crucial to adsorption surfaces have not been considered. In a similar
853 attempt di-ethylhexyl phthalate biodegradation has been modelled in the framework of ADM1. In
854 this study, first order sorption/desorption models have been used to simulate the batch and
855 continuous data. Although the findings and approach of the study are inspirational, they serve little
856 purpose in modeling complex physicochemistry of TEs. Moreover, the study focused only on the
857 biodegradation of DEHP by working on monod kinetics.
858
859 In the absence of a elaborate modeling approach for TE adsorption in AD, it is important to review
860 the existing literature in the parallel areas of waste treatment and chemistry. To start, few excellent
861 studies (Crawford, 1999; Daughney and Fein, 1998; Hauduc et al., 2015; Konhauser, 2007; Papirio
862 et al., 2017; Smith, 2008).
863
864 2.3.4 Modeling TE-Biouptake in AD
865 TE-biouptake is one of the major fluxes, in addition to precipitation/dissolution, complexation and
866 adsorption, which influence the speciation of TEs in an aquatic system (Worms et al., 2006). The
29
867 extent of TE associated with the enzymes in methanogenesis pathway makes TE-biouptake
868 processes crucial to maintain an optimum intracellular concentration (Glass and Orphan, 2012;
869 Jansen et al., 2007; Worms et al., 2006; Yekta et al., 2016). Any shortage of TEs in the local milieu
870 or impaired TE-biouptake affects the methanogenesis pathways in anaerobic microbes (Glass and
871 Orphan, 2012). Modelling such complex phenomenon of TE transport across the microbial
872 membrane has been successful in the past, where elaborate transport microbial transport
873 mechanism have been mathematically modelled (Rotureau et al., 2015). Generally such models
874 are developed for single celled pure culture microbial system. These studies investigate the toxicity
875 level of various chemical species for a particular organism. For example, metal toxicity for algae
876 has been extensively modelled (Heijerick et al., 2002). The “Biotic Ligand Model” is the widely
877 used model framework to assess the toxic effect of various chemical species on an microorganism.
878 Similarly, an advanced version is the, “Free Ion Activity Model” which considers effect of free
879 ion on the microbial activity (Fermoso et al., 2015; Qiu et al., 2014; Worms et al., 2006). To best
880 of my knowledge, no ADM1 based TE biouptake model has been reported. Further, the complex
881 physicochemistry associated with the TEs in the liquid-solid phase has never been integrated to
882 biouptake in case of ADM1 models. Nonetheless, there are limited efforts to model the biouptake
883 of TEs during biodegradation of simple substrate like methanol (Jansen, 2004). In this study
884 (Jansen, 2004), biouptake of Ni and Co has been modelled in form of a two step mechanism which
885 involves, first, binding of Ni and Co to the surface of biomass and second, internalization by
886 transporters. Michaelis-Menten kinetics was used to represent the two step biouptake mechanism.
887 Here the TEs concentration in the bulk phase was linked to the growth kinetics by Monod kinetics.
888 However, to predict the internal concentration of TEs a Droop kinetic was used (Droop, 1983;
889 Jansen, 2004).
890 Microbial growth rate was defined as:

𝑑𝑥
= 𝜇𝑥, (2.12)
𝑑𝑡
891 where 𝜇 is the specific growth rate; 𝑡 is the time; 𝑥 is the biomass concentration. The growth
892 limitation due to substrate and available metal ions was represented as follows:
𝐶𝑀𝑒𝑂𝐻 (𝑄𝑐𝑜 − 𝑄𝑐𝑜,𝑚𝑖𝑛 ) (𝑄𝑁𝑖 − 𝑄𝑁𝑖,𝑚𝑖𝑛 )
𝜇 = 𝜇𝑚𝑎𝑥 , (2.13)
𝐶𝑀𝑒𝑂𝐻 + 𝐾𝑀𝑒𝑂𝐻 𝑄𝑐𝑜 𝑄𝑁𝑖
893 where 𝜇𝑚𝑎𝑥 is the maximum growth rate constant; 𝐶𝑀𝑒𝑂𝐻 is the primary substrate concentration;
894 𝐾𝑀𝑒𝑂𝐻 is the substrate Monod constant; 𝑄𝑐𝑜 is the cobalt content of the cell; 𝑄𝑐𝑜,𝑚𝑖𝑛 is the
895 minimum Co content cell; 𝑄𝑁𝑖 is the Ni content of the cell and 𝑄𝑁𝑖,𝑚𝑖𝑛 is minimum Ni content of
896 the cell. The competition of metal ions for the same transporter was taken into account through the
897 uptake flux rate as follows:
898
𝐶𝐶𝑜2+
⁄𝐾
𝐽𝑀,𝐶𝑜 2+
𝐽𝐶𝑜,𝑖𝑛 = 𝐽𝑚𝑎𝑥 , (2.14)
𝐶𝑁𝑖 2+ 𝐶𝐶𝑜2+
⁄𝐾 + ⁄𝐾 +1
𝐽𝑀,𝑁𝑖 2+ 𝐽𝑀,𝐶𝑜 2+
30
899 where 𝐽𝑚𝑎𝑥 is the maximum uptake flux; 𝐾𝐽𝑀,𝐶𝑜2+ is the Co Michaelis-Menten constant; 𝐾𝐽𝑀,𝑁𝑖 2+
900 is the Ni Michaelis-Menten constant; 𝐶𝐶𝑜2+ is the free Co concertation in the bulk phase; 𝐶𝑁𝑖 2+ is
901 the free Ni concentration in the bulk phase. The excretion of metal ions was modelled as a first
902 order process as follows:
𝐽𝐶𝑜,𝑒𝑓𝑓 = 𝑘max _𝑒𝑓𝑓,𝐶𝑜 (𝑄𝐶𝑜 − 𝑄𝐶𝑜,𝑒𝑓𝑓,𝑚𝑖𝑛 ), (2.15)
903 where 𝑘max _𝑒𝑓𝑓,𝐶𝑜 is the efflux rate constant and 𝑄𝐶𝑜,𝑒𝑓𝑓,𝑚𝑖𝑛 is a threshold value of Co inside the
904 cell below which no efflux takes place. The mass balances were established based on the above
905 mentioned rates. The mass balance for metal content of the cells was represented as follows:
𝑑𝑄𝐶𝑜
= 𝐽𝐶𝑜,𝑖𝑛 (𝑡) − 𝐽𝐶𝑜,𝑒𝑓𝑓 (𝑡) − 𝜇 (𝑡)𝑄𝐶𝑜 (𝑡). (2.16)
𝑑𝑡
906 The amount of total dissolved metal was accounted for with the following mass balance:
𝑑𝑄𝐶𝑜,𝑡𝑜𝑡𝑎𝑙
= − (𝐽𝐶𝑜,𝑖𝑛 (𝑡) − 𝐽𝐶𝑜,𝑒𝑓𝑓 (𝑡)) 𝑥(𝑡). (2.17)
𝑑𝑡
907
908
31
909 CHAPTER 3. MODEL DEVELOPMENT
910 A full structured kinetic mathematical model based on IWA ADM1, was developed, progressively
911 in stages (Figure 6), in order to quantitatively relate the processes affecting TE dynamics, assess
912 the relationship between trace metal speciation and bioavailability, and quantify the effects of TEs
913 on microbial growth and biogas production rate. The mathematical model incorporated
914 precipitation/dissolution, complexation and adsorption reactions in the form of different sub-
915 models. The extended model maintains the structure of ADM1 for the biochemical compartment
916 as well as physicochemical processes and considers the anaerobic digester as a continuously stirred
917 batch system with three phases, viz. solid, liquid and gas.
918
919
920 Figure 6. An overview of different stages of model development carried out in this thesis.
921 The proposed model assumes TE addition to the reactor will bring about a change in the
922 equilibrium of the various biochemical processes of the reactor. Tracing the change in the
923 equilibrium between the biotic and abiotic phase due to the addition of TEs is the main focus of
924 this work. To reflect the effect of TEs in AD, the anaerobic digestion model has been extended to
925 processes of precipitation, adsorption, complexation and biouptake of trace elements. The final
32
926 version of the model is a structured model based on ADM1 and includes additional blocks or sub-
927 models aimed at improving the physicochemical compartment of ADM1. In the biochemical
928 module, a consortium of microorganisms degrades and assimilates organic waste in five distinct
929 steps of disintegration, hydrolysis, acidogenesis, acetogenesis and methanogenesis to produce
930 methane, carbon dioxide and hydrogen. All the steps in the biochemical module are represented
931 by Monod kinetics except for disintegration, hydrolysis and biomass decay which are modeled as
932 first order kinetics. The microbial uptake of TE and the TE inhibition on special microbial activities
933 have been defined as well. In the physicochemical module, ion association/dissociation reactions
934 and liquid-gas transfer reactions are considered for liquid-liquid processes and liquid-gas processes
935 respectively. Equilibrium reactions are modeled by a set of differential equations. Henry's gas law
936 reproduces the liquid-gas transfer processes at the respective phase. Development of each specific
937 model has been described in the following sections. But before that, let us revisit the model
938 presentation format used in the standard IWA model ADM1 which has been extensively used to
939 describe and represent the proposed models here.
940
941 3.1 General model structure and presentation
942 The models developed in this study are structured models based on the ADM1. The general
943 framework of ADM1 allows representing the precipitation, complexation and adsorption models
944 uniformly. A simple representation of the substrate (𝑆𝑖 )-biomass (𝑋𝑖 )-product (𝑃𝑖 ) has been
945 presented here to introduce the basic methodology of modeling biomass growth due to uptake, as
946 also conceptualized in ADM1. The model construction starts with defining the various reactions
947 to be considered. As shown in figure 3.1, 𝑛 substrates are taken up by the cells and converted into
948 𝑚 products and various biomass constituent. 𝐽 number of reactions/processes are carried out for
949 such conversions.
950
951
952 Figure 7: A general description of substrate-biomass-product model.
953 The number of reactions and processes involved for each of the three models (precipitation,
954 complexation and adsorption) varies. Stoichiometric coefficients for the various components of
33
955 the model are designated. For example, 𝛼 for stoichiometry for biochemical reactions. Due to
956 multiplicity of components and processes two indices for stoichiometric coefficients have been
957 used to express the component involved and the process number. For example, 𝛼𝑖,𝑗 is the
958 stoichiometric coefficients for 𝑖 𝑡ℎ component for the 𝑗𝑡ℎ process. So, in the model framework,
959 stoichiometric coefficients for all the components including soluble and particulate have been
960 introduced in the 𝐽 processes, wherever deemed necessary.
961 Table 5. A general matrix representation of model

Component → 𝒊 1 2 3 4 5 n
Rate (𝝆𝒋 )
𝑗 Process ↓ 𝑺𝟏 𝑺𝟐 𝑺𝟑 𝑿𝟏 𝑿𝟐 𝑿𝒏
𝑃1 B −𝑣𝑖,𝑗 𝑣𝑖,𝑗 𝜌1
𝑃2 A −𝑣𝑖,𝑗 𝑣𝑖,𝑗 𝜌2
𝑃3 T −𝑣𝑖,𝑗 𝑣𝑖,𝑗 𝜌3
𝑃𝑗 P −𝑣𝑖,𝑗 𝑣𝑖,𝑗 𝜌𝑗
962
963 The models have been represented in form of matrix (Table 5) which represent reaction terms for
964 each component, subdivided by processes. Moving vertically through the matrix the process index
965 (𝑗) changes; while moving horizontally, the component index (𝑖) changes. The left column
966 describes the process index, while in the topmost row is for component index and nomenclature.
967 The process rates for each process is mentioned in the right column. The remaining rows are for
968 stoichiometric coefficients that describe the influence of that row’s process on individual
969 components.
970 For each component (liquid, solid and gas) in the model the mass balance can be expressed as:
971
972 𝐴𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 = 𝐼𝑛𝑝𝑢𝑡 − 𝑂𝑢𝑡𝑝𝑢𝑡 + 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛
973
974 The accumulation of a particular component in the model can be described as a function of input,
975 output of mass and the reaction which it undergoes. The physical characteristics of the system
976 determine the mass flow across the system. The reaction here specifies, processes such as growth,
977 hydrolysis and decay.
978 3.2 ADM1 based mathematical model for TE precipitation/dissolution in anaerobic digestion
979 processes
980 The first model, ADM1-P, includes four types of processes: protonation/deprotonation,
981 biochemical processes, precipitation/dissolution and gas transfers. TE precipitation/dissolution
982 reactions are the newest addition to the original ADM1 framework. TE biouptake, production and
983 release have been modeled as well. The dependence of biochemical processes on physicochemical
984 reactions has been improved by adding new inorganic components as a part of precipitation
985 reactions taking place in the AD. These inorganic components influence the biochemical processes
34
986 by affecting the pH of the system. Although total free TEs concentration is not directly related to
987 the pH, precipitation affects the total free TEs present in an anaerobic digester. Taking these into
988 account, new chemical equilibrium association/dissociation, ion pairing and precipitation reactions
989 have been added to ADM1. Overall, the model tracks the dynamics of 62 state variables which
990 constitute the differential mass balance equations governing the TEs, substrates, products and
991 bacterial groups involved in the AD of solid waste. The solid precipitates formed during the AD
992 are CaCO3, MgCO3, NiCO3, CoCO3, FeCO3, Ca3(PO4)2, Ni3(PO4)2, Fe3(PO4)2, Co3(PO4)2, NiS,
993 CoS, FeS and MgNHPO4. Complex organic matter has been also considered as a source for
994 phosphorous, sulfur and TEs which arise in the system from the disintegration step. Therefore,
995 phosphorous has been supposed to be released as HPO43-, which represents the most abundant form
996 in the pH range of 6 to 14. Similarly, the released sulfur has been considered in the form of HS -
997 which has been depicted as the abundant form in the pH range of 6 to 12. Finally, three in-silico
998 experiments were performed to demonstrate the model prediction capability.
999 3.2.1 The mathematical model
1000 For each component considered in the liquid phase, the gas and solid phase, the mass balance
1001 equations can be written in general form as follows:
1002
dVliq Si
= qin Sin,i − qout Si +Vliq (δi ρA,i (t, 𝐒) − ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ))
dt
m1
+ Vliq (∑ αi,j ρbio,j 𝐼𝑖 (t, 𝐒, 𝐗) )

j=1
m2 m2
+Vliq (∑ γi,j ρprec,j (t, 𝐒, 𝐗 𝐩 ) + ∑ γi,j ρdissol,j (t, 𝐒, 𝐗 𝐏 )),

j=1 j=1
i = 1, … , n1 , t > 0,
(3.1)
m1
dVliq X,i
= qin Xin,i − qout Xi + Vliq (∑ αi,j ρbio,j 𝐼𝑖 (t, 𝐒, 𝐗) ),
dt
j=1
i = n1 + 1, … , n2 , t > 0,
(3.2)
dSgas,i Sgas,i qgas
=− + Vliq ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ),
dt Vgas
i = 1, … , n1 , t > 0,
35
(3.3)
dVliq Xp,i in
= qin Xp,i − qout Xp,i + Vliq (𝛼̅𝑖 ρprec,i (t, 𝐒, 𝐗 𝐩 ) + 𝛽𝑖̅ ρdissol,i (t, 𝐒, 𝐗 𝐩 )),
dt
i = 1, … , m2 , t > 0,
(3.4)
1003 where:
1004 n1 is the number of soluble components;
1005 n2 − n1 is the number of particulate components, including the microbial species and the inert
1006 material;
1007 m1 is the number of biochemical processes considered;
1008 m2 is the number of precipitation/dissolution processes considered;
1009 αi,j is the stoichiometric coefficient of species i on biochemical process j;
1010 γi,j is the stoichiometric coefficient of species i on precipitation/dissolution process j;
1011 δi,j is the stoichiometric coefficient for acid-base reaction for ith on biochemical process j;
1012 Si is the 𝑖 𝑡ℎ soluble component, 𝐒 = (S1 , … , Sn1 );
1013 Xi is the ith particulate component, 𝐗 = (Xn1+1 , … , Xn2 );
1014 Xp,i is the ith precipitate, 𝐗 𝐩 = (Xp,1 , … , Xp,m2 );
1015 Sgas,i is the ith soluble gas component, 𝐒𝐠𝐚𝐬 = (Sgas,1 , … , Sgas,n2 );
1016 ρA,i (t, 𝐒) is the process rate for acid base reactions for ith soluble component;
1017 ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ) is the gas transfer process rate for ith soluble component;
1018 ρbio,j (t, 𝐒, 𝐗) is the rate for jth biochemical process;
1019 ρprec,j (t, 𝐒, 𝐗 𝒑 ) is the rate for jth precipitation process;
1020 ρdissol,j (t, 𝐒, 𝐗 𝒑 ) is the rate for jth dissolution process;
1021 Vliq is the volume of liquid in the reactor.
1022 Vgas is the volume of gas in the reactor
36
1023 Effect of TEs on biochemical processes has been modeled by introducing an additional term
1024 (𝐼𝑀𝑒 2+ ) in the inhibition expressions as follows:
𝐼1 = 𝐼𝑝𝐻 ∙ 𝐼𝐼𝑁_𝑙𝑖𝑚 ∙ 𝐼𝑀𝑒 2+ , (3.5)

𝐼2 = 𝐼𝑝𝐻 ∙ 𝐼𝐼𝑁_𝑙𝑖𝑚 ∙ 𝐼𝐻2 ∙ 𝐼𝑀𝑒 2+ , (3.6)
𝐼3 = 𝐼𝑝𝐻 ∙ 𝐼𝐼𝑁_𝑙𝑖𝑚 ∙ 𝐼𝐻2 ∙ 𝐼𝑁𝐻3 ∙ 𝐼𝑀𝑒 2+ , (3.7)
1025
1026 where 𝐼𝑖 , 𝑖 = 1,2,3 are the established inhibition functions used in ADM1 framework (Batstone
1027 et al., 2002), 𝐼𝑝𝐻 is the inhibition factor due to pH, 𝐼𝐼𝑁_𝑙𝑖𝑚 is the inhibition factor for inorganic
1028 nitrogen , and 𝐼𝑁𝐻3 is the inhibition factor for ammonia and 𝐼𝑀𝑒 2+ is expressed as:
𝑎1 ∙ 𝑀𝑒 2+ + 𝑎2 (3.8)
𝐼𝑀𝑒 2+ = ,
𝑀𝑒 2+ + 𝑏1 ∙ 𝑀𝑒 2+ + 𝑏2
1029 where, 𝑎1 , 𝑎2 , 𝑏1 , 𝑏2 are assumed constants and 𝑀𝑒 2+ .is the concentration of the free TE. The
1030 function takes into account both the stimulation and inhibiting action of TEs on biochemical rate
1031 depending on the concentration range.
1032 The gas-transfer for H2, CH4, CO2 and H2S has been taken into account. The equations of liquid
1033 to gas mass transfer have been included in the proposed model as suggested in ADM1
𝜌𝑇,𝑖 = 𝑘𝑙 𝑎(𝑆𝑙𝑖𝑞,𝑖 − 𝐾𝐻,𝑖 𝑃𝑔𝑎𝑠,𝑖 ) (3.9)
1034 where the terms 𝑘𝑙 𝑎, 𝑆𝑙𝑖𝑞,𝑖 , 𝐾𝐻,𝑖 and 𝑃𝑔𝑎𝑠,𝑖 denote the volumetric gas transfer coefficient, the
1035 dynamic state variable for the liquid component which has a corresponding soluble gas, the
1036 Henry’s constant and the summation of partial pressure exerted by the individual biogas
1037 components (H2, CO2, CH4 and H2S) respectively.
1038 In addition to the six acid-base processes considered in the original ADM1, 6 new acid/base pairs
1039 have been implemented in the ADM1-P. Bicarbonate/carbonate, phosphoric acid/dihydrogen
1040 phosphate, dihydrogen phosphoric acid/hydrogenphosphate, hydrogenphosphoric acid/phosphate,
1041 hydrogen sulfide/bisulfide and bisulfide/sulfide acid pairs are the new acid-base system. The
1042 kinetic acid/base reaction rates are described by the following equation:
𝜌𝐴⁄ = 𝑘𝐴⁄ (𝑆𝑙𝑖𝑞.𝐴− ∙ 𝑆𝑙𝑖𝑞.𝐻 + − 𝐾𝑎 ∙ 𝑆𝑙𝑖𝑞.𝐻𝐴 ) (3.10)

𝐵 𝐵
1043 where 𝑘𝐴⁄ and 𝐾𝑎 are rate coefficient for the base to acid reaction and acid association and
𝐵
1044 dissociation constant respectively, 𝑆𝑙𝑖𝑞.𝐴− is the state variable for the conjugate base and 𝑆𝑙𝑖𝑞.𝐻𝐴 is
1045 the dynamic state variable for the respective acid.
37
1046 Accordingly, the charge balance has been modified to take into account the effects of additional
1047 components as:
𝑆𝐴𝐶 − 𝑆𝑃𝑅𝑂 − 𝑆𝐵𝑈 − 𝑆𝑉𝐴−

𝜃 = 𝑆𝐶𝐴𝑇 + 𝑆𝑁𝐻4 − − − − − 𝑆𝐻𝐶𝑂3 − 2 ∙ 𝑆𝐶𝑎2+ + 2 ∙ 𝑆𝑀𝑔2+ + 2
64 112 160 208
∙ 𝑆𝑁𝑖 2+ + 2 ∙ 𝑆𝐶𝑜 2+ + 2 ∙ 𝑆𝐹𝑒 2+ − 2 ∙ 𝑆𝐻2 𝑃𝑂4− − 2 ∙ 𝑆𝐻𝑃𝑂42− − 3 ∙ 𝑆𝑃𝑂43−
− 𝑆𝐻𝑆 1− − 𝑆𝑆 2− − 𝑆𝐴𝑁
(3.11)
+
1048 balance can be explicated in terms of SH as follows:
𝜃 1
1049 𝑆𝐻 + = − 2 + 2
√𝜃 2 + 4𝐾𝑤
1050 where, θ is the net charge in the system, SH+ is the hydrogen ion concentration and KW is the
1051 ionic product of water.
1052 The general precipitation-dissolution equilibrium reaction is represented as follows:
+ −
𝑥𝑀𝑉 + 𝑦𝐴𝑉 ↔ 𝑀𝑥 𝐴𝑦 (3.12)
1053 where, x and y are the number of cations and anions and v+ and v- are the valences. Precipitation
1054 occurs only when the solution is supersaturated with participating ionic species. The reverse
1055 reaction, dissolution, takes place when the system is undersaturated.
1056 The concentration based kinetic rate of the precipitation process has been modelled as following:
𝑣
1⁄
𝑣 1⁄
(3.13)
𝑣+ 𝑣
′ 𝑚+ ) 𝑎 − ]𝑣 − ] ′
𝜌𝑝𝑟𝑒𝑐 = 𝑘𝑟,𝑚 −
[[[(𝑀 ]∙[𝐴 − (𝑘𝑠𝑝,𝑚 −
) ]
𝑣+𝐴𝑣 𝑣+𝐴𝑣
1057 where, Mv+ and Av- denote the concentrations of cations and anions, k’r,Mv+Av- is the precipitation
1058 rate constant, k’sp,Mv+Av- is the solubility product, v+ and v- are the total number of cationic and
1059 anionic charges respectively and v  v   v  . The precipitation processes are reversible in the
1060 model.
1061 When the system is undersaturated, the dissolution reaction can be represented as:
1⁄ 𝑣
+ + 𝑣 1⁄ 1 (3.14)
′ [[[(𝑀𝑚 𝑣 ] 𝑎− 𝑣 − ] ′ 𝑣]
𝜌𝑑𝑖𝑠𝑠 = 𝑘𝑟,𝑚 ) ∙[𝐴 ] − (𝑘𝑠𝑝,𝑚 )
𝑣+𝐴𝑣
−
𝑣 +𝐴 𝑣 − 𝐾𝑝𝑟𝑒𝑐
1 + 𝑀 +𝐴 −
𝑣 𝑣
38
1062 where, Kprec is an assumed constant. At undersaturation, dissolution takes place till equilibrium is
1063 reached or until the mineral precipitate in the reactor becomes zero.
1064 3.2.2 Model implementation and simulations
1065 The implementation of the precipitation and dissolution reactions in the physicochemical
1066 framework of ADM1 was implemented progressively. First the original ADM1 was implemented
1067 in the differential equation (DE) mode of implementation. DE implementation of ADM1 was
1068 carried out with an aim to add large number of new state variables related to the
1069 precipitation/dissolution, sulfur, phosphorus and carbonate system of reactions. Moreover, the
1070 role of TE on AD has been incorporated in ADM1 by adding a dose response function to the
1071 microbial activity. Further, new acid-base processes involving sulfur, phosphorus and inorganic
1072 carbon was introduced. Most importantly, new solid-liquid precipitation/dissolution step was
1073 added. All these model improvements were made possible by collecting thermodynamic
1074 parameters of the new physicochemical processes added. These parameters was collected from
1075 literature (Musvoto et al., 2000b; Roussel, 2012). The ordinary differential equation (ODE) system
1076 of equations instituting the new precipitation/dissolution mode was implemented in a original code
1077 and solved using ODE 15s solver of MATLAB® platform. Further, model simulations in form of
1078 scenarios were carried out to assess the model prediction capability to represent changes in the
1079 physicochemistry of the anaerobic digestion system.
1080 Further details regarding dynamic state variables, model parameters and initial conditions can be
1081 found in publication II.
1082
1083 3.3 ADM1 based mathematical model for TE complexation in anaerobic digestion processes
1084 The second model, ADM1-PC, considers a three phase AD system (solid-liquid-gas). Reactions
1085 occurring in the liquid phase are central to the TE complexation module and they can be classified
1086 in two groups. The first one includes the enzyme-catalysed biological reactions which occur at a
1087 slower rate. Biotic processes, including hydrolysis, acidogenesis, acetogenesis, and
1088 methanogenesis belong to this category. The second group includes the liquid phase chemical
1089 reactions, which occur at a faster rate. Protonation/deprotonation and TE-complexation with VFAs
1090 and EDTA fall into this category. Precipitation and redissolution take place at the solid/liquid
1091 interphase and have been modeled according to the ADM1-P model. H2, CH4, CO2 and H2S are
1092 considered in the gas phase.
1093 The proposed model incorporates the complexation reactions of TEs with VFAs and the synthetic
1094 organic chelator, EDTA. Although there are other organic chelators in AD environment (thiol
1095 containing amino acids (Yekta et al., 2016), humic substances (Plaza et al., 2006) and soluble
1096 microbial products (Aquino and Stuckey, 2004)), complexation reactions of VFAs and EDTA
1097 were chosen due to the familiarity (Callander and Barford, 1983a; Lizarralde et al., 2015a) of the
39
1098 chemical species involved in the complexation reactions and the availability of thermodynamic
1099 constants (Callander and Barford, 1983b; Smith et al., 2004). Moreover, the effects of synthetic
1100 chelators in AD systems have been widely reported (Thanh et al., 2015) while there has been no
1101 effort on modelling such processes in AD systems.
1102 3.3.1 The mathematical model
1103 Mass balance for components in liquid phase, the gas phase and the solid phase can be written in
1104 general form as follows:
dVliq Si
= qin Sin,i − qout Si +Vliq (δi ρA,i (t, 𝐒) − ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 )) +
dt
m1 m2
+ Vliq (∑ αi,j ρbio,j (t, 𝐒, 𝐗) + ∑ βi,j ρcmplx,j (t, 𝐒)) +

j=1 j=1
m3 m3

j=1 j=1
i = 1, … , n1 , t > 0,
(3.15)
m1
dVliq X,i
= qin Xin,i − qout Xi + Vliq (∑ αi,j ρbio,j (t, 𝐒, 𝐗) ),
dt
j=1
i = n1 + 1, … , n2 , t > 0,
(3.16)
dSgas,i Sgas,i qgas
dt Vgas
i = 1, … , n1 , t > 0,
(3.17)
dVliq Xp,i in
= qin Xp,i − qout Xp,i + Vliq (𝛼̅𝑖 ρprec,i (t, 𝐒, 𝐗 𝐩 ) + 𝛽𝑖̅ ρdissol,i (t, 𝐒, 𝐗 𝐩 )),
dt
i = 1, … , m3 , t > 0,
(3.18)
1105 where:
1107 n2 − n1 is the number of particulate components, including the microbial species and the inert
1108 material;
40
1110 m2 is the number of complexation processes considered;
1112 αi,j is the stoichiometric coefficient of species i on biochemical process j;
1113 βi,j is the stoichiometric coefficient of species i on complexation process j;
1114 γi,j is the stoichiometric coefficient of species i on precipitation/dissolution process j;
1115 δi,j is the stoichiometric coefficient for acid-base reaction for ith on biochemical process j;
1116 Si is the 𝑖 𝑡ℎ soluble component, 𝐒 = (S1 , … , Sn1 );
1117 Xi is the ith particulate component, 𝐗 = (Xn1+1 , … , Xn2 );
1118 Xp,i is the ith precipitate, 𝐗 𝐩 = (Xp,1 , … , Xp,m3 );
1120 ρA,i (t, 𝐒) is the process rate for acid base reactions for ith soluble component;
1121 ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ) is the gas transfer process rate for ith soluble component;
1122 ρbio,j (t, 𝐒, 𝐗) is the rate for jth biochemical process;
1123 ρcomplx,j (t, 𝐒, 𝐗) is the rate for jth complexation process;
1124 ρprec,j (t, 𝐒, 𝐗) is the rate for jth precipitation process;
1125 ρdissol,j (t, 𝐒, 𝐗) is the rate for jth dissolution process;
1126 Vliq is the volume of liquid in the reactor.
1127 Vgas is the volume of gas component in the reactor.
1128 The formation and dissociation of TE complexes constitute the main feature of the proposed model
1129 and are considered to be kinetically controlled. The complexes result from the reaction between
1130 TE cations and EDTA or volatile fatty acids. The complexation reaction involving a generic
1131 [MeEDTA]2- complex is used here to illustrate the kinetic framework adopted in the complexation
1132 module of the proposed model. New state variables constituting the EDTA acid/base system and
1133 complexes have been introduced in the complexation module. The approach used is based on
1134 (Willet and Rittmann, 2003). The complexation reactions considered are as follows:
41
𝑘1
𝑀𝑒 2+ + 𝐸𝐷𝑇𝐴4− → [𝑀𝑒𝐸𝐷𝑇𝐴]2− (3.19)
𝑘 −1
[𝑀𝑒𝐸𝐷𝑇𝐴]2− → 𝑀𝑒 2+ + 𝐸𝐷𝑇𝐴4− (3.20)
1135 where, k1 is the [𝑀𝑒𝐸𝐷𝑇𝐴]2− formation rate constant in [M -1 h-1]; k-1 is the dissociation constant
1136 in [h-1]. The equations (3.19) and (3.20) can be rewritten as:
𝐾
[𝑀𝑒𝐸𝐷𝑇𝐴] −2
𝑀𝑒 2+ + 𝐸𝐷𝑇𝐴4− ↔ [𝑀𝑒𝐸𝐷𝑇𝐴]2− (3.21)
1137 where, 𝐾𝑀𝑒𝐸𝐷𝑇𝐴 is the equilibrium constant for [𝑀𝑒𝐸𝐷𝑇𝐴]2− in M . The rate equation for the
-1
1138 mechanism in equation (3.21) can be written as:
𝜌𝑐𝑜𝑚𝑝 = 𝐶𝑥 (𝑘1 𝑆𝐸𝐷𝑇𝐴4− ∙ 𝑆𝑀𝑒 2+ − 𝑘−1 ∙ 𝑆[𝑀𝑒𝐸𝐷𝑇𝐴]2− ) (3.22)

1139 where, 𝑆𝑀𝑒 2+ , 𝑆𝐸𝐷𝑇𝐴4− and 𝑆[𝑀𝑒𝐸𝐷𝑇𝐴]2− are the dynamic state variable for the free TE
1140 concentration, organic ligand and TE-organic complex respectively and 𝐶𝑥 is the complexation
1141 kinetic rate constant. The overall formation constant of a TE-EDTA complex is set at a particular
1142 value and the reverse rate constant is calculated from the values of the equilibrium constant,
1143 otherwise known as stability constant (Callander and Barford, 1983b; Willet and Rittmann, 2003).
1144 The formation rate constant, 𝑘1 , for all the species have been assumed to be 10 × 1010 [mol -1 h -
1
1145 ], based on the formation rate constant values available in literature (Willet and Rittmann, 2003)
1146 and the dissociation rate constant, 𝑘−1 , have been calculated from the stability constant by a simple
1147 relation as:
𝑘1
𝐾[𝑀𝑒𝐸𝐷𝑇𝐴]2− = (3.23)
𝑘−1
1148 The organic complexation reactions considered in the proposed model are the predominant ones
1149 occurring in an anaerobic digester supplemented with the synthetic organic ligand EDTA.
1150 The TE-VFA complexes considered in the proposed model are those most likely to be formed in
1151 an anaerobic environment (Lizarralde et al., 2015a). TE-VFA complexation follows the same
1152 kinetic model as TE-EDTA complexation processes. The stability constant of the complexes
1153 determines the extent of the complex formation and occurrence in the anaerobic digester. In
1154 general, the stability constants of TE-VFA complexes are significantly lower than the TE-EDTA
1155 complexes. However, TE-VFA complexation reactions have been incorporated due to the presence
1156 of large concentration of VFAs in AD systems.
1157 The effect of TE on microbial kinetics have included as an additional inhibition terms to the ADM1
1158 inhibition terms as 𝐼𝑀𝑒 2+ and is expressed as:
2+
𝑎1 ∙ (𝑀𝑒 2+ + 𝐶 𝑀𝑒 ) + 𝑎2
𝐼𝑀𝑒 2+ = 2+ 2+ , (3.24)
(𝑀𝑒 2+ + 𝐶𝑀𝑒 )2 𝑏1 ∙ (𝑀𝑒 2+ + 𝐶𝑀𝑒 ) + 𝑏2
1159 where, 𝑎1 , 𝑎2 , 𝑏1 , 𝑏2 are assumed constants which can been adjusted to obtain a desirable optimum
1160 dose-response function at a particular TE concentration, 𝑀𝑒 2+ denotes the concentration of
42
2+
1161 bioavailable TEs within the bulk liquid and 𝐶 𝑀𝑒 is the fraction of TE contained in the complex
1162 [Me-EDTA]2- (Hu et al., 2008; Vintiloiu et al., 2013). The function considers both the stimulation
1163 and inhibiting action of TEs on the biochemical rates depending on the concentration range.
1165 The implementation of the model involved integration of complexation reactions in the already
1166 developed precipitation model. In this regard, new state variables and process rates representing
1167 TE-EDTA and TE-VFA complexation were introduced in the ADM1 framework. The
1168 physicochemical reactions were updated to include the association-dissociation reactions of
1169 various components and species of EDTA in liquid phase. Further the TE-dose response function
1170 was updated to incorporate the effect of reduced TEs in the liquid phase on the microbial kinetics,
1171 due to complexation reactions. Parameters values governing complexation kinetics, precipitation
1172 kinetics and dissolution kinetics were incorporated from the literature (Callander and Barford,
1173 1983a; Musvoto et al., 2000b; Roussel, 2012). The model was also verified by chemical
1174 equilibrium calculations performed on Visual MINTEQ for complexation reactions. The system
1175 of ODE was written in an original code and solved using ODE 15s solver of the MATLAB®
1176 platform. Specific model simulation scenarios were carried out to check the extent of complexation
1177 reaction in an environment dominated by mineral precipitation.
1178 Further details regarding the dynamic state variables, model parameters and initial conditions can
1179 be found in publication III.
1180
1181 3.4 ADM1 based mathematical model for TE speciation in anaerobic digestion processes
1182 The third model, ADM1-PCD, simulates the speciation of TEs in AD processes. The model is
1183 aimed to analyze adsorption, precipitation and complexation, as three primary TE flux, in
1184 conjugation with methane production during anaerobic biodegradation of complex organic matter.
1185 Overall, the proposed model tracks the dynamics of 175 state variables which constitute the
1186 components of the proposed AD model. For each component considered in the liquid phase (i.e.
1187 microbial species, dissolved compounds, adsorbed species, precipitates and complexes), gas phase
1188 and solid phase the general mass balance can be written as:
m5
dVliq Si
= qin Sin,i − qout Si +Vliq (∑ δi,j ρA,j (t, 𝐒) − ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ))
dt
𝑗=1
(3.25)
m1 m2
+ Vliq (∑ αi,j ρbio,j (t, 𝐒, 𝐗) + ∑ βi,j ρcmplx,j (t, 𝐒))

j=1 j=1
43
m3 m3 m4
̅ 𝐚𝐝𝐬 )),
+Vliq (∑ γi,j ρprec,j (t, 𝐒, 𝐗 𝐏 ) + ∑ γi,j ρdissol,j (t, 𝐒, 𝐗 𝐏 ) + ∑ εi,j ρs,j (t, 𝐒, 𝐗 𝐚𝐝𝐬 , 𝐗
j=1 j=1 j=1
i = 1, … , n1 , t > 0,
m1
dVliq Xi
dt (3.26)
j=1
i = n1 + 1, … , n2 , t > 0,
dSgas,i Sgas,i qgas

dt Vgas
(3.27)
i = 1, … , n1 , t > 0,
dVliq Xp,i
= qin Xpin − qout Xp,i + Vliq (𝛼̅𝑖 ρprec,i (t, 𝐒, 𝐗 𝐏 ) + β̅𝑖 ρdissol,i (t, 𝐒, 𝐗 𝐩 )),
dt
(3.28)
i = 1, … , m3 , t > 0,
1189 where:
1191 n2 − n1 is the number of particulate components, including complex organic matter and inert;
1193 m2 is the number of complexation processes considered;
1195 m4 is the number of adsorption/desorption processes considered;
1196 m5 is the number of acid-base reactions considered;
1197 αi,j is the stoichiometric coefficient of the ith species on biochemical process j;
1198 βi,j is the stoichiometric coefficient of the ith species on complexation process j;
1199 γi,j is the stoichiometric coefficient of the ith species on precipitation/dissolution process j;
1200 δi,j is the stoichiometric coefficient of the ith species on acid-base reaction j;
44
1201 εi,j is the stoichiometric coefficient of the ith species on adsorption/desoprtion reaction j;
1202 ̅𝑖 and β̅𝑖 are the stoichiometric coefficients for the precipitate formation and dissolution;
α
1203 Si is the ith soluble component, 𝐒 = (S1 , … , Sn1 );
1204 Xi is the ith particulate component, 𝐗 = (Xn+1 , … , Xn2 );
1205 Xp,i is the ith precipitate, 𝐗 𝐩 = (X p,1 , … , Xp,m3 );
1206 Xads,i is the concentration of free binding sites for a specific adsorbent component, 𝐗 𝐚𝐝𝐬 =
1207 (Xads,1 , … , Xads,n3 );
1208 ̅ ̅ 𝐚𝐝𝐬 =
Xads,i is the concentration of occupied binding sites for a specific adsorbent component, 𝐗
1209 ̅ ads,1 , … , X
(X ̅ ads,n );
3
1210 n3 is the number of adsorbing components, including microbial species, inert and precipitates
1211 (n2 − n1 + m3 );
1213 ρA,j (t, 𝐒) is the process rate for acid base reaction j;
1214 ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ) is the gas transfer process rate for the ith soluble component;
1215 ρbio,j (t, 𝐒, 𝐗) is the rate for the jth biochemical process;
1216 ρcomplx,j (t, 𝐒, 𝐗) is the rate of the jth complexation process;
1217 ρprec,j (t, 𝐒, 𝐗 𝒑 ) is the rate for the jth precipitation process;
1218 ρdissol,j (t, 𝐒, 𝐗 𝒑 )is the rate for the jth dissolution process;
1219 ̅ 𝐚𝐝𝐬 ) is the rate for the jth sorption/desorption process;

ρs,j (t, 𝐒, 𝐗 𝐚𝐝𝐬 , 𝐗
1220 Vliq and Vgas are the liquid and gas volume of the reactor.
1221 The model considers a three phase AD system (solid-liquid-gas). Reactions occurring in the liquid
1222 to solid phase are central to the TE adsorption module and they can be classified in three groups:
1223 adsorption on biomass, inert and precipitates. Protonation/deprotonation and TE-complexation
1224 with VFAs and EDTA have been considered in the complexation module (Maharaj et al., 2019).
1225 Precipitation and redissolution taking place in the solid phase and have been modeled according
1226 to (Maharaj et al., 2018). H2, CH4, CO2 and H2S are considered in the gas phase.
45
1227 TE adsorption has been modeled using adsorption rate equations that are solved over time for
1228 concentrations of kinetically controlled complexes. Adsorption is modeled as a reversible process.
1229 Adsorption of TE ions (Ni and Co) on mineral surface has been considered only for FeS.
1230 Adsorptions of TEs on inert particulate matter and biomass have been considered as well.
1231 The sorption/desorption processes are kinetically controlled and can be represented as follows:
𝑘1
1232 𝑀𝑒 2+ + 𝑀𝑒𝑆 → (𝑀𝑒 ≡ 𝑀𝑒𝑆) (3.29)
𝑘−1
1233 (𝑀𝑒 − 𝑀𝑒𝑆) → 𝑀𝑒 2+ + 𝑀𝑒𝑆 (3.30)
1234 Where, k1 is the 𝑀𝑒 − 𝑀𝑒𝑆 formation rate constant; k-1 is the dissociation constant. Two pairs of
1235 reactions in (3.29-30) have equilibrium reactions and constants defined by:
𝐾𝑀𝑒𝑀𝑒𝑆
1236 𝑀𝑒 2+ + 𝑀𝑒𝑆 ↔ (𝑀𝑒 ≡ 𝑀𝑒𝑆) (3.31)
1237 Where, 𝐾𝑀𝑒𝑀𝑒𝑆 is the equilibrium constant for 𝑀𝑒 − 𝑀𝑒𝑆. The rate equation for the mechanism
1238 in equation (3.31) and introduced in equation (3.25) 𝜌𝑠 = 𝜌𝑎𝑑𝑠 − 𝜌𝑑𝑒𝑠 can be written as:
1239 𝜌𝑎𝑑𝑠𝑜𝑟𝑝 = 𝐾𝑥 ∙ 𝑘1 𝑋ads ∙ 𝑆𝑀𝑒 2+ (3.32)
1240 𝜌𝑑𝑒𝑠𝑜𝑟𝑝 = 𝐾𝑥 ∙ 𝑘−1 ∙ 𝑋̅ads (3.33)
1241 where, SMe2+, 𝑋ads and 𝑋̅ads are the dynamic state variable for free TE concentration, free binding
1242 site concentration, adsorbed complex and Kx is the sorption/desorption kinetic rate constant. Note
1243 that 𝑋𝑎𝑑𝑠 and 𝑋𝑑𝑒𝑠 constitute additional state variables of the system, whose dynamics is explicitly
1244 tracked through the sorption/desorption rate. The current model formulation considers the biomass,
1245 inert and precipitates as having similar binding characteristics. It should also be noted that the
1246 model does not distinguish between binding sites based on chemical characteristics. The idea here
1247 is to adapt a universal behaviour for inert, biomass and precipitates based on binding site density
1248 which can be latter improved to incorporate specific binding characteristics such as carboxylate,
1249 phosphate and sulphate surfaces.
1251 The parameters and stability constant of various adsorbed species have been collected from the
1252 literature and assumed based on modelling experience. The system of ordinary differential
1253 equations constituting the extended-ADM1 model has been implemented in an original code and
1254 has been solved using the algorithm ODE 15s, a multistep, variable-order solver on MATLAB®
1255 platform. Numerical simulations of specific scenarios have been performed to test model
1256 reliability.
46
1257 Further details regarding dynamic state variables, model parameters and initial conditions can be
1258 found in publication IV.
1259 CHAPTER 4. DISCUSSION

1260 4.1 ADM1 based mathematical model for TE precipitation/dissolution in anaerobic digestion
1261 processes
1262 Paper I focus on tracing the dynamics of TE species, mineral precipitates in particular. In this
1263 regard, a full ADM1 based kinetic framework has been developed to implement the precipitation
1264 processes in AD systems. Specifically, new precipitation and chemical equilibrium
1265 association/dissociation reactions have been added to the ADM1 framework and consequently the
1266 charge balance has been updated as per the requirement. The developed model takes into account
1267 the inorganic components (carbonate, phosphate and supplied) in liquid phase as well as mineral
1268 precipitates in solid phase. Mineral precipitates for Mg and Ca have been also considered.
1269 Moreover, the formation of struvite has been included in the model. A second order kinetic mineral
1270 precipitation model, based on (Koutsoukos et al., 1980) has been recruited to trace the formation
1271 of mineral precipitates. The model simulations were verified with the predictions of a chemical
1272 equilibrium solver, Visual MINTEQ. The observations presented in Paper II highlight the effects
1273 of TE precipitation and uptake on cumulative methane production. Further, a set of simulations
1274 projected the predictive capacity of the model in terms of TEs starvation and effect of change in
1275 Sulphur and phosphorous initial concentration. Overall, components and reactions for mineral
1276 precipitation have been incorporated, the prediction of precipitation sub-module has been verified
1277 and finally numerical simulations have been carried out to demonstrate the dominance of mineral
1278 precipitates among inorganic species in the AD system.
1279 The model results presented in this study (Paper II) demonstrate the relation between the
1280 physicochemical and biochemical processes in AD systems. The observations presented above
1281 highlight the effects that TEs precipitation/redissolution and uptake have on change in pH and on
1282 overall methane production. Objectives such as understanding speciation of TE in AD system
1283 through process models or elucidating the interaction of various inorganic species in an AD system
1284 requires complex model structure that can extend system boundaries to include solid-liquid-gas
1285 phase. For instance, the work of Xavier et al. (2016) extended ADM1 and demonstrated the
1286 interactions among phosphorous, sulphur and iron and their potential effect on total biogas
1287 production (CO2, CH4, H2 and H2S) (Flores-Alsina et al., 2016b). Similarly, Zhang et al. (2015)
1288 resolved discrepancies in inorganic nitrogen and carbon balances and included an improved
1289 physicochemical framework to describe solid phase precipitation reactions (Y. Zhang et al., 2015).
1290 The model was validated with experimental work of Musvoto et al (2000a) for precipitation of
1291 calcium and magnesium in batch reactor ( Musvoto et al., 2000a). Although such works are
1292 important in terms of process modeling and introduce new model structure, they lack of an accurate
1293 model description which can take into account the major TEs dynamics. The simulations
1294 demonstrate that out of all the chemical systems considered in the model, i.e sulphate, phosphate
47
1295 and carbonate, sulphur predominate in the solid phase of the anaerobic batch digester (Paper II).
1296 Similarly, with respect to all precipitates, when considered along with complexation reaction, FeS
1297 is also the major precipitate found in the system (Paper III) as Fe forms insoluble salts with sulphur.
1298 Theoretically all metals should precipitate as metal sulfide. Solubility product values of NiS and
1299 CoS are close but higher than FeS. Using sulphate reducing bacteria to produce sulfide, it has been
1300 shown that 99.8% of iron and zinc precipitated as MeS and X-ray diffraction spectroscopy analysis
1301 revealed that FeS, FeS2 and ZnS are the predominant compounds (van der Veen et al., 2007).
1302 Precipitation of sulfide with iron has been studied more than any other metals. The two main
1303 precipitates that have been observed in AD are pyrite and amorphous FeS (Kaksonen et al., 2003).
1304 According to pourbaix diagram, pyrite is the main compound found in anaerobic digesters
1305 (Roussel, 2012). However, amorphous FeS is found predominantly before the start of pyretization
1306 process which has been agreeably illustrated by the model simulations.
1307 In summary, mineral precipitates along with free TE species in AD processes have been quantified
1308 by a new approach, to model physicochemical processes (Paper II), in an extended version of the
1309 original ADM1. An original tailor code to solve the aqueous chemistry and precipitation reactions
1310 has been used and verified with external chemical equilibrium software (Visual MINTEQ). These
1311 modifications in ADM1 equip it to measure the amount of TE present at a particular point of time
1312 in the AD reactor. This extends the ability of the ADM1 based models to be considered as a tool
1313 in TE supplementation.
1314 4.2 ADM1 based mathematical model for TE complexation in anaerobic digestion processes
1315 Paper III is a direct logical extension of Paper II. In Paper II focus was on the modeling of
1316 precipitation reactions which affect the dynamics of TEs in AD systems. However, as already
1317 mentioned, the TE dynamics also depends on complexation reactions. Complexation of TEs with
1318 organic ligands present in the system as well as the synthetic organic ligands was felt necessary
1319 based on literature survey. It is now well established that TEs complexation with synthetic organic
1320 chelator (EDTA/NTA/EDDS) plays an important role in process optimization and imparts process
1321 stability with increased production of methane (Aquino and Stuckey, 2007; Hu et al., 2008; Ortner
1322 et al., 2015; Thanh et al., 2015; Tsapekos et al., 2018; Vintiloiu et al., 2013). The major
1323 shortcomings related to designing a supplementation strategy for TEs and EDTA have three basic
1324 reasonings, which are: the fate of EDTA and TEs in AD systems; what is the dominant form of
1325 species at a particular period of time in reactor operation; and to which extent the synthetic
1326 chelators should be dosed to improve the biogas production. The primary aim for all these
1327 reasonings comes out from a single question, that is, is it possible to decrease the amount of TE
1328 supplementation if dosed in addition to synthetic chelators? Despite a significant amount of
1329 research efforts being dedicated to address these questions (Choong et al., 2016; Thanh et al., 2015;
1330 van Hullebusch et al., 2016), there is no consolidated understanding of the dynamic partitioning
1331 of EDTA and TEs in AD systems among different inorganic fractions (Fermoso et al., 2015; van
1332 Hullebusch et al., 2016). Hence EDTA dosing depends largely on case to case basis and it is
1333 affected by the type of substrate, the initial TE dosage and the stage of reactor operation. The
48
1334 problem of low detection level can also be added to this. Thus, a mathematical model based on
1335 ADM1, as presented in Paper II, may be recruited to study the effect of TE-complexation reactions
1336 on an AD system. This introduces a novel attempt to incorporate TE complexation reactions in the
1337 framework of AD to study the dynamics and effect of TE on methane production. Nevertheless,
1338 there have been some recent advances in showcasing the effect of synthetic chelators using
1339 experimental set ups (Fermoso et al., 2008; Qing H. Hu et al., 2008; Qing Hao Hu et al., 2008;
1340 Schiavon Maia et al., 2017; P. Thanh et al., 2017; P. M. Thanh et al., 2017; Vintiloiu et al., 2013).
1341 In this part of the research a new model based on ADM1 approach has been proposed to simulate
1342 TEs complexation, along with precipitation/dissolution and their effect on the anaerobic batch
1343 methane production. TEs complexation reactions with VFAs and EDTA have been incorporated
1344 in the extended ADM1 model (Paper II) which considers TE precipitation/dissolution reactions,
1345 biouptake of TEs and disintegration of TE from organic complex matter in conjugation to
1346 biodegradation processes. The model has been implemented in form of a system of ordinary
1347 differential equations which tracks the dynamics of 90 state variables by linking the kinetics of
1348 biochemical processes with physicochemical processes which constitute the components of the
1349 proposed AD model. New inorganic components of the EDTA acid-base chemical system in
1350 aqueous solution have been defined, new complexation species resulting out of the reactions
1351 between VFAs and EDTA with TEs have been included. The charge balance has been modified
1352 accordingly to consider the effect of the additional acid-base components. Later, the complexation
1353 model has been verified by chemical equilibrium composition calculations performed by the
1354 geochemical modeling suite, Visual MINTEQ. Finally, the verified model was analyzed in 3
1355 distinct scenarios to establish the prediction capability with respect to establish a relation between
1356 TE speciation and methane production.
1357 The major accomplishment of this work (Paper III) is the ability of the proposed mathematical
1358 model to simulate the effect of addition of both TEs and EDTA in an AD system. The model is
1359 able to estimate the effect of the combined addition of TEs and EDTA and to find out the ratio
1360 which results in the maximum methane production rate. The results showcased that the addition
1361 of EDTA to the anaerobic digester with sub-optimal concentration of TE stimulates the
1362 biodegradation and methane production. In contrast, when an optimal concentration of TEs was
1363 dosed into the system, the stimulating effect of EDTA on the bio-degradation and methane
1364 production was negligible. The model simulation results are coherent with the experimental studies
1365 carried out by various research groups. Recently Thanh et al. (2017) carried out supplementation
1366 studies for only Fe2+ in presence of a green chelating agent, EDDS. It was found that the addition
1367 of EDDS at a molar ration of 1:1 to Fe2+ affected the bioavailabity of Fe2+ and improved the
1368 methane production rate by 11.5% and methane production yield by 13.25%. The study did not
1369 consider the addition of Co and Ni. This discrepancy in only dosing Fe 2+ comes from the fact that
1370 large amount of FeS is formed as compared to the NiS and CoS. Secondly, the study focused on
1371 designing a strategy for Fe supplementation because the Fe salts are cheaper as compared to Ni
1372 and Co. Another study (Zhang et al., 2015), with similar objective of improving methane
49
1373 production, used a cocktail of Fe, Co, Ni and Mo for batch and semi-continuous reactor. Dosing
1374 such a cocktail of essential TEs in addition to EDDS resulted in a 50% reduction in amount of TE
1375 supplementation. This is precisely what was observed in the model simulations where the addition
1376 of EDTA to sub-optimal concentration of TEs resulted in higher degree of methane production as
1377 compared to EDTA supplied with optimal concentration of TEs. This is a result of natural
1378 consequence as the stimulating concentration of TEs have already been achieved. Therefore, any
1379 further addition of EDTA do not result in a significant increase in methane production. It should
1380 be noted here that the dose response function in complexation model assumes both free TEs and
1381 TE-complexes as stimulatory species. However, when sub-optimal TE concentration is added,
1382 there is still room for achieving stimulatory concentration. This is fulfilled by the addition of
1383 EDTA which eventually results in a significant increase in methane production. The only point of
1384 difference is that, in our case we used EDTA as a chelating agent in place of EDDS.
1385 The model prediction efficiency was also reflected in the simulation set comprising Ca and Mg.
1386 Our study found that with increase in dosing of calcium and magnesium the production of methane
1387 decreases. It is apparent that such a change in methane production can be attributed to the increase
1388 in basic ionic species in the AD system. Here calcium and magnesium. The decrease in methane
1389 production rate can be explained by the diminished availability of carbonates in the liquid phase
1390 resulting from increase in carbonate mineral precipitation in solid phase. The results illustrate a
1391 direct relation between increase in formation of TE-complexes with EDTA addition. However,
1392 when Ca and Mg were added to the system, the amount of total mineral precipitates (carbonate,
1393 sulfide and phosphate) increases. Here, it can be inferred that increase in EDTA addition increase
1394 the solubility of the mineral phases and directs the TE fluxes towards the liquid phase. However
1395 increase in light metal concentration force TEs to stay in solid phase in form of mineral
1396 precipitates. This is an interesting observation. The model was able to capture the chelation
1397 capacity of EDTA in the AD system. In a recent study (Thanh et al., 2017) carried out for
1398 bioavailability of Fe in presence of various proportions of TEs, found out that methane production
1399 increases with addition of Ca and Mg up to 4 ×10-3 M. Unfortunately, the study did not involve
1400 the effect of Ca and Mg addition on TE-EDTA/EDDS complexation. The study focused on
1401 changes in methane production due to addition of: a) Ca+Mg ; b) Fe+Co+Ni and c) Cu+Zn+Mn.
1402 The study concluded a direct relation between addition of Ca and Mg (upto 4 ×10-3 M) to methane
1403 production. In our case, the increase in mineral precipitation, and therefore, decrease in
1404 bioavailable TEs in the AD reactor, was attributed to the inhibited rate of methane production. The
1405 point to include and investigate the effect of Ca and Mg with addition of EDTA is the fact that Ca
1406 and Mg forms major light element species in the AD system which CaCO 3 as the most widely
1407 accepted precipitate. In this regard, scrutinizing the effect of Ca and Mg in presence of EDTA may
1408 reveal the relationship between calcium precipitation and calcium complexation. Such a relation,
1409 if established by experimental results, may suggest a new check point for trouble shooting in the
1410 case of a system failure due to increased alkalinity.
50
1411 A note of caution is due here since the proposed model has not been validated with reactor or plant
1412 wide data. This is related to the unavailability of suitable sensors to detect the change in
1413 concentration of the ionic species particularly for TEs. Nevertheless, there are experimental
1414 protocols which quantify the species in discussion, but, the amount of noise associated with the
1415 experimental determination renders the data hard to incorporate in decision making processes. In
1416 this regard, quantification methods based on mechanistic mathematical models, as shown in this
1417 study, may help to overcome quantification limitations. Further, these simulation findings cannot
1418 be extrapolated to all reactor set up or to any plant wide application. To make such robust model
1419 which can be applied to all reactor set up and plant wide application, it is necessary to first
1420 scrutinize other important physicochemical processes, such as adsorption, and then selectively
1421 keep the sensitive parameters and variables while removing insignificant variables. There have
1422 been few attempts at arriving at such a reduced mathematical model (Xue et al., 2015). However,
1423 any such attempt has not been reported for models incorporating new physicochemical processes
1424 like precipitation, complexation and adsorption of TEs.
1425 There is abundant room for further development and analysis of the complexation model at
1426 different levels. At reactor operation level, the current batch AD model can be reformatted to
1427 simulate the dynamics of complexation processes for a continuous reactor operation. Here playing
1428 with residence time becomes crucial to find out possible interaction between various TE species.
1429 Further, at individual component level of the model, new complexation reactions can be
1430 incorporated. For example, addition of other chelating agents such as NTA and EDDS will give
1431 more power to the user to decide which chelating agent should be used at a particular period of
1432 reactor operation. Moreover, the relative chelating power of various chelating agents based on the
1433 ability to stimulate microbial activity can be verified and ordered.
1434 4.3 ADM1 based mathematical model for TE speciation in anaerobic digestion processes
1435 Paper IV laid down an original methodology to integrate sorption processes in the already extended
1436 ADM1 in Paper II and Paper III. It is well established that sorption of TEs on biomass and inert
1437 plays a crucial role in deciphering the dynamics of TEs in AD systems. In anaerobic digestion
1438 system the various organic matter, such as biomass have both hydrophobic and hydrophilic
1439 moieties in the cell wall (Konhauser, 2007). Such functional groups, carboxylate, hydroxyl
1440 phosphate play a major role in electrostatic interactions with the various ionic species present in
1441 the local aqueous environment (Stumm and Morgan, 1996). So, in order to decode the speciation
1442 of TEs in a given system it is necessary to consider the flux of TE participating in the sorption
1443 process. Incorporation of TE sorption in standardized waste water model has already been
1444 achieved in case of activated sludge models and for some biopharmaceuticals in the case of ADM1
1445 (Fountoulakis et al., 2006). However, there have been no attempts to incorporate sorption of TEs
1446 in the ADM1 framework. One major reason of this lack relies on the unavailability of experimental
1447 data. Secondly, the addition of sorption processes would make the ADM1 framework more
1448 complex. Nevertheless, the dynamics and speciation of TEs is a function of multiple complex
1449 physicochemical and biochemical processes taking place at different velocity. So, in order to arrive
51
1450 at a full proof model which can be later recruited for a model based control of TE dosing, the
1451 analysis of sorption processes is necessary.
1452 The TE adsorption model (Paper IV) primarily consists of components, species and process rates
1453 to model the formation of free binding sites and subsequent TE adsorption to form occupied
1454 binding sites. TE adsorption has been modeled using adsorption rate equations (3.32 and 3.33) that
1455 are solved over time for concentrations of kinetically controlled adsorption complexes. A
1456 reversible second order kinetic model has been considered for adsorption (forward) and desorption
1457 (backward). Three types of adsorption surfaces have been considered: 1) Biomass; 2) Inerts; and
1458 3) Precipitate (FeS). Seven biomass species, as originally formulated in the ADM1, are considered
1459 to participate in the adsorption/desorption reactions with all the five metals considered (TEs: Fe,
1460 Ni and Co; Light metals: Ca and Mg). Inerts have been construed to take part in
1461 adsorption/desorption reaction with Fe, Ni, Co, Mg and Ca, while, only Co and Ni have been
1462 considered to participate with precipitates (FeS). Thus, arriving at a system which constitute 35
1463 adsorption desorption reactions for biomass, 5 adsorption/desorption reactions for inert and 2 for
1464 adsorption/desorption on FeS. A structured kinetic model based on ADM1, including separate
1465 modules of TE precipitation and complexation, was implemented. The free binding sites are
1466 supposed to be produced during biomass, inert and precipitate formation. The formation of
1467 occupied binding sites is related to the TEs sorption. The biomass decay results in inert formation
1468 and TEs release from the occupied binding sites. In order to incorporate flexibility in modeling
1469 adsorption to fit varied data sets, the concept of binding site density was introduced for biomass,
1470 inert and precipitates. Further, competition of TEs for a particular free binding site was also
1471 introduced. To simplify the model, over parameterization was avoided by implementing a
1472 particular value of stability constant for all the adsorption species formed.
1473 The model sensitivity on the binding site density is investigated. The binding site density is the
1474 amount of free binding site available for binding of TEs per unit of biomass. Binding site density
1475 has been defined for the biomass, inert and precipitate. Although the chemical nature of these
1476 surfaces is distinct and varies with stage of reactor operation, the proposed model considers a
1477 singular binding site characteristic. It was found that the most significant changes in the speciation
1478 of TE have been found by varying the inert binding site density. It is interesting to note here that
1479 change in binding site density of precipitate did not produce any result in the methane production.
1480 Rather, the methane production did not change at all for change in binding site density of
1481 precipitates. The reasons for such a contradiction are multiple. First, the amount of precipitates
1482 which undergo binding with TEs is very low as compared to quantitative percentage of inert and
1483 biomass present in the system. So, less precipitate will attract less TEs for adsorption and the
1484 eventual effect on methane production is negligible. Secondly, the model considers the sorption of
1485 only Ni and Co on FeS. This might be a reason for absence of any effect on methane production.
1486 This is because, unfortunately, adsorption of Ni and Co on FeS are the only adsorption reactions
1487 on precipitate which have been reported to occur in AD. Finally, the binding site density value
1488 assumed for precipitates is lower than the binding site density for biomass and inert. This is
52
1489 realistic as organic surfaces such as in case of biomass and inert will harbor more binding site
1490 functional groups as compared to the newly formed inorganic homogenous precipitate.
1491 Similarly, simulations performed on the model established the predicting capability of the model.
1492 In scenario 1, effect of adsorbent quantity on TE adsorption in AD was studied. With increase in
1493 adsorbent amount in the anaerobic system, the amount of sulfide precipitate decreased with
1494 increase in initial complex organic matter. Moreover, sulfide fraction for Fe was higher than Ni
1495 and Co for corresponding initial complex organic matter concentration. The majority of Ni and Co
1496 was found with EDTA complexes. What is important to note here is the slow increase in TE in
1497 inert fraction with increase in initial complex organic matter concentration (adsorbent). In scenario
1498 2, effect of initial Fe, Ni and Co concentration on TE adsorption in anaerobic digestion was studied.
1499 In this scenario, sulfide fraction for Fe and Co was higher than Ni. Ni was found more with
1500 complex fraction. The amount of Fe adsorbed on inert matter is found to be higher at both lower
1501 and higher initial concentration of Fe, in comparison at optimal value of Fe. Further analysis of
1502 the fractionation pattern shows that with increase in initial Fe concentration the carbonate fraction
1503 increases. In scenario 3, effect of initial Ca-Mg concentration on TE adsorption in AD was
1504 evaluated. The simulations showcased that sulfide fraction is predominantly found with Fe. EDTA
1505 complexation fraction was found across all the simulation other than for Fe. Fe was found
1506 associated more with inert fraction than biomass fraction for lower Ca-Mg concentration. For
1507 simulation scenario 4, effect of initial EDTA concentration on TE adsorption in the AD system
1508 was studied. The model simulations showed that out of all the TE fractions Fe was found
1509 predominantly in inert fraction with lower initial concentration of EDTA.
1510 The proposed model is not full-proof. The modelling efforts required to strengthen the concept
1511 that methane production potential is a function of TE adsorption, complexation, precipitation and
1512 biouptake, may require additional model complexity in terms of significant processes and
1513 components. For an instance, consider EPS which affects the binding site density and hence the
1514 adsorption behaviour. Nevertheless, the goal of this study was to build a model framework based
1515 on ADM1 which can simulate the effect of adsorption, complexation and precipitation processes
1516 on AD. This model can be used to formulate a TE dosing scheme or a reduced version of this can
1517 be used as a model predictive control of TE dosing in AD. At this stage, with unavailability of
1518 proper experimental data to calibrate and validate such a model, it is necessary to individuate and
1519 define other biochemical and physicochemical processes (EPS and redox state) which can affect
1520 the speciation of TEs in anaerobic digesters. The use of such dynamic model based on ADM1 to
1521 predict the speciation of TEs in AD systems is still under development and some limitations and
1522 gap areas in this study can be summarized below:
1523 1) The experimental data needed to calibrate and validate the model should be communicated
1524 with the researchers performing experiments.
1525 2) Amino acids and proteins can be considered as a source for TEs.
1526 3) Redox state of the system can be modeled which can improve the TEs speciation
1527 prediction.
53
1528 4) The effect of diffusion phenomenon on TE across biofilms within the framework of ADM1
1529 may be studied to assess the role of speciation in anaerobic systems.
1530 5) Production of soluble microbial product as a part of biochemical processes can be
1531 considered.
1532 6) Incorporating a rigorous modeling approach to define and predict binding site density by
1533 taking help of surface complexation models.
1534 The proposed model has highlighted the fact that a complete unified TE speciation model based
1535 on ADM1, which exclusively includes adsorption processes, is still to be meticulously defined,
1536 developed and validated. Some of the process variables added to the TE speciation model are
1537 impossible to be determined experimentally in full-scale anaerobic reactors. The model developed
1538 in this work have been evaluated based on various numerical scenarios and the ability to mimic
1539 experimental data has been stressed. However, the absence of reactor data from a real anaerobic
1540 digester unit is a major set-back. Nevertheless, such a mechanistic model based on the actual
1541 physicochemistry of the reactor system establishes and advances the knowledge regarding the
1542 speciation of TEs in anaerobic system. In this backdrop, it is extremely difficult to prescribe any
1543 comprehensive but easily implementable integrated mathematical model framework for TE
1544 speciation for anaerobic digestion system. However, few elementary ideas can be consolidated in
1545 a full-proof TE speciation model into the standardized waste water treatment model frameworks.
1546 For example, empirical models based on black box can be used to measure unknown variables for
1547 such a complicated model. These deliberations may lead to the genesis of mechanistic-data driven
1548 hybrid model frameworks in future.
1549 4.4 Sources of gaps and potential error in modelling TE speciation in AD
1550 AD modelling is prone to challenges and bottle necks arise in the form of objective and assumption
1551 during model calibration and validation of the proposed model. In such cases uncertainty analysis
1552 provides a promising solution. The difficulty in modeling TE dynamics with regard to
1553 precipitation, complexation and adsorption are multifold. The major impediments of modeling TE
1554 speciation in the framework of ADM1 encompass: a) skillful consideration of parameters for
1555 precipitation, complexation and adsorption models; b) cautious selection of initial conditions to
1556 correctly represent a model scenario in the light of the research objective which could bring out
1557 the prediction ability of the model; c) comprehending and representing complex cascading
1558 biophysicochemical processes in simple mathematical terms employing minimum number of
1559 parameters and variables; d) availability of data; e) selection of left out potential sensitive model
1560 component and species; f) verifying model simulations with chemical equilibrium calculations; g)
1561 identifying and setting appropriate and justifiable TE stimulation and inhibition concentration
1562 range for dose response functions; h) selecting and incorporating correct chemical component and
1563 species in charge balance expression to effectively calculate pH.
1564 TE Initial Conditions
54
1565 For process models, initial conditions are generally obtained from experimental studies and
1566 existing literature. When the model carries reactions of varied time range, initial conditions play
1567 an important role in model optimization. This aspect becomes more crucial when the processes
1568 spread across the biological, chemical and physical boundaries. This makes it very challenging.
1569 For example, in our case, we have relied largely on literature and heuristics to scrutinize initial
1570 conditions for various inorganic components in the model. Further, it is difficult to measure the
1571 exact mineral/ionic content of sludge with respect to TEs. In such a scenario, a common solution
1572 is trying with many sets of initial condition with a hit and trial method. This is time consuming
1573 and may result in faulty and irrelevant model simulation scenarios.
1574 Physicochemical Model Complexity
1575 In addition to the cascading biochemical processes in AD systems, physicochemical processes are
1576 significant and complex. The complexity of the overall model reactions increases manifold when
1577 biochemical reactions are linked with physicochemical reactions. This interlinking is generally
1578 facilitated by the estimation of pH of the system. And it is well established notion that the pH of
1579 the AD system is subject to various internal and external perturbations. TE dosing is one of such
1580 desired and allowed perturbations. The ability of the model to correctly demonstrate the effect of
1581 change in concentration of a component is difficult. Likewise, considering a particular type of
1582 hydrolysis model is tricky and complex. For example, Batstone 2002 suggested use of surface-
1583 based kinetics would result in better prediction when disintegration hydrolysis are in question.
1584 Additionally, ADM1 already uses a simplified first order kinetics for such processes. Further,
1585 incorporation of release, uptake and metabolism of TE in the biochemical framework is
1586 challenging. This is primarily due to the variation in substrate type and composition.
1587 TE Speciation Data
1588 The efficiency, stability and applicability of the proposed models will significantly increase if
1589 reliable data is made accessible for the calibration, validation and training. Obtaining data set either
1590 from literature or independent experimental research is cumbersome and demanding. Particularly,
1591 when processes like precipitation, complexation and adsorption are in question. The accuracy of
1592 the model prediction will increase with increase in data availability. Most of the state variable used
1593 in original ADM1 are measurable, but it takes significant amount of time to experimentally
1594 measure such variables. As a solution to this problem, all the unknowns should be considered in
1595 the model parameter set. Then reliable data sets from experiments should be used to calibrate and
1596 validate the proposed model to estimate the rest of the unknowns. Online data monitoring systems
1597 can be recruited to produce data in real time. VFAs, pH, sulfate, phosphate and carbonate species
1598 data can be generated. Moreover, software sensors can be employed to trace the dynamics of
1599 process parameters which are critical to precipitation/dissolution, complexation and adsorption
1600 processes in AD. Software sensors provide indirect measurements and are a growing trend in AD
1601 system monitoring and analysis but it has not yet been applied in the case of TE speciation.
1602 Currently, Extended Kalman filters, Extended Luenberger observers, Adaptive observers and
55
1603 Asymptotic observers are widely used software sensors techniques in AD process monitoring.
1604 Further, research collaborations with an agenda of data sharing may address the inaccessibility and
1605 scarcity of data sets required for various new avenues carried out in modeling AD processes.
1606 TE Parameter Interpretation
1607 The Ksp values as required in precipitation model are available and has been sufficiently used in
1608 our model. Similarly, the stability constants for various complexation species are widely available.
1609 However, kinetic rate constants for all these processes have been assumed based on experience
1610 and literature. The kinetic rate constants for the precipitation, complexation and adsorption
1611 processes have been defined based on the fact that complexation and adsorption are faster
1612 processes as compared to precipitation. This assumption masks any potential outlier which can be
1613 evaluated. Increasing the number of parameters increases the freedom of model fitting to
1614 experimental data. Also, an increase in number of parameters with limited data results in
1615 overfitting. Therefore, as a meaningful practice in modeling procedure, particularly in the case of
1616 AD, any new addition of parameter should have a strong underlying mechanistic theory. Such a
1617 practice will enhance the speed and accuracy for model parameter calibration.
1618 Modelling new TE component and species
1619 The primary rationality of undertaking modelling of physicochemical processes in AD with respect
1620 to TE stems from the fact that AD modelling necessarily do not consider all important processes
1621 which are essential for model prediction. For example, immediately after ADM1 release, it was
1622 felt that solid state reactions involving precipitations have largely been ignored. Whether this
1623 affects the model prediction, is all together a different discussion. But the ability of a mathematical
1624 tool to make analysis and predictions for unrelated variables was significantly lowered. Higher
1625 number of processes and features on the ADM1 may result in a better analysis and may expand
1626 the current knowledge regarding the interaction of physicochemical and biochemical processes
1627 and the ways to control it. For example, the lactic acid pathway is not taken into consideration so
1628 was the precipitation processes in the original ADM1. In our case, efforts have been made to
1629 correctly identify important process variables and significant parameters to be included in the
1630 model structure. These new species and components are essential to define the physicochemistry
1631 of TEs. However, it should be appreciated that the modelling dynamics of any new species depends
1632 on the model objective and complexity of the biophysicochemical interactions involved. In this
1633 case it can be stated that, even though the proposed ADM1 based models are all-inclusive, they
1634 require further model developments.
1635
1636
1637
56
1638
1639 CHAPTER 5. CONCLUSION AND OUTLOOK

1640 The main objective of the research work presented in this thesis was to investigate the dynamics,
1641 speciation and role of TEs in AD systems by developing a new mathematical model able to take
1642 into account all these factors. The widely adopted ADM1 neglects the effects of TEs
1643 precipitation/dissolution, complexation and adsorption on methane production, as well as TE
1644 biouptake, release and dose response. On the other hand, standalone chemical equilibrium models
1645 predict only the steady state speciation of TEs in AD environment and mostly neglect the
1646 interactions between the biochemical and physicochemical compartments. The models developed
1647 in this thesis fall between these two extremes: they have been formulated in the framework of
1648 ADM1 for that concerns the biochemical compartment, and make use of thermodynamic
1649 parameters (solubility products, stability constants for complexes and adsorption) to predict the
1650 dynamic fate of TEs in AD systems.
1651 The model presented in Paper II is a structural kinetic model for TE precipitation/dissolution with
1652 a hormesis type TE dose-response effect on microbial activity. The model takes into account
1653 multiple mineral precipitation reactions, chemical ionization reactions for pH calculation,
1654 biodegradation reactions and liquid-gas transfer processes for gas components. The model analysis
1655 conducted in Paper I showed model capability to qualitatively predict the TE dynamics in an AD
1656 system, when the precipitation/dissolution reactions predominate over the overall physicochemical
1657 processes. TEs will exist in free, carbonate, phosphate and sulfide form. Furthermore, the
1658 numerical simulations showed that sulfide components have higher potential to form precipitate
1659 and dominate the AD system at neutral pH, even if carbonate and phosphate are present in higher
1660 quantity. This modelling outcome corelates with the experimental experiences in which sulfide
1661 precipitates have been confirmed to be a process destabilizing factor in the normal course of reactor
1662 operation. The instability caused is largely due to the fluctuation in the bio-geo-physicochemistry
1663 of TEs which arises due to the increase in sulfide precipitates. The model developed in Paper II
1664 was further extended in Paper III to include additional physicochemical processes. The numerical
1665 results revealed that the incorporation of TE complexation reactions is significant for a proper
1666 description of the multitude of processes involved in AD and provides useful insights in TEs
1667 dynamics, contributing to the definition of a more efficient TE dosing strategy. The numerical
1668 simulations affirmed that dosing of EDTA with TE is beneficial when EDTA is dosed with
1669 suboptimal TE dosing. Dosing EDTA with optimal TE initial concentrations does not produce any
1670 significant change in methane production. Verification of precipitation/dissolution and
1671 complexation sub-models by a chemical equilibrium calculation software was carried out in Paper
1672 II and Paper III respectively. The time limit for model prediction was set to a higher value to mimic
1673 a steady state behavior of the system. A higher degree of correspondence was observed between
1674 the MINTEQ chemical equilibrium calculations and the model predictions. Due to presence of
1675 large amount of organic and inorganic ligands (biomass, inert and mineral precipitates), TE
57
1676 speciation should be considered as a function of adsorption processes in addition to
1677 precipitation/dissolution, biouptake and complexation. The model presented in Paper IV
1678 numerically studied the effect of adsorption processes on TE dynamics and eventually methane
1679 production, by varying the amount of complex organic matter, TEs, Ca/Mg, EDTA and binding
1680 site density. The numerical studies confirmed the adaptability of the proposed model to a varied
1681 range of binding site density. However, some elements of uncertainty are related to the absence of
1682 adsorption constants for various species considered and to the arbitrary of the initial conditions
1683 which are generally difficult to estimate experimentally.
1684 Mathematical models are a valuable tool in AD research. They offer a window of opportunity in
1685 the form of qualitative and or/quantitative analysis and at the same time provide information on
1686 the overall reactor performance with respect to any specific process parameter or process variable.
1687 Mathematical modeling exercise builds and extends a model framework within which progressive
1688 modification and developments can be carried out. The models presented in this thesis focused and
1689 traced the significance of TE speciation in AD and focused on the TEs effect on methane
1690 production. The latter is generally easy to access on a reactor level study. However, to evaluate
1691 relationship between TE speciation and cumulative methane production is hard task if conducted
1692 merely on an experimental basis. The modeling exercise proved that new process parameters and
1693 variables for defining TE speciation can be incorporated in the well-established ADM1 framework.
1694 With TE dynamics in the center of the study, many interesting model applications can be
1695 reimagined and evaluated. The modeling community interested in TE dynamics in AD is new but
1696 growing, which would benefit from further model developments. Mechanistic mathematical
1697 models in conjugation with experimental experience will help to consolidate and/or improve the
1698 existing understanding of the fate of TEs in AD systems. Such hybrid models will have the
1699 potential to solve intricate process instability problems associated with the physicochemistry of
1700 TEs. However, this is rarely a straightforward task.
1701 The two fundamental questions in TE speciation modeling in the ADM1 framework are the
1702 complexity of biogeochemical processes occurring inside an AD reactor, and the availability of
1703 experimental data for model calibration and validation. TEs are subject to multiple simultaneous
1704 processes which determine their fate in anaerobic environments. An older school of thought
1705 attributed the effect of TE on AD to the total amount of TE available in the reactor system.
1706 However, a new school of thought have emerged in which free TEs have been given the central
1707 role in deciphering the TE effect on microbial activity. Nevertheless, precipitation/dissolution,
1708 complexation and adsorption processes in AD are largely under investigation. Mathematical
1709 modelling of such processes is close to non-existent. Many researchers use various experimental
1710 techniques in conjugation to steady state chemical equilibrium calculations to decipher the exact
1711 interactions and physicochemistry of TE. Such efforts are futile if a consolidated strategy and
1712 guideline which can work better at different reactor operating conditions cannot be developed. In
1713 this work, TEs physicochemistry has been described as a collective function of
1714 precipitation/dissolution, adsorption, complexation, and biouptake in the form of chemical kinetic
58
1715 reactions based on components and species. However, such a detailed approach to
1716 physicochemistry of TE lacks the availability of experimental validation at this stage.
1717 Although the models presented in the Paper II, Paper III and Paper IV, are representative of
1718 precipitation/dissolution, complexation and adsorption, there are multiple areas of improvement.
1719 A first model improvement is correlated to the incorporation of the ionic activity. Davis
1720 approximation, which is widely used in geochemical equilibrium models may be used for activity
1721 correction. Davis method of correction is suitable for the range of ionic strength generally
1722 encountered in AD environments. The model estimation for concentration can be easily corrected
1723 by multiplying or dividing with equilibrium coefficients with activity coefficients. In case of
1724 higher ionic strengths (I > 3) Debyee-Huckel approach of approximation to ionic activity may be
1725 used.
1726 A natural extension of the present study would be to consider in the numerical simulations of a
1727 continuous reactor operation instead of a batch system. To the best of my knowledge, there has
1728 been no work in this regard. Some plant wide modeling approaches mainly focused on resource
1729 recovery have been developed, which incorporate precipitation/dissolution reactions, but such
1730 models are primarily restricted to Fe and neglect Ni and Co dynamics. Modelling TE dynamics in
1731 continuous reactors will allow the investigation into the nature, role and outcome of perturbations
1732 caused by sudden addition/removal of TEs in an AD system. Eventually, such a continuous version
1733 of the proposed model may be used to recruit a model predictive control for TE dosing. An
1734 inquisitive look into the methods from the field of data science could be fruitful in the development
1735 and application of a mechanistic-data driven hybrid model. An additional area of improvement is
1736 related to the calibration and validation of the proposed models. Adequate amount of data sets may
1737 be experimentally generated and or extracted from appropriate literature. These data sets should
1738 essentially specify the effect of TE dose on pH and methane production. Further, uncertainty
1739 analysis on the data sets may be performed. The model calibration should predict parameter sets
1740 with better accuracy which can explain all the data trends. As a starting point for model calibration,
1741 the kinetic parameters newly introduced for the precipitation, complexation and adsorption
1742 processes should be investigated through a local and or sensitivity analysis.
1743 Process optimization is not new to wastewater engineering but designing an “intelligent” dosing
1744 strategy based on the speciation of TEs in AD system will require skills from conventional fields
1745 of mathematics, biology, biogeochemistry, computer science and informatics, bioprocess
1746 engineering, process optimization as well as emerging fields of statistical modeling and machine
1747 learning. Such a concerted effort can be made efficient by following the FAIR (Findable,
1748 Accessible, Interoperable and Reusable) principles of Research Data Alliance, which bolsters the
1749 awareness of all the stakeholders. The probable solution of TE speciation and bioavailability in
1750 AD environments provides one of many perfect departure points towards, “data science in
1751 environmental engineering”.
1752
59
1753 6. BIBILIOGRAPHY
1754 Abdelgadir, A., Chen, X., Liu, J., Xie, X., Zhang, J., Zhang, K., Wang, H., Liu, N., 2014.
1755 Characteristics, process parameters, and inner components of anaerobic bioreactors. Biomed
1756 Res. Int. 2014. https://doi.org/10.1155/2014/841573
1757 Ahring, B.K., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acids as indicators of process
1758 imbalance in anaerobic digestors. Appl. Microbiol. Biotechnol. 43, 559–565.
1759 https://doi.org/10.1007/BF00218466
1760 Allison, J.D., Brown, D.S., Novo-Gradac, K.J., 1991. Minteqa2/Prodefa2, a Geochemical
1761 Assessment Model for Environmental Systems: Version 3.0 User’S Manual. U. S. Environ.
1762 Prot. Agency EPA/600/3-91/021 115.
1763 Appels, L., Baeyens, J., Degrève, J., Dewil, R., 2008. Principles and potential of the anaerobic
1764 digestion of waste-activated sludge. Prog. Energy Combust. Sci. 34, 755–781.
1765 https://doi.org/10.1016/j.pecs.2008.06.002
1766 Appels, L., Lauwers, J., Degrve, J., Helsen, L., Lievens, B., Willems, K., Van Impe, J., Dewil,
1767 R., 2011. Anaerobic digestion in global bio-energy production: Potential and research
1768 challenges. Renew. Sustain. Energy Rev. 15, 4295–4301.
1769 https://doi.org/10.1016/j.rser.2011.07.121
1770 Aquino, F., Stuckey, D.C., 2008. Integrated model of the production of soluble microbial
1771 products ( SMP ) and extracellular polymeric substances ( EPS ) in anaerobic chemostats
1772 during transient conditions 38, 138–146. https://doi.org/10.1016/j.bej.2007.06.010
1773 Aquino, S.F., Stuckey, D.C., 2007. Bioavailability and Toxicity of Metal Nutrients during
1774 Anaerobic Digestion. J. Environ. Eng. 133, 28–35. https://doi.org/10.1061/(ASCE)0733-
1775 9372(2007)133:1(28)
1776 Aquino, S.F., Stuckey, D.C., 2004. Soluble microbial products formation in anaerobic
1777 chemostats in the presence of toxic compounds. Water Res. 38, 255–266.
1778 https://doi.org/10.1016/j.watres.2003.09.031
1779 Ball, J.W., 1991. WATEQ4F Database, The United States Geological Survey (USGS), Open-File
1780 Report 91–183 (2001). U.S. Geol. Surv. Open-File.
1781 https://doi.org/10.1107/S1600536811054717
1782 Banerjee, R., Ragsdale, S.W., 2003. The many faces of vitamin B12. Annu. Rev. Biochem. 72,
1783 209–247. https://doi.org/10.1146/annurev.biochem.72.121801.161828
1784 Barat, R., Montoya, T., Seco, A., Ferrer, J., 2011. Modelling biological and chemically induced
1785 precipitation of calcium phosphate in enhanced biological phosphorus removal systems.
1786 Water Res. 45, 3744–3752. https://doi.org/10.1016/j.watres.2011.04.028
1787 Barat, R, Montoya, T., Seco, A., Ferrer, J., 2011. Modelling biological and chemically induced
1788 precipitation of calcium phosphate in enhanced biological phosphorus removal systems 5,
1789 0–8. https://doi.org/10.1016/j.watres.2011.04.028
60
1790 Bartacek, J., Fermoso, F.G., Lens, P.N.L., 2008. Cobalt toxicity in anaerobic granular sludge :
1791 influence of chemical speciation. J. Ind. Microbiol. Biotechnol. 35, 1465–1474.
1792 https://doi.org/10.1007/s10295-008-0448-0
1793 Batstone, D.J., Amerlinck, Y., Ekama, G., Goel, R., Grau, P., Johnson, B., Kaya, I., Steyer, J.P.,
1794 Tait, S., Takacs, I., Vanrolleghem, P.A., Brouckaert, C.J., Volcke, E., 2012. Towards a
1795 generalized physicochemical framework. Water Sci. Technol. 66, 1147–1161.
1796 https://doi.org/10.2166/wst.2012.300
1797 Batstone, D.J., Keller, J., 2003. Industrial applications of the IWA anaerobic digestion model
1798 No.1 (ADM1). Water Sci. Technol. 47, 199–206.
1799 Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S. V., Pavlostathis, S.G., Rozzi, A.,
1800 Sanders, W.T., Siegrist, H., Vavilin, V.A., 2002. The IWA Anaerobic Digestion Model No
1801 1 (ADM1). Water Sci. Technol. 45, 65–73. https://doi.org/10.2166/wst.2008.678
1802 Batstone, D.J., Keller, J., Steyer, J.P., 2006. A review of ADM1 extensions, applications, and
1803 analysis: 2002-2005. Water Sci. Technol. 54, 1–10. https://doi.org/10.2166/wst.2006.520
1804 Batstone, D.J., Puyol, D., Flores-Alsina, X., Rodríguez, J., 2015. Mathematical modelling of
1805 anaerobic digestion processes: applications and future needs. Rev. Environ. Sci. Biotechnol.
1806 14, 595–613. https://doi.org/10.1007/s11157-015-9376-4
1807 Callander, I.J., Barford, J.P., 1983a. Precipitation, chelation, and the availability of metals as
1808 nutrients in anaerobic digestion. I. Methodology. Biotechnol. Bioeng. XXV, 1947–1957.
1809 https://doi.org/10.1002/bit.260250806
1810 Callander, I.J., Barford, J.P., 1983b. Precipitation, chelation, and the availability of metals as
1811 nutrients in anaerobic digestion. II. Applications. Biotechnol. Bioeng. XXV, 1959–1972.
1812 https://doi.org/10.1002/bit.260250806
1813 Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: A review.
1814 Bioresour. Technol. 99, 4044–4064. https://doi.org/10.1016/j.biortech.2007.01.057
1815 Choong, Y.Y., Ismail, N., Abdullah, A.Z., Yhaya, M.F., 2016. Impacts of trace element
1816 supplementation on the performance of anaerobic digestion process: A critical review.
1818 Crawford, J., 1999. Geochemical Modelling – A Review of Current Capabilities and Future
1819 Directions. Technology.
1820 Dai, X., Hu, C., Zhang, D., Dai, L., Duan, N., 2017. Impact of a high ammonia-ammonium-pH
1821 system on methane-producing archaea and sulfate-reducing bacteria in mesophilic
1822 anaerobic digestion. Bioresour. Technol. 245, 598–605.
1823 https://doi.org/10.1016/j.biortech.2017.08.208
1824 Daughney, C.J., Fein, J.B., 1998. The Effect of Ionic Strength on the Adsorption of H+, Cd2+,
1825 Pb2+, and Cu2+ by Bacillus subtilis and Bacillus licheniformis: A Surface Complexation
1826 Model. J. Colloid Interface Sci. 198, 53–77. https://doi.org/10.1006/jcis.1997.5266
61
1827 Demirel, B., Scherer, P., 2011. Trace element requirements of agricultural biogas digesters
1828 during biological conversion of renewable biomass to methane. Biomass and Bioenergy 35,
1829 992–998. https://doi.org/10.1016/j.biombioe.2010.12.022
1830 Dong, B., Liu, X., Dai, L., Dai, X., 2013. Changes of heavy metal speciation during high-solid
1831 anaerobic digestion of sewage sludge. Bioresour. Technol. 131, 152–8.
1833 Droop, M.R., 1983. 25 Years of Algal Growth Kinetics A Personal View XXVI, 99–112.
1834 Esposito, G., Frunzo, L., Panico, A., Pirozzi, F., 2012. Enhanced bio-methane production from
1835 co-digestion of different organic wastes. Environ. Technol. 1–8.
1836 https://doi.org/10.1080/09593330.2012.676077
1837 Fabiyi, M., Novak, R.A., Jampala, S., Shah, P., 2015. Enhanced biogas production using ozone –
1838 lysis reactor design and integration with anaerobic digestion of biosolids. Proc. Water
1839 Environ. Fed. 2014, 1–10. https://doi.org/10.2175/193864714816196754
1840 Federovich, V., Lens P., Kalyuzhnyi S., 2003. Extension of Anaerobic Digestion Model No . 1.
1841 Appl. Biochem. Biotechnol. 109, 33–45. https://doi.org/10.1385/ABAB:109:1-3:33
1842 Feng, X.M., Karlsson, A., Svensson, B.H., Bertilsson, S., 2010. Impact of trace element addition
1843 on biogas production from food industrial waste - Linking process to microbial
1844 communities. FEMS Microbiol. Ecol. 74, 226–240. https://doi.org/10.1111/j.1574-
1845 6941.2010.00932.x
1846 Fermoso, F.G., Bartacek, J., Ching, L., Lens, P., 2008a. Supplementation of cobalt to UASB
1847 reactors by pulse dosing : CoCl2 versus CoEDTA2− pulses. Biochem. Eng. J. 42, 111–119.
1848 https://doi.org/10.1016/j.bej.2008.06.005
1849 Fermoso, F.G., Bartacek, J., Jansen, S., Lens, P.N.L., 2009. Metal supplementation to UASB
1850 bioreactors: from cell-metal interactions to full-scale application. Sci. Total Environ. 407,
1851 3652–3667. https://doi.org/10.1016/j.scitotenv.2008.10.043
1852 Fermoso, F.G., Bartacek, J., Jansen, S., Lens, P.N.L., 2008b. Metal supplementation to UASB
1853 bioreactors : from cell-metal interactions to full-scale application. Sci. Total Environ. 407,
1854 3652–3667. https://doi.org/10.1016/j.scitotenv.2008.10.043
1855 Fermoso, F.G., Collins, Æ.G., Bartacek, J., Flaherty, Æ.V.O., Lens, Æ.P., 2008c. Role of nickel
1856 in high rate methanol degradation in anaerobic granular sludge bioreactors 725–737.
1857 https://doi.org/10.1007/s10532-008-9177-3
1858 Fermoso, F.G., Hullebusch, E.D. Van, Guibaud, G., Collins, G., Svensson, B.H., Vink, J.P.M.,
1859 Esposito, G., Frunzo, L., 2015. Fate of Trace Metals in Anaerobic Digestion, in: Gubitz, G.,
1860 Bauer, A., Bochmann, G., Gronauer, A., Weiss, S. (Eds.), Biogas Science and
1861 Technology.Advances in Biochemical Engineering/Biotechnology. pp. 171–195.
1862 https://doi.org/10.1007/978-3-319-21993-6
1863 Fermoso, F.G., Roussel, J., Mucha, A.P., Esposito, G., 2019. Trace Elements in Anaerobic
1864 Biotechnologies, Trace Elements in Anaerobic Biotechnologies. IWA Publishing.
62
1865 https://doi.org/10.2166/9781789060225
1866 Fezzani, B., Ben Cheikh, R., 2009. Extension of the anaerobic digestion model No. 1 (ADM1) to
1867 include phenol compounds biodegradation processes for simulating the anaerobic co-
1868 digestion of olive mill wastes at mesophilic temperature. J. Hazard. Mater. 172, 1430–8.
1869 https://doi.org/10.1016/j.jhazmat.2009.08.017
1870 Filbert, J.W., W, J., 2011. Procedures and Problems of Digester Startup. Water 39, 367–372.
1871 Filgueiras, A. V., Lavilla, I., Bendicho, C., 2002. Chemical sequential extraction for metal
1872 partitioning in environmental solid samples. J. Environ. Monit. 4, 823–857.
1873 https://doi.org/10.1039/b207574c
1874 Finazzo, C., Harmer, J., Bauer, C., Jaun, B., Duin, E.C., Mahlert, F., Goenrich, M., Thauer, R.K.,
1875 Van Doorslaer, S., Schweiger, A., 2003. Coenzyme B induced coordination of coenzyme M
1876 via its thiol group to Ni(I) of F430 in active methyl-coenzyme M reductase. J. Am. Chem.
1877 Soc. 125, 4988–9. https://doi.org/10.1021/ja0344314
1878 Flores-Alsina, X., Solon, K., Kazadi Mbamba, C., Tait, S., Gernaey, K. V., Jeppsson, U.,
1879 Batstone, D.J., 2016a. Modelling phosphorus (P), sulfur (S) and iron (Fe) interactions for
1880 dynamic simulations of anaerobic digestion processes. Water Res. 95, 370–382.
1882 Flores-Alsina, X., Solon, K., Kazadi Mbamba, C., Tait, S., Gernaey, K. V., Jeppsson, U.,
1883 Batstone, D.J., 2016b. Modelling phosphorus (P), sulfur (S) and iron (Fe) interactions for
1884 dynamic simulations of anaerobic digestion processes. Water Res.
1886 Fountoulakis, M.S., Stamatelatou, K., Batstone, D.J., Lyberatos, G., 2006. Simulation of DEHP
1887 biodegradation and sorption during the anaerobic digestion of secondary sludge. Water Sci.
1888 Technol. 54, 119–128. https://doi.org/10.2166/wst.2006.533
1889 Fricke, K., Santen, H., Wallmann, R., Hüttner, A., Dichtl, N., 2007. Operating problems in
1890 anaerobic digestion plants resulting from nitrogen in MSW. Waste Manag. 27, 30–43.
1891 https://doi.org/10.1016/j.wasman.2006.03.003
1892 Galí, A., Benabdallah, T., Astals, S., Mata-Alvarez, J., 2009. Modified version of ADM1 model
1893 for agro-waste application. Bioresour. Technol. 100, 2783–2790.
1895 Gavala, H.N., Angelidaki, I., Ahring, B.K., 2003. Kinetics and Modeling of Anaerobic Digestion
1896 Process. Adv. Biochem. Eng. Biotechnol. 81, 57–91.
1897 Glass, J.B., Orphan, V.J., 2012. Trace metal requirements for microbial enzymes involved in the
1898 production and consumption of methane and nitrous oxide. Front. Microbiol. 3, 1–20.
1899 https://doi.org/10.3389/fmicb.2012.00061
1900 Gonzalez-Gil, G., Jansen, S., Zandvoort, M.H., Van Leeuwen, H.P., 2003. Effect of yeast extract
1901 on speciation and bioavailability of nickel and cobalt in anaerobic bioreactors. Biotechnol.
1902 Bioeng. 82, 134–142. https://doi.org/10.1002/bit.10551
63
1903 Gonzalez-gil, G., Kleerebezem, R., 1999. Effects of Nickel and Cobalt on Kinetics of Methanol
1904 Conversion by Methanogenic Sludge as Assessed by On-Line CH 4 Monitoring Effects of
1905 Nickel and Cobalt on Kinetics of Methanol Conversion by Methanogenic Sludge as
1906 Assessed by On-Line CH 4 Monitoring 65, 1789–1793.
1907 Guendouz, J., Buffière, P., Cacho, J., Carrère, M., Delgenes, J.-P., 2010. Dry anaerobic digestion
1908 in batch mode: Design and operation of a laboratory-scale, completely mixed reactor. Waste
1909 Manag. 30, 1768–1771. https://doi.org/10.1016/j.wasman.2009.12.024
1910 Gustavsson, J., 2012. Cobalt and Nickel Bioavailability for Biogas Formation, PhD Thesis,
1911 University of Linkoping, Sweden.
1912 Gustavsson, J., Shakeri Yekta, S., Sundberg, C., Karlsson, A., Ejlertsson, J., Skyllberg, U.,
1913 Svensson, B.H., 2013. Bioavailability of cobalt and nickel during anaerobic digestion of
1914 sulfur-rich stillage for biogas formation. Appl. Energy 112, 473–477.
1915 https://doi.org/10.1016/j.apenergy.2013.02.009
1916 Hao, T. wei, Xiang, P. yu, Mackey, H.R., Chi, K., Lu, H., Chui, H. kwong, van Loosdrecht,
1917 M.C.M., Chen, G.H., 2014. A review of biological sulfate conversions in wastewater
1918 treatment. Water Res. 65, 1–21. https://doi.org/10.1016/j.watres.2014.06.043
1919 Hauduc, H., Smith, S., Szabo, A., Murthy, S., Tak, I., 2015. A dynamic physicochemical model
1920 for chemical phosphorus removal 3.
1921 He, X.T., Traina, S.J., Logan, T.J., 1992. Chemical properties of municipal solid waste
1922 composts. J. Environ. Qual. 21, 318–329.
1923 https://doi.org/10.2134/jeq1992.00472425002100030003x
1924 Heijerick, D.G., De Schamphelaere, K.A.C., Janssen, C.R., 2002. Biotic ligand model
1925 development predicting Zn toxicity to the alga Pseudokirchneriella subcapitata: Possibilities
1926 and limitations. Comp. Biochem. Physiol. - C Toxicol. Pharmacol. 133, 207–218.
1927 https://doi.org/10.1016/S1532-0456(02)00077-7
1928 Hickey, R.F., Vanderwielen, J., Switzenbaum, M.S., 1989. The effect of heavy metals on
1929 methane production and hydrogen and carbon monoxide levels during batch anaerobic
1930 sludge digestion. Water Res. 23, 207–218. https://doi.org/10.1016/0043-1354(89)90045-6
1931 Hille, R. Van, Foster, T., Storey, A., Duncan, J., 2004. Heavy metal precipitation by sulphide and
1932 bicarbonate : evaluating methods to predict anaerobic digester overflow performance.
1933 Jarvis, AP, Dudgeon, BA Younger, P. L. mine water 2004–Proceedings Int. Mine Water
1934 Assoc. Symp. 2, 141–150.
1935 Hu, Q.H., Li, X.F., Liu, H., Du, G.C., Chen, J., 2008. Enhancement of methane fermentation in
1936 the presence of Ni2+ chelators. Biochem. Eng. J. 38, 98–104.
1937 https://doi.org/10.1016/j.bej.2007.07.002
1938 Hullebusch, E.D. Van, Guibaud, G., Simon, S., Lenz, M., Yekta, S.S., Fermoso, F.G., Jain, R.,
1939 Duester, L., Roussel, J., Guillon, E., Skyllberg, U., Almeida, C.M.R., Garuti, M., Frunzo,
1940 L., Esposito, G., Carliell-, C., Ortner, M., Collins, G., 2016. Methodological approaches for
64
1941 fractionation and speciation to estimate trace element bioavailability in engineered
1942 anaerobic digestion ecosystems : An overview. Crit. Rev. Environ. Sci. Technol. 46, 1324–
1943 1366. https://doi.org/10.1080/10643389.2016.1235943
1944 Jansen, S., 2004. Speciation and bioavailability of cobalt and nickel in anaerobic wastewater
1945 treatment, Leerstoelgroep Fysische Chemie en Kolloidkunde.
1946 Jansen, S., Gonzalez-gil, G., Leeuwen, H.P. Van, 2007. The impact of Co and Ni speciation on
1947 methanogenesis in sulfidic media — Biouptake versus metal dissolution 40, 823–830.
1948 https://doi.org/10.1016/j.enzmictec.2006.06.019
1949 Jiang, B., 2006. The effect of trace elements on the metabolism of methanogenic consortia.
1950 Jimenez, J., Latrille, E., Harmand, J., Robles, A., Ferrer, J., Gaida, D., Wolf, C., Mairet, F.,
1951 Bernard, O., Alcaraz-Gonzalez, V., Mendez-Acosta, H., Zitomer, D., Totzke, D., Spanjers,
1952 H., Jacobi, F., Guwy, A., Dinsdale, R., Premier, G., Mazhegrane, S., Ruiz-Filippi, G., Seco,
1953 A., Ribeiro, T., Pauss, A., Steyer, J.P., 2015. Instrumentation and control of anaerobic
1954 digestion processes: a review and some research challenges. Rev. Environ. Sci. Biotechnol.
1955 14, 615–648. https://doi.org/10.1007/s11157-015-9382-6
1956 Kaksonen, A.H., Riekkola-Vanhanen, M.L., Puhakka, J.A., 2003. Optimization of metal sulphide
1957 precipitation in fluidized-bed treatment of acidic wastewater. Water Res. 37, 255–266.
1958 https://doi.org/10.1016/S0043-1354(02)00267-1
1959 Kazadi Mbamba, C., Batstone, D.J., Flores-Alsina, X., Tait, S., 2015a. A generalised chemical
1960 precipitation modelling approach in wastewater treatment applied to calcite. Water Res. 68,
1961 342–353. https://doi.org/10.1016/j.watres.2014.10.011
1962 Kazadi Mbamba, C., Tait, S., Flores-Alsina, X., Batstone, D.J., 2015b. A systematic study of
1963 multiple minerals precipitation modelling in wastewater treatment. Water Res. 85, 359–370.
1965 Khalid, A., Arshad, M., Anjum, M., Mahmood, T., Dawson, L., 2011. The anaerobic digestion of
1966 solid organic waste. Waste Manag. 31, 1737–1744.
1967 https://doi.org/10.1016/j.wasman.2011.03.021
1968 Kleerebezem, R., van Loosdrecht, M.C.M., 2006. Critical analysis of some concepts proposed in
1969 ADM1. Water Sci. Technol. 54, 51. https://doi.org/10.2166/wst.2006.525
1970 Kong, X., Wei, Y., Xu, S., Liu, J., Li, H., Liu, Y., Yu, S., 2016. Inhibiting excessive acidification
1971 using zero-valent iron in anaerobic digestion of food waste at high organic load rates.
1973 Konhauser, K., 2007. Introduction to geomicrobiology, European Journal of Soil Science.
1974 https://doi.org/10.1111/j.1365-2389.2007.00943_4.x
1975 Koutsoukos, P., Amjad, Z., Tomson, M.B., Nancollas, G.H., 1980. Calcium Phosphates. A
1976 Constant Composition Study. J. Am. Chem. Soc. 102, 1553–1557.
1977 https://doi.org/10.1021/ja00525a015
65
1978 Kumar, A.G., Kirubagaran, R., Nagesh, N., Rubio, J.P., Sekaran, G., 2010. Influence of
1979 fermentation metabolites on redox potential in anaerobic digestion of proteinaceous solid
1980 wastes by Synergistes sp. Eng. Life Sci. 10, 368–373.
1981 https://doi.org/10.1002/elsc.200900057
1982 Lauwers, J., Appels, L., Thompson, I.P., Degrève, J., Impe, J.F. Van, Dewil, R., 2013.
1983 Mathematical modelling of anaerobic digestion of biomass and waste : Power and
1984 limitations. Prog. Energy Combust. Sci. 39, 383–402.
1985 https://doi.org/10.1016/j.pecs.2013.03.003
1986 Lizarralde, I., Fernández-Arévalo, T., Brouckaert, C., Vanrolleghem, P., Ikumi, D.S., Ekama,
1987 G.A., Ayesa, E., Grau, P., 2015a. A new general methodology for incorporating physico-
1988 chemical transformations into multi-phase wastewater treatment process models. Water
1989 Res. 74, 239–256. https://doi.org/10.1016/j.watres.2015.01.031
1990 Lizarralde, I., Fernández-Arévalo, T., Brouckaert, C., Vanrolleghem, P., Ikumi, D.S., Ekama,
1991 G.A., Ayesa, E., Grau, P., 2015b. A new general methodology for incorporating physico-
1992 chemical transformations into multi-phase wastewater treatment process models. Water
1993 Res. 74, 239–256. https://doi.org/10.1016/j.watres.2015.01.031
1994 Lokshina, L.Y., Vavilin, V.A., Kettunen, R.H., Rintala, J.A., Holliger, C., Nozhevnikova, A.N.,
1995 2001. Evaluation of kinetic coefficients using integrated monod and haldane models for
1996 low-temperature acetoclastic methanogenesis. Water Res. 35, 2913–2922.
1997 https://doi.org/10.1016/S0043-1354(00)00595-9
1998 Maharaj, B.C., Mattei, M.R., Frunzo, L., Hullebusch, E.D. va., Esposito, G., 2019. ADM1 based
1999 mathematical model of trace element complexation in anaerobic digestion processes.
2000 Bioresour. Technol. 253–259. https://doi.org/10.1016/j.biortech.2018.12.064
2001 Maharaj, B.C., Mattei, M.R., Frunzo, L., van Hullebusch, E.D., Esposito, G., 2018. ADM1 based
2002 mathematical model of trace element precipitation/dissolution in anaerobic digestion
2003 processes. Bioresour. Technol. 267, 666–676.
2005 Mancini, G., Papirio, S., Riccardelli, G., Lens, P.N.L., Esposito, G., 2018. Trace elements dosing
2006 and alkaline pretreatment in the anaerobic digestion of rice straw. Bioresour. Technol. 247,
2007 897–903. https://doi.org/10.1016/j.biortech.2017.10.001
2008 Mata-álvarez, J., 2015. Anaerobic Digestion of Organic Solid Wastes . An Overview of Research
2009 Achievements and Perspectives. https://doi.org/10.1016/S0960-8524(00)00023-7
2010 Maurer, M., Boller, M., 1999. Modelling of phosphorus precipitation in wastewater treatment
2011 plants with enhanced biological phosphorus removal. Water Sci. Technol. 39, 147–163.
2012 https://doi.org/10.1016/S0273-1223(98)00784-7
2013 Mazumder, T.K., Nishio, N., Fukuzaki, S., Nagai, S., 1987. Production of extracellular vitamin
2014 B-12 compounds from methanol by Methanosarcina barkeri. Appl. Microbiol. Biotechnol.
2015 26, 511–516. https://doi.org/10.1007/BF00253023
66
2016 Mbamba, K., Christian, Tait, S., Flores-Alsina, X., Batstone, D.J., 2015. A systematic study of
2017 multiple minerals precipitation modelling in wastewater treatment. Water Res. 85, 359–370.
2019 McCartney, D.M., Oleszkiewicz, J.A., 1991. Sulfide inhibition of anaerobic degradation of
2020 lactate and acetate. Water Res. 25, 203–209. https://doi.org/10.1016/0043-1354(91)90030-T
2021 Mora-Naranjo, N., Meima, J.A., Haarstrick, A., Hempel, D.C., 2004. Modelling and
2022 experimental investigation of environmental influences on the acetate and methane
2023 formation in solid waste. Waste Manag. 24, 763–773.
2024 https://doi.org/http://dx.doi.org/10.1016/j.wasman.2004.04.006
2025 Morel, F., Morgan, J., 1972. A Numerical Method for Computing Equilibria in Aqueous
2026 Chemical Systems 6.
2027 Mudhoo, A., Kumar, S., 2013. Effects of heavy metals as stress factors on anaerobic digestion
2028 processes and biogas production from biomass. Int. J. Environ. Sci. Technol. 10, 1383–
2029 1398. https://doi.org/10.1007/s13762-012-0167-y
2030 Musvoto, E. V., Ekama, G.A., Wentzel, M.C., Loewenthal, R.E., 2000a. Extension and
2031 application of the three-phase weak acid/base kinetic model to the aeration treatment of
2032 anaerobic digester liquors. Water SA 26, 417–438.
2033 Musvoto, E. V., Wentzel, M.C., Loewenthal, R.E., Ekama, G.A., 2000b. Integrated chemical-
2034 physical processes modelling - I. Development of a kinetic-based model for mixed weak
2035 acid/base systems. Water Res. 34, 1857–1867. https://doi.org/10.1016/S0043-
2036 1354(99)00334-6
2037 Nguyen, D., Gadhamshetty, V., Nitayavardhana, S., Khanal, S.K., 2015. Automatic process
2038 control in anaerobic digestion technology: A critical review. Bioresour. Technol. 193, 513–
2039 522. https://doi.org/10.1016/j.biortech.2015.06.080
2040 O’Flaherty, V., Collins, G., Mahony, T., 2006. The microbiology and biochemistry of anaerobic
2041 bioreactors with relevance to domestic sewage treatment. Rev. Environ. Sci. Biotechnol. 5,
2042 39–55. https://doi.org/10.1007/s11157-005-5478-8
2043 Ohlinger, K. N., Young, T.M., Schroeder, E.D., 1998. Predicting struvite formation in digestion.
2044 Water Res. 32, 3607–3614. https://doi.org/10.1016/S0043-1354(98)00123-7
2045 Ohlinger, K.N., Young, T.M., Schroeder, E.D., 1998. Predicting struvite formation in digestion.
2046 Water Res. 32, 3607–3614. https://doi.org/10.1016/S0043-1354(98)00123-7
2047 Oleszkiewicz, J.A., Sharma, V.K., 1990. Stimulation and inhibition of anaerobic processes by
2048 heavy metals-A review. Biol. Wastes 31, 45–67. https://doi.org/10.1016/0269-
2049 7483(90)90043-R
2050 Ortner, M., Rameder, M., Rachbauer, L., Bochmann, G., Fuchs, W., 2015. Bioavailability of
2051 essential trace elements and their impact on anaerobic digestion of slaughterhouse waste.
2052 Biochem. Eng. J. 99, 107–113. https://doi.org/10.1016/j.bej.2015.03.021
67
2053 Osuna, M.B., Zandvoort, M.H., Iza, J.M., Lettinga, G., Lens, P.N.L., 2003. Effects of trace
2054 element addition on volatile fatty acid conversions in anaerobic granular sludge reactors.
2055 Environ. Technol. (United Kingdom) 24, 573–587.
2056 https://doi.org/10.1080/09593330309385592
2057 Oude Elferink, S.J.W.H., Visser, A., Hulshoff Pol, L.W., Stams, A.J.M., 1994. Sulfate reduction
2058 in methanogenic bioreactors. FEMS Microbiol. Rev. 15, 119–136.
2059 https://doi.org/10.1016/0168-6445(94)90108-2
2060 Papirio, S., Frunzo, L., Mattei, M.R., Ferraro, A., 2017. Sustainable Heavy Metal Remediation.
2061 https://doi.org/10.1007/978-3-319-61146-4
2062 Parker, W.J., Wu, G.-H., 2006. Modifying ADM1 to include formation and emission of
2063 odourants. Water Sci. Technol. 54, 111–117. https://doi.org/10.2166/wst.2006.532
2064 Paulo, L.M., Stams, A.J.M., Sousa, D.Z., 2015. Methanogens, sulphate and heavy metals: a
2065 complex system. Rev. Environ. Sci. Biotechnol. 14, 537–553.
2066 https://doi.org/10.1007/s11157-015-9387-1
2067 Peiris, B.R.H., Rathnasiri, P.G., Johansen, J.E., Kuhn, A., Bakke, R., 2006. ADM1 simulations
2068 of hydrogen production. Water Sci. Technol. 53, 129–137.
2069 https://doi.org/10.2166/wst.2006.243
2070 Pind, P.F., Angelidaki, I., Ahring, B.K., Stamatelatou, K., Lyberatos, G., 2003. Monitoring and
2071 control of anaerobic reactors. Adv. Biochem. Eng. Biotechnol. 82, 135–82.
2072 Plaza, C., Brunetti, G., Senesi, N., Polo, A., 2006. Fluorescence characterization of metal ion-
2073 humic acid interactions in soils amended with composted municipal solid wastes. Anal.
2074 Bioanal. Chem. 386, 2133–2140. https://doi.org/10.1007/s00216-006-0844-0
2075 Qiu, H., Vijver, M.G., van Gestel, C.A.M., He, E., Peijnenburg, W.J.G.M., 2014. Modeling
2076 cadmium and nickel toxicity to earthworms with the free ion approach. Environ. Toxicol.
2077 Chem. 33, 438–446. https://doi.org/10.1002/etc.2454
2078 Rittmann, B.E., Banaszak, J.E., VanBriesen, J.M., Reed, D.T., 2002. Mathematical modeling of
2079 precipitation and dissolution reactions in microbiological systems. Biodegradation 13, 239–
2080 250. https://doi.org/10.1023/A:1021225321263
2081 Rotureau, E., Billard, P., Duval, J.F.L., 2015. Evaluation of metal biouptake from the analysis of
2082 bulk metal depletion kinetics at various cell concentrations: Theory and application.
2083 Environ. Sci. Technol. 49, 990–998. https://doi.org/10.1021/es505049f
2084 Roussel, J., 2012. Metal behaviour in anaerobic sludge digesters supplemented with trace
2085 nutrients. PhD Thesis, Univ. Birmingham, United Kingdom.
2086 Roussel, J., Fermoso, F.G., Collins, G., Van Hullebusch, E., Esposito, G., Mucha, A.P., 2018.
2087 Trace element supplementation as a management tool for anaerobic digester operation:
2088 benefits and risks. Water Intell. Online 17, 15. https://doi.org/10.2166/9781780409429
2089 Scherer, P., Sahm, H., 1981. Effect of trace elenments and vitamins on the growth of
68
2090 Methanosarcina barkeri. Acta Biotechnol. 1, 57–65; ISSN: 1618-2863.
2091 Schiavon Maia, D.C., Niklevicz, R.R., Arioli, R., Frare, L.M., Arroyo, P.A., Gimenes, M.L.,
2092 Pereira, N.C., 2017. Removal of H2S and CO2 from biogas in bench scale and the pilot scale
2093 using a regenerable Fe-EDTA solution. Renew. Energy 109, 188–194.
2094 https://doi.org/10.1016/j.renene.2017.03.023
2095 Schmidt, T., Nelles, M., Scholwin, F., Pröter, J., 2014. Trace element supplementation in the
2096 biogas production from wheat stillage - Optimization of metal dosing. Bioresour. Technol.
2097 168, 80–85. https://doi.org/10.1016/j.biortech.2014.02.124
2098 Schwarz, A.O., Rittmann, B.E., 2007. A biogeochemical framework for metal detoxification in
2099 sulfidic systems. Biodegradation 18, 675–692. https://doi.org/10.1007/s10532-007-9101-2
2100 Scott, W.D., Wrigley, T.J., Webb, K.M., Science, E., 1991. A computer model of struvite
2101 solution chemistry. Talanta 38, 889–895.
2102 Shakeri, S., Gustavsson, J., Svensson, B.H., Skyllberg, U., 2012. Talanta Sulfur K-edge XANES
2103 and acid volatile sulfide analyses of changes in chemical speciation of S and Fe during
2104 sequential extraction of trace metals in anoxic sludge from biogas reactors. Talanta 89, 470–
2105 477. https://doi.org/10.1016/j.talanta.2011.12.065
2106 Silva, F., Nadais, H., Prates, A., Arroja, L., Capela, I., 2009. Modelling of anaerobic treatment of
2107 evaporator condensate (EC) from a sulphite pulp mill using the IWA anaerobic digestion
2108 model no. 1 (ADM1). Chem. Eng. J. 148, 319–326.
2109 https://doi.org/10.1016/j.cej.2008.09.002
2110 Smith, R.M., Martell, A.E., Motekaitis, R.J., 2004. NIST critically selected stability constants of
2111 metal complexes database. NIST Stand. Ref. Database 46.
2112 Smith, S., 2008. Phosphate Complexation Model and Its Implications for Chemical Phosphorus
2113 Removal 80, 428–438. https://doi.org/10.2175/106143008X268443
2114 Solon, K., Flores-Alsina, X., Mbamba, C.K., Volcke, E.I.P., Tait, S., Batstone, D., Gernaey, K.
2115 V., Jeppsson, U., 2015. Effects of ionic strength and ion pairing on (plant-wide) modelling
2116 of anaerobic digestion. Water Res. 70, 235–245.
2118 Steyer, J.-P., Buffière, P., Rolland, D., Moletta, R., 1999. Advanced control of anaerobic
2119 digestion processes through disturbances monitoring. Water Res. 33, 2059–2068.
2120 https://doi.org/10.1016/S0043-1354(98)00430-8
2121 Stuckey, D.C., 2012. Recent developments in anaerobic membrane reactors. Bioresour. Technol.
2122 122, 137–148. https://doi.org/10.1016/j.biortech.2012.05.138
2123 Stumm, W., Morgan, J.J., 1996. Aquatic chemistry, 3rd ed. John wiley and sons, Inc, Newyork.
2124 Stupperich, E., Kräutler, B., 1988. Pseudo vitamin B12 or 5-hydroxybenzimidazolyl-cobamide
2125 are the corrinoids found in methanogenic bacteria. Arch. Microbiol. 149, 268–271.
2126 https://doi.org/10.1007/BF00422016
69
2127 Tessier, A., Campbell, P.G.C., Bisson, M., 1979. Sequential Extraction Procedure for the
2128 Speciation of Particulate Trace Metals. Anal. Chem. 51, 844–851.
2129 https://doi.org/10.1021/ac50043a017
2130 Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2017a. Effect of operating conditions on
2131 speciation and bioavailability of trace metals in submerged anaerobic membrane
2132 bioreactors. Bioresour. Technol. 243, 810–819.
2134 Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2017b. Effect of Ethylenediamine-N , N’-
2135 disuccinic acid (EDDS) on the speciation and bioavailability of Fe 2+ in the presence of
2136 sulfide in anaerobic digestion. Bioresour. Technol. 229, 169–179.
2138 Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2015. Trace metal speciation and
2139 bioavailability in anaerobic digestion : A review. Biotechnol. Adv. 34, 122–136.
2140 https://doi.org/10.1016/j.biotechadv.2015.12.006
2141 Thauer, R.K., 1998. Biochemistry of methanogenesis : a tribute to marjory stephenson. Ordinary
2142 Meet. Soc. Gen. Microbiol. 2377–2406.
2143 Tsapekos, P., Alvarado-Morales, M., Tong, J., Angelidaki, I., 2018. Nickel spiking to improve
2144 the methane yield of sewage sludge. Bioresour. Technol. 270, 732–737.
2146 Tugtas, A.E., Tezel, U., Pavlostathis, S.G., 2006. An extension of the Anaerobic Digestion
2147 Model No. 1 to include the effect of nitrate reduction processes. Water Sci. Technol. 54,
2148 41–9.
2149 Uemura, S., 2010. Mineral Requirements for Mesophilic and Thermophilic Anaerobic Digestion
2150 of Organic Solid Waste. Int. J. Environ. Res. 4, 33–40.
2151 van der Veen, A., Fermoso, F.G., Lens, P.N.L., 2007. Bonding from analysis of metals and sulfur
2152 fractionation in methanol-grown anaerobic granular sludge. Eng. Life Sci. 7, 480–489.
2153 https://doi.org/10.1002/elsc.200720208
2154 Van Hullebusch, E.D., Gieteling, J., Zhang, M., Zandvoort, M.H., Daele, W. Van, Defrancq, J.,
2155 Lens, P.N.L., 2006. Cobalt sorption onto anaerobic granular sludge: Isotherm and spatial
2156 localization analysis. J. Biotechnol. 121, 227–240.
2157 https://doi.org/10.1016/j.jbiotec.2005.07.011
2158 van Hullebusch, E.D., Guibaud, G., Simon, S., Lenz, M., Shakeri Yekta, S., Fermoso, F.G., Jain,
2159 R., Duester, L., Roussel, J., Guillon, E., Skyllberg, U., Almeida, C.M.R., Pechaud, Y.,
2160 Garuti, M., Frunzo, L., Esposito, G., Carliell-Marquet, C., Ortner, M., Collins, G., 2016.
2161 Methodological approaches for fractionation and speciation to estimate trace element
2162 bioavailability in engineered anaerobic digestion ecosystems: an overview. Crit. Rev.
2163 Environ. Sci. Technol. 46, 1324–1366.
2164 https://doi.org/http://dx.doi.org/10.1080/10643389.2016.1235943
70
2165 Vintiloiu, A., Boxriker, M., Lemmer, A., Oechsner, H., Jungbluth, T., Mathies, E., Ramhold, D.,
2166 2013. Effect of ethylenediaminetetraacetic acid (EDTA) on the bioavailability of trace
2167 elements during anaerobic digestion. Chem. Eng. J. 223, 436–441.
2168 https://doi.org/10.1016/j.cej.2013.02.104
2169 Waldron, K.J., Robinson, N.J., 2009. How do bacterial cells ensure that metalloproteins get the
2170 correct metal? Nat. Rev. Microbiol. 7, 25–35. https://doi.org/10.1038/nrmicro2074
2171 Wijekoon, K.C., Visvanathan, C., Abeynayaka, A., 2011. Effect of organic loading rate on VFA
2172 production, organic matter removal and microbial activity of a two-stage thermophilic
2173 anaerobic membrane bioreactor. Bioresour. Technol. 102, 5353–5360.
2175 Willet, A.I., Rittmann, B.E., 2003. Slow complexation kinetics for ferric iron and EDTA
2176 complexes make EDTA non-biodegradable. Biodegradation. 14, 105–121.
2177 https://doi.org/10.1023/A:1024074329955
2178 Wintsche, B., Jehmlich, N., Popp, D., Harms, H., Kleinsteuber, S., 2018. Metabolic adaptation of
2179 methanogens in anaerobic digesters upon trace element limitation. Front. Microbiol. 9, 1–
2180 10. https://doi.org/10.3389/fmicb.2018.00405
2181 Wood, D.K., Tchobanoglous, G., 1975. Trace elements in biological waste treatment. J. Water
2182 Pollut. Control Fed. 47, 1933–1944.
2183 Worms, I., Simon, D.F., Hassler, C.S., Wilkinson, K.J., 2006. Bioavailability of trace metals
2184 to aquatic microorganisms: importance of chemical, biological and physical processes
2185 on biouptake. Biochimie 88, 1721–1731. https://doi.org/10.1016/j.biochi.2006.09.008
2186 Wrigley, T.J., Scoit, W.D., Webb, K.M., 1992. An improved computer model of struvite. Talanta
2187 39, 1597–1603.
2188 Xu, F., Li, Y., Wang, Z.W., 2015. Mathematical modeling of solid-state anaerobic digestion.
2189 Prog. Energy Combust. Sci. 51, 49–66. https://doi.org/10.1016/j.pecs.2015.09.001
2190 Xue, L., Li, D., Xi, Y., 2015. Nonlinear Model Predictive Control of anaerobic digestion process
2191 based on reduced ADM1. Control Conf. (ASCC), 2015 10th Asian 1–6. https://doi.org/978-
2192 1-4799-7862-5
2193 Yang, W., Young, S., Munoz, A., Palmarin, M.J., 2019. Dynamic modeling of a full-scale
2194 anaerobic mesophilic digester start-up process for the treatment of primary sludge. J.
2195 Environ. Chem. Eng. 7, 103091. https://doi.org/10.1016/j.jece.2019.103091
2196 Yekta, S.S., 2014. Chemical Speciation of Sulfur and Metals in Biogas Reactors Implications for
2197 Cobalt and Nickel Bio-uptake Processes.
2198 Yekta, S.S., Lindmark, A., Skyllberg, U., Danielsson, Å., Svensson, B.H., 2014a. Importance of
2199 reduced sulfur for the equilibrium chemistry and kinetics of Fe (II), Co (II) and Ni (II)
2200 supplemented to semi-continuous stirred tank biogas reactors fed with stillage. J. Hazard.
2201 Mater. 269, 83–88. https://doi.org/10.1016/j.jhazmat.2014.01.051
71
2202 Yekta, S.S., Skyllberg, U., Danielsson, Å., Björn, A., Svensson, B.H., 2016. Chemical speciation
2203 of sulfur and metals in biogas reactors—Implications for cobalt and nickel bio-uptake
2204 processes. J. Hazard. Mater. 324, 110–116. https://doi.org/10.1016/j.jhazmat.2015.12.058
2205 Yekta, S.S., Svensson, B.H., Björn, A., Skyllberg, U., 2014b. Thermodynamic modeling of iron
2206 and trace metal solubility and speciation under sulfidic and ferruginous conditions in full
2207 scale continuous stirred tank biogas reactors. Appl. Geochemistry 47, 61–73.
2208 https://doi.org/10.1016/j.apgeochem.2014.05.001
2209 Yekta, S.S., Ziels, R.M., Björn, A., Skyllberg, U., Ejlertsson, J., Karlsson, A., Svedlund, M.,
2210 Willén, M., Svensson, B.H., 2017. Importance of sulfide interaction with iron as regulator
2211 of the microbial community in biogas reactors and its effect on methanogenesis, volatile
2212 fatty acids turnover, and syntrophic long-chain fatty acids degradation. J. Biosci. Bioeng.
2213 123, 597–605. https://doi.org/10.1016/j.jbiosc.2016.12.003
2214 Zandvoort, B.M.H., Hullebusch, E.D. Van, Fermoso, F.G., Lens, P.N.L., 2006. Trace Metals in
2215 Anaerobic Granular Sludge Reactors : Bioavailability and Dosing Strategies. Eng. Life Sci.
2216 6, 293–301. https://doi.org/10.1002/elsc.200620129.
2217 Zeikus, J.G., Weimer, P.J., Nelson, D.R., Daniels, L., 1975. Bacterial methanogenesis: Acetate
2218 as a methane precursor in pure culture. Arch. Microbiol. 104, 129–134.
2219 https://doi.org/10.1007/BF00447312
2220 Zhang, L., Jahng, D., 2012. Long-term anaerobic digestion of food waste stabilized by trace
2221 elements. Waste Manag. 32, 1509–1515. https://doi.org/10.1016/j.wasman.2012.03.015
2222 Zhang, L., Lee, Y.W., Jahng, D., 2011. Anaerobic co-digestion of food waste and piggery
2223 wastewater: Focusing on the role of trace elements. Bioresour. Technol. 102, 5048–5059.
2225 Zhang, W., Zhang, L., Li, A., 2015. Enhanced anaerobic digestion of food waste by trace metal
2226 elements supplementation and reduced metals dosage by green chelating agent [S, S]-EDDS
2227 via improving metals bioavailability. Water Res. 84, 266–277.
2229 Zhang, Y., Piccard, S., Zhou, W., 2015. Improved ADM1 model for anaerobic digestion process
2230 considering physico-chemical reactions. Bioresour. Technol. 196, 279–289.
2232 Zhao, C.M., Campbell, P.G.C., Wilkinson, K.J., 2016. When are metal complexes bioavailable?
2233 Environ. Chem. 13, 425–433. https://doi.org/10.1071/EN15205
2234
2235
2236
2237
72
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
73
Chapter 5
Mathematical modelling of trace
element dynamics in anaerobic
digestion environments
Bikash Chandra Maharaj1,2,4*, Maria Rosaria Mattei2,
Luigi Frunzo2, Artin Hatzikioseyian3, Eric D. van
Hullebusch4,5, Piet N. L. Lens4 and Giovanni Esposito6
1
University of Cassino and the Southern Lazio, Department of Civil and
Mechanical Engineering, Cassino, Italy
2
University of Naples Federico II, Department of Mathematics and
Applications ‘Renato Caccioppoli’, Naples, Italy
3
Laboratory of Environmental Science and Engineering, School of Mining
and Metallurgical Engineering, National Technical University of Athens
(NTUA), 9 Heroon Polytechniou, 15780, Athens, Greece
4
IHE Delft Institute for Water Education, Department of Environmental
Engineering and Water Technology, The Netherlands
5
Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Université
Paris Diderot, UMR 7154, CNRS, F-75005 Paris, France
6
University of Naples Federico II, Department of Civil, Architectural and
Environmental Engineering, Naples, Italy
*Corresponding author: Department of Civil and Mechanical Engineering,
University of Cassino and Southern Lazio, Via di Basio 43, 03043 Cassino, Italy
ABSTRACT
Trace elements (TEs) are essential for microbial activity in anaerobic environments.
They are often added to improve the biogas production rate and yield. Dosing of TEs
in anaerobic digestion (AD) systems is largely based on trial-and-error approach as
© 2019 The Authors. This is an Open Access book chapter distributed under the terms of the Creative
Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided
the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). The chapter is from the
book Trace Elements in Anaerobic Biotechnologies, Fernando G. Fermoso, Eric van Hullebusch, Gavin
Collins, Jimmy Roussel, Ana Paula Mucha and Giovanni Esposito (Eds.).
doi: 9781789060225_0101
Downloaded from https://iwaponline.com/ebooks/book-pdf/572061/wio9781789060225.pdf

by guest
102 Trace Elements in Anaerobic Biotechnologies
no general guidelines exist to date. This is primarily because the fate of TEs in AD
environments still remains poorly understood. This knowledge gap is due to the
multiple and complex biogeochemical processes influencing TEs chemistry, TE
physicochemical interactions with biotic and abiotic surfaces, as well as uptake
by microbial community. A mathematical model based on TE-dosing
optimization can be recruited to tackle such a complex problem. In this regard,
the major physicochemical processes involved in determining the fate of TEs in
an anaerobic-digestion environment need to be reviewed and consolidated with a
suitable modelling approach. This chapter enlists and describes the most
important physicochemical processes such as precipitation, adsorption, and
aqueous complexation, as well as the bio-uptake mechanisms involving TEs in
AD systems with the aim of summarizing the main modelling contributions to
determine the fate of TEs in engineered anaerobic-digestion environments.
KEYWORDS: modeling, anaerobic digestion, trace elements
5.1 INTRODUCTION
Nutrients are essential for living organisms to carry out catabolic and
anabolic biochemical reactions. The microbial requirement in terms of nutrients
has been manipulated in all bioprocess technologies (England, 2013). Nutrients
can be classified in two broad categories: macro- and micronutrients.
Macronutrients are the bulk of the energy molecules required by the microbes,
while micronutrients support the metabolism in the form of structural molecules.
Micronutrients also enhance the rate and specificity of biochemical reactions.
Accordingly, balanced macro- and micronutrients are required for ideal growth
conditions and are essential for efficient and stable biogas production (Fermoso
et al., 2008). Any imbalance in these factors can affect the activity and syntrophy
of microorganisms in an anaerobic digester (Gustavsson, 2012; Jiang, 2006) and
hence limit biogas production. The presence of oxygen, the accumulation of
volatile fatty acids (VFAs) and the presence of toxic compounds/elements (i.e.
ammonium, high concentrations of TEs) are the major disturbances reported for
AD systems (Chen et al., 2008). Among all the microbial communities prevailing
in an anaerobic environment, methanogens are the most sensitive to
environmental perturbations and hence their activity is easily disturbed.
Table 5.1 lists the most important micronutrients or trace elements (TEs)
involved in AD systems and the stimulating concentration range reported in
some experimental studies performed with pure microbial strains (Glass &
Orphan, 2012). In this regard, Fe, Co an Ni have been recognized as essential
micronutrients (Glass & Orphan, 2012; Oleszkiewicz & Sharma, 1990; Thanh
et al., 2015; Uemura, 2010). The effect of TEs on AD has been extensively
reported (Fermoso et al., 2008; Gonzalez-Gil & Kleerebezem, 1999; Jiang, 2006;
Mudhoo & Kumar, 2013; Roussel, 2012; Thanh et al., 2015; Zandvoort et al.,

by guest
Mathematical modelling of trace-element dynamics 103
2006). Cobalt forms the core of the vitamin B12 which acts as a cofactor for the
enzyme methylase involved in methane production (Banerjee & Ragsdale, 2003;
Stupperich & Kräutler, 1988; Mazumder et al., 1987). The coenzyme F430
containing Ni is essential for the functioning of the methylcoenzyme M
reductase, which is involved in the reduction of coenzyme M to methane in
methanogens (Finazzo et al., 2003; Thauer, 1998). Iron is involved in the
transport system of the methanogenic archaea for the conversion of CO2 to CH4,
where it functions both as an electron acceptor and donor (Thanh et al., 2015),
and also binds to sulfide to form precipitates which in turn reduce the hydrogen
sulfide content of the anaerobic digester (Gustavsson, 2012; Hille et al., 2004;
Kong et al., 2016; Oude Elferink et al., 1994; Shakeri et al., 2012). Similarly,
Mn acts as an electron acceptor in anaerobic processes. Zn, Cu, and Ni have all
been found in hydrogenase (Thanh et al., 2015). Trace elements such as W and
Mo are also found in enzymes such as formate dehydrogenase involved in
formate formation from propionate by propionate oxidizers (Thanh et al., 2015).
Tungsten can also be considered as essential, but its low concentration in the
sludge in comparison with the other metals (Fe, Ni and Co) reduces the interest
for supplementing this metal (Fermoso et al., 2009; Oleszkiewicz & Sharma, 1990).
Table 5.1 List of TEs with optimal dissolved concentration in growth media for pure
methanogenic cultures grown on different substrates (adapted from Glass & Orphan,
2012).
Metal Substrate Concentration (µM) Species

Fe H2/CO2 300–500 Methanospirillum hungatei
H2/CO2 .15 Methanococcus voltae
Acetate 100 Methanothrix soeggenii
Methanol 50 Methanosarcina bakeri
Ni H2/CO2 1 Methanobacterium thermoautotrophicum
H2/CO2 0.2 Methanococcus voltae
Acetate 2 Methanothrix soehngenii
Co Acetate 2 Methanothrix soehngenii
Methanol 1 Methanosarcina barkeri
Mo Acetate 2 Methanothrix soehngenii
W H2/CO2 1 Methanocorpusculum parvum
Formate 0.5 Methanocorpusculum parvum
Within biological systems, the physicochemical processes are those that are not
directly mediated by microbial activity, but affect the biochemical processes to a

by guest
large extent (Batstone et al., 2012). Physicochemical processes including

precipitation, co-precipitation, adsorption, surface complexation and gas
production are among the most important taking place in AD. These processes
occur in the extracellular environment and affect the overall performance of the
system by mediating a change in the proton balance (hence in pH) of the
bioreactor (Fermoso et al., 2015). In particular, the physicochemical processes are
directly related to the fate of TEs in AD and, therefore, to the bioavailability of
TEs for bio-uptake. Some of the main physicochemical mechanisms related to
TEs in AD are presented in Figure 5.1.
Figure 5.1 Schematic representation of various processes affecting the dynamics of

TEs in an engineered anaerobic system.
The underlying principles of the physicochemical processes are relatively well

understood (Batstone et al., 2012; Stumm & Morgan, 1996). However, existing
AD models, including the Anaerobic Digestion Model 1 (ADM1) as the most
representative (Batstone et al., 2002), lack a thoughtful implementation of many
of those processes occurring in AD (Batstone et al., 2012; Fermoso et al., 2015;
Flores-Alsina et al., 2016; Maharaj et al., 2018; Mbamba et al., 2015a, b; Xu
et al., 2015; Zhang et al., 2015).
The understanding and efforts to model and quantify the fate of TEs in AD are
isolated and patchy. This chapter attempts to consolidate the available knowledge
and efforts on modelling the fate of TEs in AD and envisages a set of processes
necessary to quantify the TE dynamics. The chapter presents the principles
and methodologies to incorporate the effect of TEs in the structure of AD
models. Figure 5.2 summarizes a conceptual interaction diagram between

by guest
methane-producing microbes, sulfate-reducing bacteria and TEs. It is clearly

depicted that the effect of TEs should be studied in an integrated competing
biotic and abiotic environment in an AD model. For all the processes to be
discussed in the following sections, the relative models have been classified in
two broad categories: the first corresponds to the mathematical models that have
been formulated for the specific process independently from the consideration
of an AD environment, while the second contemplates the models that can be
considered to be an extension of the original ADM1 model. This option is
justified by the significant impact that the ADM1 model has on AD literature.
Indeed, in the timeline of the published models, ADM1 is considered a
significant milestone (Batstone et al., 2002). ADM1 was the result of a
collaborative work from the International Water Association (IWA) task group
for mathematical modelling of AD processes. The authors tried to unify and
consolidate the knowledge and modelling experience accumulated in the
literature until 2002, when the scientific and technical report of ADM1 was first
published (Batstone et al., 2002). ADM1 is considered even today a state-of-
the-art model, including a significant number of biological, chemical and
physicochemical processes (Batstone et al., 2002). The model was conceived as a
general modelling framework for AD processes and as such has been used
successfully without modifications (Blumensaat & Keller, 2005) or in a modified
version (Galí et al., 2009; Parker & Wu, 2006; Peiris et al., 2006) by many
researchers to simulate experimental results. ADM1 has also been used as a
complementary module to activated sludge models such as activated sludge
model 1 (ASM1), activated sludge model 2d (ASM2d) or activated sludge model
3 (ASM3) for plant-wide wastewater treatment processes (Copp et al., 2003;
Kauder et al., 2007; Lopez-Vazquez et al., 2013; Nopens et al., 2009; Rosen
et al., 2006).
Note that the majority of mathematical models under consideration in this work
do not explicitly consider TEs as model components. However, such models are
discussed as they introduce a general framework that could be extended/adapted
to the case of TEs in AD systems. Furthermore, as the mathematical modelling
of the fate of TEs in AD systems cannot neglect the explicit consideration of
sulfur and phosphorus dynamics, which play a crucial role in precipitation and
adsorption processes, the main mathematical models accounting for sulfur and
phosphorus in anaerobic environments have been reviewed as well.
This chapter is organized as follows: in Section 5.2 the main models for sulfur
and phosphorus dynamics in anaerobic environments are described. In Section 5.3
the main physicochemical processes affecting TE availability in AD systems are
described and the modelling approaches proposed over the years are reviewed and
summarized. Specifically, Section 5.3.1 is dedicated to precipitation modelling; in
Section 5.3.2 adsorption mechanisms and the main modelling approaches to this
topic are presented; Section 5.3.3 is devoted to aqueous complexation and the
relative modelling approaches. The chapter has been extended with Section 5.4

by guest
which refers to the mathematical modelling of TEs bio-uptake and reviews the
mathematical formulas usually adopted for dose-response functions.
Figure 5.2 Conceptual interactions between methanogenic activity, sulfate reduction

and trace elements positive (+) or negative (−) effects on biogas production.
5.2 MATHEMATICAL MODELLING OF SULFUR AND

PHOSPHORUS CYCLES IN ANAEROBIC DIGESTION
SYSTEMS
5.2.1 Sulfur modelling
Sulfur represents a dominant element in anaerobic digesters, counting on multiple
sources. In particular, sulfide precipitates have been identified as one of the major
sinks for sulfur-related compounds and subsequent removal of heavy metals
toxicity (Lawrence & McCarty, 1965). Due to the significant participation of
sulfur compounds in physicochemical processes affecting TEs fate and
bioavailability in AD systems, it is necessary to consider sulfur transformations
such as sulfate reduction in AD models (Paulo et al., 2015). In this regard, ADM1
lacks the representation of sulfur bioprocesses. The extension of any AD models
with sulfate-reduction processes is considered prerequisite when TEs are also
included in the system. This is related to the fact that most TEs form stable
insoluble solid sulfide precipitates under sulfidogenic conditions. This process

by guest
significantly affects the bioavailable fraction of TEs and cannot be ignored (Paulo
et al., 2015). Nonetheless, some studies have been carried out for modelling
sulfate reduction in AD. Sulfate-reduction modelling necessitates the definition of
important biochemical, gas transfer, and physicochemical process rates to link
new state variables. These processes and state variables are mentioned along with
the modelling efforts discussed in the following sections.
Sulfate-reduction models have been developed and calibrated for specific
purposes for different substrates and for simulating specific systems. Some follow
a different principle (other than ADM1) in their conception and hence they are
classified in this chapter as ‘stand-alone sulfate-reduction models’ (Kalyuzhnyi &
Fedorovich, 1997, 1998; Knobel & Lewis, 2002; Poinapen & Ekama, 2010;
Ristow et al., 2002; Vavilin et al., 1995). Conversely, models which are embedded
in the structure of ADM1 have been termed as ‘ADM1-based sulfate-reduction
models’ (Barrera et al., 2015; Federovich et al., 2003; Flores-Alsina et al., 2016).
These two groups of models are discussed in the following sections.
5.2.1.1 Stand-alone sulfate-reduction models

Sulfate-reduction models have evolved independently as well as part of more
complex AD models. These models do not follow the ADM1 structure as the
theoretical basis, for some of them were already set before the inception of
ADM1. In an attempt to model the physicochemical reaction system for pH
prediction, a simulation model for AD was developed (Vavilin et al., 1995). The
model kept track of the sulfide formation and its effect on pH. It also considered
two sulfate-reducing microbial groups which separately use acetate and
propionate as organic substrate to produce sulfide from sulfate. A reduced model
of self-oscillating dynamics in an AD system with sulfate reduction was
developed in (Kalyuzhnyi & Fedorovich, 1998). The model considered only two
groups of microbes, methanogens and sulfate reducers. Substrate-limiting
functions were described by typical Monod equations. The model considered
hydrogen sulfide inhibition as a function of non-ionized hydrogen sulfide.
Finally, the model was validated with experimental data. The competition
between sulfate-reducing bacteria (SRB) and methanogens by using data from a
synthetic high sulfate-containing wastewater treated in a UASB reactor was
evaluated in (Kalyuzhnyi & Fedorovich, 1998). There was a good correlation
between model simulations and experimental data. Competition for acetate was
emphasized. The modelling results were obtained under variations of hydraulic
retention time (HRT), SO2−4 /COD ratio, initial proportion of SRB/methanogens
in seed sludge, efficiency of retention of SRB, and sludge retention time.
Similarly, the competition between methanogens and SRB in a UASB reactor
was studied to develop a dispersed plug-flow model (Kalyuzhnyi et al., 1998).
Data from literature were used to calibrate the model with emphasis on acetate
competition. The model was able to predict system performance with regard to

by guest
variations in the liquid upward velocity as an important control parameter. However,

when pH changes occurred the results were not correctly predicted. The model of
Knobel and Lewis (2002) was calibrated and validated for a number of reactor
configurations: packed bed, UASB and gas lift reactor. Three simulation studies
were carried out in steady-state and dynamic conditions. The first two simulation
tests were able to predict concentration of sulfate and COD in the effluent.
AQUASIM was used (Ristow et al., 2002) in a recycling sludge bed reactor to
simulate the AD process including sulfate reduction. Acid-mine drainage and
primary sludge were used as sulfate-containing wastewater and carbon source,
respectively. Data from pilot plant agreed well with model simulations. Influence
of sludge recycle ratio, SO2−4 , COD ratio and HRT on sulfate-reduction process
was studied. A two-phase (aqueous-gas) kinetic model for sulfate-reduction
processes using primary sewage sludge as carbon source was developed in
Poinapen & Ekama 2010. The kinetic model was calibrated with experimental
data starting from different SO2−4 /COD ratios (Poinapen & Ekama 2010). Model
simulations showed better fitting with experimental data. In addition, model
simulations showed that at ambient temperature of 20°C, the hydrolysis rate is
significantly reduced as compared with 35°C. The hydrolysis rate of the primary
sewage sludge biodegradable particulate organism is the same under
methanogenic and sulfidogenic conditions. The primary sewage sludge and
biodegradable particulate organics are carbon deficient for biological sulfur
reduction. Moreover, the model only considered gas stripping for H2S. The
efficiency of retention of SRB, HRT, SO2− 4 , butyrate, propionate, and acetate
were predicted. However, the removal efficiencies for COD, SO2− 4 , hydrogen
sulfide concentrations in the gas phase were not predicted (Barrera et al., 2013).
In summary, ‘stand-alone sulfur-reduction models’ have been developed and
calibrated to study the impact of operating parameters and initial concentrations
under specific cases.
5.2.1.2 ADM1-based sulfate-reduction models

Sulfate-reduction processes as an extension of ADM1 have been implemented
in Federovich et al. (2003). The model was calibrated against experimental
data from literature and was able to predict sulfate removal in the AD process,
the concentrations of butyrate, propionate and acetate, as well as methane
and biomass production. The extended model included sulfate reduction and
biogenic sulfide production by SRB. SRB compete with methanogens for
intermediates produced during the biodegradation processes. The extension added
four different SRB species (XSRB butyrate , XSRB propionate , XSRB acetate , XSRB hydrogen )
which oxidize butyrate, propionate, acetate and hydrogen, respectively, coupled
with sulfate reduction according to the following reactions:
XSRB butyrate
2C3 H7 COO− + SO2−
4 −−−−−− 4CH3 COO− + H+ + HS− (5.1)

by guest
XSRB propionate
4C2 H5 COO− + 3SO2−
4 −−−−−− 4CH3 COO− + 4HCO− +
3 + H + 3HS
−
(5.2)
XSRB acetate
CH3 COO− + SO2−
4 −−−−−− 2HCO−
3 + HS
−
(5.3)
XSRB hydrogen
+ −
4H2 + H + SO2−
4 −−−−−− HS + 4H2 O (5.4)
The following rate equations for reactions (5.1)–(5.4) have been adopted:
Si SSO4
ri = kmax,i · · Xi · IpH · Isulfide (5.5)
KSi + Si KSO4 + SSO4
where ri is the volumetric growth rate of SRB group Xi [M L−3 T−1], kmax,i is the
maximum specific growth rate for group Xi [T−1], Si represents the concentration
of substrate i [M L−3], KSi denotes the Monod half-saturation constant for
substrate i [M L−3], SSO4 is the concentration of sulfate in liquid phase [M L−3],
KSO4 represents the Monod half-saturation constant for sulfate [M L−3], Xi is
the concentration of SRB biomass group i [M L−3], IpH is the pH inhibition
factor (dimensionless), and Isulfide represents the sulfide inhibition factor
(dimensionless). The inhibition by hydrogen sulfide was included in the rate
equation (5.5) through the term:
SH2 S
Isulfide = 1 − (if SH2 S . KI , Isulfide = 0) (5.6)
KI
where SH2S denotes the concentration of undissociated hydrogen sulfide in liquid
phase [M L−3]; and KI is the inhibition constant of undissociated hydrogen
sulfide [M L−3].
The values of the kinetic parameters for SRB were estimated in the model to
simulate the behavior of a UASB fed with sulfate up to 6 g S/L (Federovich
et al., 2003). Although this model is reported today as the most appropriate
extension of ADM1 when sulfate-removal efficiencies are of primary interest, it
was not calibrated to predict the concentrations of total aqueous sulfide,
undissociated sulfides and gas phase sulfides.
Similarly, an extension of ADM1 with sulfate reduction was proposed in
(Barrera et al., 2015). The model was calibrated and validated for a high strength
and sulfate rich wastewater (cane molasses vinasse). Butyric acid was neglected
as substrate for SRB in the model structure. Propionic acid, acetic acid, hydrogen
were considered as primary electron donors for sulfate reduction processes.
Propionate SRB, acetate SRB and hydrogenotrophic SRB were considered in
addition to the seven microbial groups of ADM1. The new biochemical reactions
introduced in the model have been depicted in the following equations:
XSRB propionate
C2 H5 COOH + 0.75H2 SO4 −−−−−− CH3 COOH + CO2 + H2 O + 0.75H2 S
(5.7)

by guest
XSRB acetate
CH3 COOH + H2 SO4 −−−−−− 2CO2 + 2H2 O + H2 S (5.8)
XSRB hydrogen
4H2 + H2 SO4 −−−−−
− H2 S + 4H2 O (5.9)
A dual Monod type kinetic was used for the uptake of the substrates. Protonation
and deprotonation of sulfuric acid and sulfide ions were considered in the liquid
phase. Gas stripping of H2 S was also considered. SH2 S was included as the
process inhibitor. Likewise, a non-competitive inhibition function for sulfides
was introduced. The model prediction works reasonably well for the process
variables with an accurate quantitative predictions of high (+10%) to medium
(10%–30%) and error ranging from 1 to 26%.
Recently in (Flores-Alsina et al., 2016), biological production of sulfide by
sulfate reducing bacteria has been included as one of the four extensions to
ADM1, the others being phosphorus metabolism, mineral precipitation and iron
transformation in an AD system. The model was applied to a set of full-scale
plant-wide data.
Selected process rates and inhibition functions, which may be used to model
sulfate reduction processes, are represented in the equations below. Readers may
refer to the literature for more details (Barrera et al., 2013).
Growth rate for SRB species:
Si SSO4
ri = kmax,i · · Xi · IpH · IH2 S (5.10)
KSi + Si KSO4 + SSO4
where, ri is the volumetric growth rate of SRB group i[ML−3 T−1 ]; k max,i is the
maximum specific growth rate of SRB group i[T−1 ]; Si is the concentration of
soluble organic components [ML−3 ]; KSi is half saturation constant [ML−3 ]; Xi is
the concentration of SRB group i[ML−3 ]; IpH is the pH inhibition function; and
IH2 S is the sulfide inhibition function.
Biomass decay rate:
rdecay,i = kdec,i · Xi (5.11)
where, rdecay,i is the kinetic rate of substrate uptake (kg COD m3 d−1 ); and k dec,i is
first order decay rate of species i(d −1 );
Acid-base rate for sulfide species:
rA/B = KA/B,H2 S (SHS− · (SH+ + Ka,H2 S ) − Ka,H2 S · SH2 S,tot ) (5.12)
where rA/B is the acid-base kinetic rate [ML−3 T−1 ]; kA/B,H2 S is the acid-base kinetic
parameter [M−1 L3 T−1 ]; SHS− is the concentration of bisulfide ion [ML−3 ]; SH+ is the
concentration of hydrogen ion [ML−3 ]; SH2 S,tot is the concentration of total H2S
species (dissociated and undissociated) in liquid phase [ML−3 ]; and Ka,H2 S is the
acid-base equilibrium coefficient [ML−3 ].

by guest
Gas transfer:
rgas = kl a · (SH2 S − KH,H2 S · Pgas,H2 S ) (5.13)
where rgas is the specific mass transfer rate of gas H2 S[ML−3 T−1 ]; kl a is the gas-
liquid mass transfer coefficient [T−1 ]; SH2 S is the concentration of undissociated
H2 S[ML−3 ]KH,H2 S is Henry’s law coefficient [M2 L−3 T−1 ]; and Pgas,H2 S is the
partial pressure of the H2 S gas[MT−1 ].
Process inhibition:
Sulfide:
2
SH S
− 2
IH 2 S = e (5.14)
KI,H S
2
where KI,H2 S is the inhibition coefficient by undissociated H2 S[ML−3 ].

1
IH 2 S = (5.15)
SH2 S
1+
KI,H2 S
1
IH 2 S = (5.16)
ln99/ KK100
SH2 S 2
1+
K2
where K100 is the concentration of H2S or pH at which the uptake rate is decreased
100 times [ML−3]; and K2 is the concentration of H2S or pH at which the uptake rate
is decreased 2 times [ML−3].
pH:
1
IpH = (5.17)
pH ln99/(K100 /K2 )
1+
K2
where pH is the resultant or predicted pH of the system.
IpH = (1 + exp(−aLL (pH − pHLL )))−1 · (1 + exp(−aUL (pH − pHUL )))−1

(5.18)
where αLL and αUL are the parameters which affect the steepness of the curve; pHLL
is the lower pH limit where microbial growth is 50% inhibited; and pHUL is the
upper pH limit where microbial growth is 50% inhibited.
1 + 2 · 100.5( pHUL −pHLL )

IpH = (5.19)
1 + 10( pH−pHUL ) + 10( pHLL −pH)

by guest
5.2.2 Phosphorus modelling

Phosphorus is one of the most abundant inorganic components of AD
systems (∼30% nutrient load according to Johnson and Shang (2006)).
Phosphorus transformations are major drivers for precipitation processes (struvite
precipitation). Although phosphorus precipitation has been addressed in
equilibrium- and kinetic-based models, phosphorus transformations have not
been linked to the biodegradation reactions in the majority of studies. Poly-
hydroxy-alkanoates (XPHA ), poly-phosphate (XPP ), and phosphate (SPO3− 4
) are the
major biochemical components which are necessary to define phosphorus
transformations. Apart from this, inorganic phosphate-water acid/base system
(SPO3−
4
/SHPO2− 4
/SH2 PO−4 /SH3 PO4) should be taken into account to track the change
in pH. Additionally, based on requirements, inorganic phosphate minerals can also
be introduced, such as, amorphous calcium phosphate (XCa3 (PO4 )2), hydroxylapatite
(XCa5 (PO4 )3 (OH) ), octacalium phosphate (XCa8 H2 (PO4 )6) and, struvite (XMgNH4 PO4). The
following lines briefly discuss efforts to model phosphorus transformations in
waste-treatment models and ADM1-based models.
Modelling the effect of pH change (due to phosphate species) and tracking
the dynamics of phosphate minerals in relation to the biodegradation are two
important aspects of modelling phosphorus in AD systems. The pH prediction
module has been expanded to take into account physicochemical effects such
as ion-activity correction, ion-pairing behavior, and weak acid-base reactions
occurring in conjugation with biological reactions (Solon et al., 2015).
Phosphorus metabolism has been implemented in the framework of ADM1 in
(Johnson & Shang, 2006). Two major methods for incorporating phosphorus in
the AD models have been identified (Solon, 2015). In the first approach, P does
not participate in biological reactions. The quantity of phosphorus within the
system, i.e. in the particulate and soluble components, is tracked through the
introduction of a new state variable SIP (concentration of inorganic phosphorus).
This acts as the only source and sink for all the inorganic phosphorous minerals and
acid-base species (Zaher & Chen, 2006; Zaher et al., 2007). In addition, it is
assumed that there are instantaneous phosphorus-related processes taking place at
the interface between the activated sludge and the AD process. The prevalence of
inorganic phosphorus and absence of biological phosphorus oversimplifies the
model structure.
Conversely, the alternate method takes into account some of the processes
which are considered in the IWA Activated Sludge Model No. 2d (ASM2d)
(Henze et al., 1999). These processes are further incorporated in ADM1.
Phosphorus metabolizing microorganisms are active in the anaerobic digester
and facilitate microbial conversion of phosphate-related compounds. Based on
this principle, a model was developed in (Wang et al., 2016). The model
investigated anaerobic fermentation of enhanced biological phosphorus removal of
sludge in terms of phosphate release and VFA production with respect to

by guest
polyphosphate-accumulating organism (PAO) activity. The proposed model

extension included: (a) the storage of 4 VFA as PHA mediated by PAOs; (b) the
effect of PHA content on disintegration rate; and (c) the mineral phosphate
precipitation. Similar to this study, a new extension was implemented which
enabled the release of phosphorus in AD (Flores-Alsina et al., 2016). For this,
phosphorus-accumulating organism (XPAO ), polyhydroxyalkanoates (XPHA ) and
polyphosphates (XPP ) constitute the new state variables. Seven new processes were
added for: (a) uptake of valerate, butyrate, propionate and acetate to form
polyhydroxyalkanoates (XPHA ); (b) decay of phosphorus accumulating organism
(XPAO ); and (c) lysis of polyhydroxyalkanoates (XPHA ) and polyphosphates
(XPP ). Further details of the model can be found in the literature (Flores-Alsina
et al., 2016).
5.3 MATHEMATICAL MODELLING OF

PHYSICOCHEMICAL PROCESSES IN ANAEROBIC
DIGESTION SYSTEMS
Despite several efforts devoted to the improvement of physicochemical modelling
and simulation in anaerobic environments, the effect and fate of TEs have not
been accounted for (Fermoso et al., 2015; Lauwers et al., 2013; van Hullebusch
et al., 2016). As mentioned, the most important processes to be included in the
AD model in order to estimate the fate of TEs are: precipitation, aqueous
complexation and surface complexation/adsorption (Batstone et al., 2012;
van Hullebusch et al., 2016). A modelling approach is envisioned that takes into
account all these intermittent processes and unifies them in the form of a defined
model structure (Figure 5.2 above). In the last decades, there have been efforts in
wastewater-treatment process modelling to predict the physico-chemical state of
the system. A list of such efforts has been provided in Table 5.2. Some of these
works are discussed in the following lines. In addition, the ADM1 extensions
considering the mentioned processes are also described.
5.3.1 Precipitation
With the term ‘non ADM1-based precipitation models’, we refer to the class of
models that have been developed to simulate the precipitation process in aqueous
environments, non-necessarily anaerobic. In some cases such models account
for the biodegradation reactions occurring in AD but they do not follow the
modelling framework introduced in ADM1. Conversely, we define ‘ADM1-based
precipitation models’, the ones that have been conceived in the framework of
ADM1 and explicitly take into account precipitation reactions not necessarily for
TEs. Table 5.3 represents a potential list of precipitation reactions which can
occur in an anaerobic environment.

by guest
by guest
114
Table 5.2 Waste-treatment models considering various processes involving geochemical components.
No. Reference Process Model Type Geochemical Application

Components
3−
1 Hanhoun et al. Precipitation Equilibrium Mg2+, NH+
4 , PO4 , H2O Phosphorus recovery Lab scale
(2011)
3−
2 Scott et al. (1991) Precipitation Equilibrium Mg2+, NH+
4 , PO4 , H2O Waste water stream, Lab scale
kidney stones
3−
3 Barat et al. (2011) Precipitation Equilibrium, Mg2+, NH+ 4 , PO 4 , H 2 O, SBR for biological Lab scale
2+ +
Kinetic CO2−
3 , Ca , K phosphorus removal
4 Musvoto et al. Acid-base reactions, Kinetic H+, OH−, H2CO3, HCO− 3 , Aeration of anaerobic Plant wide
−
(2000b) precipitation, gas CO2−
3 , H3PO4, H2 PO4 , digestion liquor from
3− + 2+
stripping, ion pairing HPO2−4 , PO4 , NH4 , Ca , spent wine UASB
Mg2+ and AD for sewage

sludge
2−
5 Maurer & Boller Precipitation Kinetic H2CO3, HCO− 3 , CO 3 , Waste Lab scale
(1999) OH−, H+, H3PO4, treatment-EBPR
2−
H2 PO−4 , HPO4 ,
2+
PO3−
4 , Ca
2+ 2− −
6 Ohlinger et al. Precipitation Equilibrium Mg , PO3− 4 , HPO4 , OH , Waste treatment Plant wide
+ +
(1998) H , NH4
Trace Elements in Anaerobic Biotechnologies
2+
7 Smith et al. (2008) Precipitation, Equilibrium H+, PO3− 4 , Fe Phosphate removal Lab scale
adsorption,
co-precipitation
3−
8 Wrigley et al. Precipitation Equilibrium Mg2+, NH+
4 , PO4 , H2O Waste water stream, Lab scale
(1992) kidney stones
9 Precipitation Kinetic Waste treatment Lab scale
by guest
Van Rensburg H+, OH−, H2CO3,
2−
et al. (2003) HCO− 3 , CO3 , H3PO4,
− 3−
H2PO4 , HPO2− 4 , PO4 ,
2+ 2+
NH+
4 , Ca , Mg
10 Tait et al. (2009) Precipitation Equilibrium Low cost sulfate Lab scale
removal
11 Hauduc et al. Acid-base; chemical ion Kinetic Multiple components Chemical Lab scale
(2015) pairing; precipitation; phosphorus removal
chemical surface
complexation;
adsorption;
co-precipitation
12 Schwarz & Acid-base reactions, Equilibrium, HS−, S−, H+, Zn2+ Metal detoxification in Lab Scale
Rittmann (2007) Precipitation, surface Kinetic, sulfidic systems
complexation,
biodegradation

2+ 2+
13 Mbamba et al. Acid base, Equilibrium, H+, NH+
4 , Ca , Mg , Calcite precipitation Lab scale
3− 3−
(2015b) precipitation-dissolution, Kinetic PO4 , PO4 in synthetic aqueous
gas stripping solution
14 Flynn et al. (2014) Surface complexation Equilibrium Co, Ni, Sr, Zn Calculation of Lab scale
metal-carboxyl
bacterial surface
stability constant
− 3+
Mathematical modelling of trace-element dynamics
15 VanBriesen & Acid-base, Equilibrium, H+, NH+

4 , OH , Fe , Citrate/Fe(III) Theoretical
Rittmann (1999) complexation and Kinetic H2CO3 systems
biodegradation
115
by guest
Table 5.3 Probable precipitation reactions in AD systems. 116
No. Name Precipitation Reaction LogK sp Citation in AD systems
+ 3+
1 Al2O3 Al2O3 + 6H ⇌ 2Al + 3H2O 19.6 Mu et al. (2011)
2 AlPO4 AlPO4 O Al3+ + PO3+
4 20.2 Peng & Colosi (2016)
3 Anhydrite CaSO4 O Ca2+ + SO2−
4 4.6 MacConnell & Collins (2009)
4 Aragonite CaCO3 O Ca2+ + CO2−
3 8.3 Musvoto et al. (2000b)
5 Brucite Mg(OH)2 + 2H+ ⇌ Mg2+ + 2H2O 17.1 Uludag-Demirer & Othman (2009)
6 CaHPO4 CaHPO4 O Ca2+ + H+ + PO3− 4 19.2 Van Langerak et al. (1999)
7 Ca4H(PO4)3 · 3H2O Ca4 H(PO4 )3 · 3H2 O O 4Ca2+ + H+ + 3PO3−
4 + 3H2 O 47.9 Zhang et al. (2010)
8 Calcite CaCO3 O Ca2+ + CO2−
3 8.4 Mbamba et al. (2015a,b)
9 Ca3(PO4)2 Ca3 (PO4 )2 O 3Ca2+ + 2PO3−
4 25.5 Ye et al. (2010)
10 Dolomite CaMg(CO3 )2 O Ca2+ + Mg2+ + 2CO2−
3 16.5 Musvoto et al. (2000b)
11 Fe(OH)2 Fe(OH)2 + 2H+ ⇌ Fe2+ + 2H2O 12.8 Dezham et al. (1988)
12 FeS FeS + H+ ⇌ Fe2+ + HS− 2.9 Fermoso et al. (2015)

13 Gibbsite Al(OH)3 + 3H+ ⇌ Al3+ + 3H2O 7.7 Qiao & Ho (1996)
14 Huntite CaMg3 (CO3 )4 O 3Mg2+ + Ca2+ + 4CO2−
3 29.9 Musvoto et al. (2000a)
15 Hydroxyapatite Ca10 (PO4 )6 (OH)5 + 5H+ O 10Ca2+ + 6PO3−
4 + 5H2 O 44.3 Maurer et al. (1999)
16 Mackinawite FeS + H+ ⇌ Fe2+ + HS− 3.6 Morse & Arakaki (1993)
17 Magnesite MgCO3 O Mg2+ + CO2−
3 7.4 Van Rensburg et al. (2003)
18 Mg(OH)2 Mg(OH)2 + 2H+ ⇌ Mg2+ + 2H2O 18.7 Huang et al. (2016)
19 Mg3(PO4)2 Mg3 (PO4 )2 O 3Mg2+ + 2PO3−
4 23.2 Musvoto et al. (2000a)
20 Newberyite MgHPO4 · 3H2 O O Mg2+ + H+ + PO3−

4 + 3H2 O 18.1 Musvoto et al. (2000a)
21 Portlandite Ca(OH)2 + 2H+ ⇌ Ca2+ 2H2O 22.7 Gu et al. (2015)
22 Siderite FeCO3 O Fe2+ + CO2−
3 10.5 Preis & Gamsjäger (2001)
3−
23 Struvite MgNH4 PO4 · 6H2 O O Mg2+ + NH+
4 + PO4 13.2 Musvoto et al. (2000a)
24 Vaterite CaCO3 O Ca2+ + CO2− 3 7.9 Musvoto et al. (2000a)
25 Vivianite Fe3 (PO4 )2 · 8H2 O O 3Fe2+ + 2PO3−
4 + 8H2 O 37.7 Roussel & Carliell-Marquet (2016)
5.3.1.1 Non-ADM1-based precipitation models

Precipitation reactions in wastewater systems have been extensively studied in the
last few decades (Barat et al., 2011; Donoso-Bravo et al., 2013; Hanhoun et al.,
2011; Mbamba et al., 2015a, b; Maurer et al., 1999; Musvoto et al., 2000a, b;
Rittmann et al., 2002; Tait et al., 2009; Van Rensburg et al., 2003). Generally,
the precipitation studies focus on nutrient removal or recovery from wastewater
streams. Recovery of phosphate is central to these studies. This is largely due to
the associated risk of water eutrophication (Hauduc et al., 2015; Jimenez et al.,
2015; Rahaman et al., 2014; Smith et al., 2008; Szabó et al., 2008). Some of these
studies have been supplemented with process models. Different modelling
techniques have been used to estimate the precipitation process and the effect
of precipitation on the biochemical system (Mbamba et al., 2015b). In general,
these modelling strategies are restricted either to thermodynamic equilibrium
calculations or to a kinetic approach, and only a minor number of these studies
adopted a combination of equilibrium and kinetic approaches to estimate the exent
of precipitation (Barat et al., 2011). These approaches and the relative
contributions have been listed in Table 5.4 and are discussed below.
The thermodynamic equilibrium method for estimating precipitation
makes use of a well established multicomponent thermodynamic equilibrium
approach (Allison et al., 1991; Morel & Morgan, 1972; Rittmann et al., 2002).
Thermodynamic models consider a system comprising of components and species.
Components can interact to form species, but individual species do not react to
form new species. This modelling approach assumes the acid-base system to
consist of a predefined number of organic/inorganic components and species. All
acid-base reactions in the system are considered to be in equilibrium. The system
of equation is solved by the Newton-Raphson algorithm using the equilibrium
constants and the law of mass action (Mbamba et al., 2015b; Solon et al., 2015). A
thermodynamic model for phosphate precipitation was developed to study the
effect of pH, ionic strength and temperature on struvite solubility (Hanhoun et al.,
2011). In the first step, experiments were conducted to determine the solubility
product of struvite from synthetic solution at various temperatures. In the second
step, an algorithm was developed to calculate the equilibrium constants. The
algorithm is based on a hybrid resolution procedure which integrates a
multi-genetic algorithm and the Newton-Raphson method in order to increase
computing efficiency. Using this approach, the precipitation of struvite was
predicted both quantitatively and qualitatively.
A predictive approach, to determine the amount of precipitate formed when
arbitrary amounts of N, Mg and P were added in wastewater system, was
developed by (Scott et al., 1991; Wrigley et al., 1992). The computer program
used NH3 as the central species for calculating the precipitated composition. It
also included a routine to calculate the activity coefficients of each species
involved. After selecting H+ and Mg2+ as main species, concentration ratios (K )

by guest
by guest
Table 5.4 Representative kinetic equations for mineral-precipitation modelling.
118
No. Reference Model Equations Notation

1 Ali & Schneider where
PSO 1/v
(2008) SI = log
Pcs • SI is the saturation index;
n • PSO is the concentration of total species;
G = KS
• Pcs is the conditional solubility product;
dM 1 • v is the number of reactants;
= KN prc L2 Sn Ln
dt 2 • G is the growth rate of crystals;
• K is a kinetic parameter;
• n is a kinetic parameter;
• N is the crystal number;
• rc is the density of the crystals;
• L is the final size of the crystal;
• S is the function of supersaturation of crystal;
• M is concentration of precipitated crystal.
gi mi + gj mj

2 Ebigbo et al. SI = where
ksp
(2012)
• SI is the saturation index;
• γi and γj are the activity coefficient of the
components;
• ksp is the solubility product;
• mi and mj are the concentration of
the components.
IAPim −1
3 Jeen et al. where
Rim = Keff,i
kim
(2012)
• Rim is the rate of mineral precipitation;
• Keff,i is the effective rate constant of precipitation;
• IAPim is the activity product;
• kim is the equilibrium constant for dissolution.
by guest
4 Mbamba et al. rcryst = kcryst Xcryst sn where
(2015a,b) n
Zi Zj 1/v • rcryst is the rate of crystallization;
s = kcryst Xcryst −1
k sp • kcryst is the kinetic rate constant for crystallization;
• Xcryst is the concentration of crystal formed;
• s is the saturation index;
• Zi and Zj are the concentrations of participating
components;
• v is the summation of charges;
• n is an experimentally determined constant;
• ksp is the solubility product.
ksp
5 Rittmann et al. Rp = k ′a 1 − Me where
Q
(2002)
• Rp is the rate of precipitation;
• Me is the concentration of cationic species
(i.e. TEs);
• k ′a is the lumped product of reaction or mass
transfer rate times;

• a is the specific surface area;
• ksp is the solubility product;
n • Q is the reaction product of precipitate.
d[Mv+ Av− ] v+ v− 1/v ′ 1/v
6 Koutsoukos = kr [[M m+ ] − [Aa− ] ] − (ksp ) where
dt
et al. (1980)
• kr is a kinetic constant;
• [M m+ ] and [Aa− ] are the ion concentration;
• [Mv+ Av− ] is the mineral precipitate concentration;
• k ′sp is the solubility product;
• v is the summation of opposite charges v + + v − ;

• n is a constant with value 2 for sparingly
soluble salts.
(Continued)
119
by guest
Table 5.4 Representative kinetic equations for mineral-precipitation modelling (Continued). 120
No. Reference Model Equations Notation
7 Maurer where
dXP X P −1
et al. (1999) = kP fi mi fj mj K P K PPT +
dt X TS • XP is the mineral precipitate;
• mi and mj are the concentrations of soluble ionic
components;
• kP is a kinetic constant for precipitation;
• KP is the hyperbolic coefficient;
• XTS is the conentration of total solids;
• fi and fj are the activity coefficients.

9 Shaojun and R = Rf − Rb = kf gi mi )n1 gj mj )n2 − kb where

Mucci (1993)
• Rf and Rb are the forward and backward reaction
rates;
• k f and kb are the forward and backward rate
constants;
• mi and mj are the concentrations of ionic

components;
• γi and γj are the activity coefficients of the
components;
• n is the reaction rate order.
were assigned to all the equilibria followed by an approximation of the coefficients

by a set of quadratic equations. Once the quadratic equations were set and solved,
the next step involved the estimation of PO3− 4 using the charge balance
(electroneutrality) of the system. Consequently, the system of equations was
solved for other species, checked for consistency and the range and output
plotted. The authors argued the importance of a theoretical estimation to guide
further laboratory and field experimentation.
Equilibrium calculation programs have also been used to study precipitation.
These programs calculate the equilibrium composition based on equilibrium
constant approach. A new method to predict struvite potential in AD systems was
developed by Ohlinger et al. (1998). This method included the ionic strength,
ionic activities, magnesium phosphate complexation effects on ion speciation and
the experimentally determined struvite solubility constant. The method was
verified using MINEQL+. MINEQL+ was used to calculate a set of theoretical
equilibrium concentration values for ions and complexes present in the system.
Theoretical total magnesium, total ammonia, and total phosphorus concentrations
were calculated. The measured theoretical concentrations and variances from lab
tests were used to produce an objective function. The objective function was
minimized to arrive at a solubility constant, pKSP = 13.26 for struvite.
The thermodynamic approach to quantify precipitation in aqueous environments
generally involves the prediction of the equilibrium composition of the system at
a given point in time. Such a method does not provide information about the
dynamics of the species or components taking part in the precipitation reaction.
Additionally, studies based on thermodynamic approach did not take into account
the biological processes taking place in the system. Hence, the influence of the
biological processes, in terms of pH, on precipitation and vice-versa was
largely neglected.
Some studies (Barat et al., 2011; Mbamba et al., 2015a,b) adopted a method
in which the thermodynamic and kinetic approaches were combined to model the
system. In practice, some of the reactions were described by adopting the
equilibrium approach whereas others processes were kinetically solved. Such a
choice largely depends on the process being modelled and the precipitation
reaction under study. More often, protonation/deprotonation reactions are
modelled as equilibrium based while slow precipitation reactions are kinetically
controlled. For instance in Barat et al. (2011), a calcium phosphate precipitation
model was incorporated into the Activated Sludge Model No.2 (ASM2d). The
model considered two types of reactions resulting in precipitation of calcium
phosphate. The aqueous phase reactions (fast) and ion-paring reactions were
modeled as equilibrium reactions by adopting the equilibrium approach, while
the aqueous to solid-phase reactions (slow) were kinetically modelled. A
surface-based kinetic was considered for precipitation reactions (solid phase).
The use of a complete kinetic approach would have required one differential
equation per species (28 in total), while the use of the equilibrium approach

by guest
reduced the problem to 7 differential equations, one for each of the component, thus
increasing the computation efficiency.
Over the course of years, full kinetic models evolved as well. The kinetic
method of estimating the precipitation of mineral phases in wastewater systems
assumes all the physicochemical reactions in the system as kinetic reactions in a
dynamic state equation set. This set of differential equations is solved with
respect to the variables describing the system. With such a method, the time
varying nature of the species and components can be effectively traced and used
to set up a decision-making strategy. For example, a dynamic model was
formulated by Maurer & Boller (1999), based on the experimental observations
that, in inactivated sludge containing high amounts of dissolved calcium
and phosphorus, a pH-sensitive and reversible precipitation process exists.
The model described the calcium and phosphate precipitation by using
protonation/deprotonation reactions of phosphoric acid, carbonic acid and water.
It also included fully reversible precipitation of hydroxydicalcium phosphate
(Ca2 HPO4 (OH)2 , HDP) and formation of hydroxyapatite (Ca5 (PO4 )3 OH, HAP)
from HDP. HAP is considered to be a stable product. All the reactions in the
system were kinetically modeled. This approach was preferred because the
effluent concentrations of the waste treatment plants are generally modeled as
dynamic state variables.
The first and foremost attempt to model the three phases of an AD system, in the
context of mineral precipitation, came from Musvoto et al. (2000a, b). In this study,
a kinetic model for the carbonate system was developed. Ion-pairing and
precipitation reactions were included in the model to describe chemical changes
due to the formation of struvite, newberyite, amorphous calcium and magnesium
carbonate. Further, stripping of CO2 and NH3 was considered. Aqueous phase
mixed weak acid/base chemistry and ion-pairs constituted the ion-pair module.
This was achieved by implementing a similar kinetic model for both the
association/dissociation acid/base reactions and the equilibria of ion-pair
formation. The aqueous phase weak acid/base submodule involved water,
carbonate, phosphate, short chain fatty acid and ammonium. In total, 15
components and 26 processes were defined to implement the aqueous phase weak
acid/base submodule. Likewise, to implement the ion-pair submodule, a total of
12 compounds and 22 processes were used. Here it should be noted that the
forward reaction rate for ion pair formation was selected to be very high (107 s−1)
and the reverse rate was calculated subsequently from the stability constant
values. As mentioned earlier, the model also included a sub-module for
multiple mineral precipitation. Attention was given to magnesium, calcium
and few other relevant minerals. Four possible magnesium phosphates
(MgNH4 PO4 · 6H2 O, MgHPO4 · 3H2 O, Mg3 (PO4 )2 · 8H2 O, Mg3 (PO4 )2 · 22H2 O),
five calcium phosphates (Ca5 (PO4 )3 OH, Ca3 (PO4 )2 , Ca8 (HPO4 )2 (PO4 )4 ·
5H2 O, CaHPO4 , CaHPO4 · 2H2 O) and other minerals (MgCO3 · 3H2 O) were
considered. A concentration-based kinetic rate formulation was used with a default

by guest
order of reaction of 2. Solubility product values were modified according to the

Debye-Hückel rule for low and medium saline waters. The solubility product
values did not consider the temperature variation. Calibration and validation of the
model was carried out by experimental batch results from a spent wine upflow
anaerobic sludge bed (USAB) digester and a sewage sludge digester. Further,
experimental values from literature were also used to fit the model simulations.
The model was implemented on the AQUASIM platform (Reichert 1994). The
quantitative estimation of precipitation of carbonates and phosphates as well as the
amount of CO2 and NH3 stripping from the digester was justified.
In a recent study (Mbamba et al., 2015a), a set of detailed experiments to support
a precipitation modelling approach was carried out. The modelling approach was
applied to the case of calcite precipitation. The model was designed keeping in
mind the adaptability to Generalized Physicochemical Modelling Framework
(Batstone et al., 2012). Furthermore, the authors discussed possible usable control
strategies and nutrient recovery models. pH titration tests and constant
composition experiments were carried out to define the base line of the model
approach and then deduce the environmental factors on the baseline, respectively.
Ion-pairing and acid-base reactions were formulated algebraically following the
mass action principle. Parameters from thermodynamic databases were used to
formulate correct stoichiometry and to assign appropriate equilibrium constants
to the chemical reactions. Chemical activities of each component were also
considered by multiplying for a correction factor. Davies equation with
temperature correction was used to derive the activity coefficients. The pH was
predicted by proton balance method, which considers calculations of total
hydrogen using a Tableau method. The resulting algebraic equations were solved
iteratively by the Newton-Raphson method. Further, precipitation and dissolution
of the chemical constituents were considered based on the Saturation Index SI
approach. This approach necessitates the use of predefined conditions based on
supersaturation to select or omit a given chemical precipitation reaction. For
example, formation of CaCO3 can be considered in the model if the saturation
index reaches a value which supports formation of precipitate. In general, three
conditions were used: SI , 0, SI = 0 and SI . 0. The semi-empirical
precipitation kinetics was based on total dissolved calcium (SCa2+), total dissolved
carbonate (SCO2− 3
) and total particulate calcium carbonate (XCaCO3). It is important
to note that CO2 stripping was not considered in the model. The reason suggested
by the authors corresponds to the low availability of inorganic carbon in liquid
phase. This was entrapped by the addition of NaOH which caused little to no
stripping of CO2 . The predictions of the aqueous phase model were compared
with Visual MINTEQ (Version 3.0, Royal Institute of Technology (KTH)). The
set of algebraic and ordinary differential equations were solved using an ODE
solver and the Newton-Raphson method.
A general methodology to include physicochemical processes in multiphase
wastewater treatment models under a plant-wide modelling framework was

by guest
developed in (Lizarralde et al., 2015). The model emphasized the definition

and selection of biochemical, chemical and physicochemical processes taking
place in a multiphase environment. The physicochemical plant-wide modelling
methodology involved: (1) the definition of the model components and
transformations; (2) the mass-transport definition for a multiphase model of
wastewater treatment plant; and (3) the numerical solution procedure. The first
step in this methodology, demands the compilation of the biochemical, chemical
and physicochemical processes. This was achieved by denominating and
specifying the relation between: (1) COD, N, P or S removal to biochemical
model; (2) acid-base equilibrium and ion-pairing equilibrium to chemical model;
and (3) liquid-gas transfer and liquid-solid transfer to physicochemical model.
Subsequently, important transformations were individuated and introduced in
the model by means of relations between dynamic state variables. The liquid-
solid transfer processes involved precipitation and dissolution processes. The
major precipitates were defined and included in the model. They include
CaCO3 , MgCO3 , Ca3 (PO4 )2 , struvite, k-struvite and newberyite. A kinetic model
based on ion activity product (IAP) was reformulated to take into account the
spontaneous nucleation in addition to the development of supersaturation,
nucleation and growth of the solids. Interestingly, dissolution was considered as the
reverse of precipitation process, which made it easier to implement. The ion-pairing
processes in the chemical model are built at the end in conjugation with acid-base
equilibrium processes. The relevant chemical components and species were selected
and represented as generic dissociation reactions. The chemical equilibrium could
be solved through ordinary differential equations or by algebraic equations.
5.3.1.2 ADM1-based precipitation models

Soon after the publication of ADM1, the model received a considerable amount
of criticism for some errors and omissions (Batstone et al., 2006; Kleerebezem &
van Loosdrecht, 2006). These concern mostly the complexity of the model
structure, the stoichiometry and inhibition kinetics of the biological processes, the
number and the uncertainty of the parameter values. Requests were also received
for the inclusion of important anaerobic processes such as sulfate reduction
(Batstone et al., 2006; Kleerebezem & van Loosdrecht, 2006), phosphorus
metabolism and physicochemical processes (mineral precipitation in particular).
These omissions were acknowledged by the authors of the model and a
‘Generalized Physicochemical Modelling Framework’ was developed (Mbamba
et al., 2015a). However, multiple studies have been carried out with modified
versions of ADM1. These studies include extensions for both biochemical
(Federovich et al., 2003; Galí et al., 2009; Parker & Wu, 2006; Peiris et al.,
2006) and physicochemical processes (Batstone & Keller, 2003; Flores-Alsina
et al., 2016; Maharaj et al., 2018; Zhang et al., 2015). Here we discuss the
ADM1 extensions to include precipitation processes.

by guest
Calcium carbonate (CaCO3 ) precipitation was implemented as a minor structural

modification to the standard ADM1 (Batstone & Keller, 2003). To predict
CaCO3 precipitation, three additional state variables (SCO2− 3
, SCa and SCaCO3 )
representing the concentrations of CO2− 3 , Ca
2+
and CaCO3 respectively, were
defined and added. Simple first-order kinetic rate equations for HCO− 3 /CO3
2−
8
acid-base reactions and CaCO3 precipitation reaction were also added. DH values
of precipitation and acid base reactions were used to calculate the change in value
of the equilibrium coefficients due to temperature. The model was used to assess
two case studies: (1) recycled paper-mill wastewater fed to UASB, and (2) gelatine
production wastewater fed to solids digester.
A more detailed physicochemical framework was implemented by (Zhang et al.,
2015) to simulate the dynamics of calcium caonate (CaCO3 ), magnesium carbonate
(MgCO3 ), struvite (MgNH4 PO3 ), magnesium phosphate (MgHPO4 ) and tricalcium
phosphate (Ca3 (PO4 )2 ). ADM1 was modified by improving the biochemical
framework and integrating a more detailed physicochemical (gas transfer and
precipitation) framework. The changes in inorganic carbon and nitrogen for decay
processes were considered and carbon and nitrogen balances were closed by
adding balance terms of inorganic carbon and nitrogen for the microbial decay
processes. Two balance terms based on a previous study (Blumensaat & Keller,
2005) were introduced into the original ADM1 to resolve the discrepancies
between carbon and nitrogen contents in the degraders and substrates. In the
physicochemical framework, an additional gas transfer process was incorporated
to account for NH3 . Further, inorganic components for acid-base reactions and
solid precipitation were added. The inorganic carbon components included
dissolved carbon dioxide, bicarbonate and carbonate. Similarly, inorganic
phosphate components such as phosphate, hydrogen phosphate, dihydrogen
phosphate and phosphoric acid were introduced into the model. Consequently,
the charge balance was modified to accommodate the new components. A
second-order irreversible precipitation model was used to improve the ability of
the model to simulate non-biologically mediated processes. The model equation
used in this study is based on the fundamental relation for crystallization process
which can describe precipitation processes better than any simple first-order rate
equations based on pseudo-equilibrium. The model was validated with literature
data. The model was used to assess the effect of calcium ions, magnesium ions,
inorganic phosphorus and inorganic nitrogen on a batch anaerobic digester.
A series of model extensions, aimed at improving the interactions of phosphorus,
sulfur and iron, mineral precipitation processes, were implemented in the framework
of ADM1. The extended ADM1 was the part of a larger plant-wide model.
The model extension (A3) included: (1) aqueous chemistry model (Solon et al.,
2015), and (2) a multiple mineral precipitation model, MMP (Mbamba et al.,
2015b). The first submodule estimates the pH by considering ionic behaviour of
non-ideality (ion-pairing and ion activity) instead of molar ion concentration. A
set of non-linear algebraic equations were used to resolve the aqueous chemistry

by guest
module. The method includes: (1) Davis equation based ionic strength correction;
(2) ion-pairing equilibrium reactions for inorganic carbon, inorganic nitrogen and
VFA; and, (3) multi-dimensional Newton-Raphson method to solve the set of
algebraic equations. In the second submodule of MMP, a reversible mechanism of
precipitation has been implemented using saturation index. This methodology
of predicting precipitation reactions has been already explained in detail (Mbamba
et al., 2015a, and Section 5.3.1.1). The MMP model simulates the dynamics of
calcite (XCaCO3), agronite (XCaCO3a), amorphous calcium phosphate (XCa3 (PO4 )2),
hydroxylapatite (XCa5 (PO4 )3 (OH)), octacalium phosphate (XCa8 H2 (PO4 )6), struvite
(XMgNH4 PO4), newberyte (XMgHPO4), magnesite (XMgCO3), k-struvite (XKMgPO4),
iron sulfide (XFeS .), iron phosphate (XFePO4 /XFe3 (PO4 )2), and aluminium phosphate
(XAlPO4).
Recently, a model based on ADM1 has been proposed which explicitly predicts
the dynamics of TEs (Fe, Co and Ni) in terms of mineral precipitation in an
anaerobic batch reactor (Maharaj et al., 2018). In the biochemical module, the
microbial uptake of TE and the TE inhibition on microbial activities have been
introduced. A dose response function has been implemented to model the
effect of TEs on anaerobic production of methane. The function considers
deficiency, activation, inhibition and toxicity of TEs on the microbial groups. In
the physicochemical module, ion association/dissociation reactions and
liquid-gas transfer reactions are considered for liquid-liquid processes and
liquid-gas processes, respectively. Equilibrium reactions are modeled by a set of
differential equations. The incorporation of the precipitation reactions in the
physicochemical module leads to the definition of new inorganic components in
the ADM1 framework. The model takes into consideration inorganic carbonate
(e.g. CO2 , HCO− 2− 3− 2−
3 and CO3 ), inorganic phosphate (e.g. PO4 , HPO4 , H2 PO4
−
− 2−
and H3 PO4 ) and inorganic sulfide (e.g. HS and S ) components in liquid
phase. The components of the three chemical systems (carbonate, phosphate and
sulfide) react, in the liquid phase, to form precipitates (e.g. CoCO3 , Co3 (PO4 )2 ,
FeS, FeS, FeCO3 , Fe3 (PO4 )2 , NiCO3 , Ni3 (PO4 )2 , NiS, MgNH4 PO4 ), whose
formation is governed by the Ksp values. A full kinetic framework has been used
to implement the precipitation process. The model was implemented on an
original code in MATLAB platform and has been solved using ‘ode15s’ routine.
The model was used to study the effect of changes in the initial concentration of
sulfur and phosphorous, the effect of nutrient starvation on methane production
and the dynamics of TEs in AD.
5.3.2 Adsorption
Adsorption of TEs in AD can proceed through three different mechanisms:
(i) adsorption of TEs on precipitates forming in the digester; (ii) complexation of
TEs on organic matter released by the microorganisms; and (iii) adsorption of
TEs on the surface of the microorganisms. In the first case, when a precipitate

by guest
is formed it provides many adsorption sites for various cations and anions. In the
second and third case, various carboxylic and amino groups of organic surfaces
provide sites for the surface complexation of TEs. In general, TE adsorption
in biogeochemical systems has been studied and reviewed to a great extent
(Konhauser, 2007; Warren & Haack, 2001). However, only few attempts to
model the adsorption process in wastewater treatment systems exist in literature
(Hauduc et al., 2015), and up to date they have not been implemented in the
ADM1 framework to study the effect of TEs adsorption on AD performance. In
the following section, we report the modelling efforts to simulate the adsorption
process in wastewater environments. For a general overview of the models
adopted for biosorption the reader can refer to Papirio et al. (2017).
The general strategy to model adsorption processes is derived from geochemical
reaction modelling techniques. In this approach, a chemical equilibrium problem is
formed comprising the major chemical reactions taking place. Further, the equations
are solved simultaneously by the Newton-Raphson algorithm. The equilibrium
constants of the individual reactions within the system are of major importance
because such models consider every physicochemical reaction as being in
equilibrium. For example, the mechanisms behind the chemical-mediated
phosphorus system was studied in (Smith et al., 2008). The study was aimed at
investigating the major factors affecting phosphorus removal and also the
effect of various operational parameters. The model took into consideration
all the possible surface reactions on hydrous ferric oxide. The basis of the model
is that iron and phosphorus share an oxygen atom. The chemical equilibrium
problem was divided in two parts for numerical reasons. The first part solves or
quantifies the thermodynamically-favored solid precipitates formed (Fe(OH)3(s)
and FePO4(s) ). Once the precipitated solids are estimated, a second calculation step
is performed solving a second equilibrium problem where phosphate is allowed to
complex with active sites (adsorption sites) on the precipitated hydrous ferric
oxide (HFO). As the authors pointed out, phosphate removal should be considered
with phosphate precipitation. However, the surface complexation modelling
formalism was adopted in this case to remove phosphate from the system in the
earlier reaction step (hydrolysis). This was done by kinetically linking the process
to the mixing.
In other cases, the kinetic approach to model the adsorption process in a
wastewater system has been adopted. In such case, the adsorption of a chemical
species is considered to be kinetically controlled. The kinetic approach to define
the physicochemical system is easier to implement because in most of the
biochemical framework, AD models are kinetically controlled rather than
equilibrium controlled. The process of chemical phosphorus removal in
wastewater was kinetically modelled by (Hauduc et al., 2015). Such dynamic
model serves as a tool to optimize chemical dosing, taking into consideration
the effluent phosphorus concentration due to regulatory reasons. The model
predicts the stoichiometry and kinetics of precipitation of hydrous ferrous oxides

by guest
(HFO), phosphate adsorption and co-precipitation. Thus, chemical equilibrium

dissociation, chemical ion pairing, physical mineral precipitation process,
chemical surface complexation, and aging of precipitates were considered in the
model. Kinetic models used for adsorption processes in wastewater treatment are
limited to pseudo-first-order rates (Lagergren type). The adsorption equation
expresses the variation of component concentration as function of a driving force
which is the difference between the amounts of the component that should be
bound at equilibrium. This is calculated based on the equilibrium constants and
actual bonded component. For monodentate and bidentate species the order of the
reaction is one and two, respectively. The adsorption sites on the flocs are not
equally accessible to phosphorus. More accessible sites are available first, which
reduces in number with the course of adsorption. Therefore the rate of adsorption
decreases. This decrease in kinetic rate due to a decrease in accessibility of
remaining sites have been modeled by multiplying the ratio (SiteF/SiteT)n to the
overall rate, where, n is 1 for monodentate and 2 for bidentate species. The ratio
is the amount of free sites (SiteF) to the amount of total sites (SiteT) with a value
between 0 and 1.
A biogeochemical framework (CCBATCH) was implemented and expanded
by adding surface complexation reactions to the already developed
biodegradation and chemical equilibrium sub models (Schwarz & Rittmann,
2007). The surface complexation sub-model included both electrostatic and
non-electrostatic surface complexation reactions. Metal complexation by active
cells, inert biomass (solids) was described by the electrostatic surface
complexation model. The non-electrostatic sub-model was used to describe
metal complexation reactions of extracellular polymeric substances (EPS) and
biomass-associated-product (BAP). The model assumed a negative charge on
the surface of the cell which is due to the protonation/deprotonation of
carboxyl, phosphoryl and hydroxyl groups. The interaction of these negatively
charged surface species with the adsorbing protons was quantified according to
the capacitance model of microbial surface. Further, to understand the
biological and non-biological ligand interaction with cell surface leading to
metal detoxification a three-step procedure was adopted. Stoichiometry and
kinetics of production of reactive ligands by the microbes were mentioned in
the first step. In the second step, reactivity of the ligands towards metal was
described which indirectly described the speciation of the metal in the system.
Lastly, expanded CCBATCH was used to carry out titration reactions
simulations. The biogeochemical analyses emphasized the fact that over various
organic and inorganic microbial products, sulfide is a major promoter for Zn
detoxification. It was shown that, in presence of sulfide, detoxification of Zn
occurs while in absence of sulfide, Zn complexes with biogenic organic ligands
providing a mechanism for detoxification.
A surface-complexation modelling approach to describe the metal ion interactions
with microbial surface has also been studied in (Daughney & Fein, 1998). This

by guest
modelling technique makes use of stability constants to model metal-microbial

surface interaction. However, stability constants for metal-microbial surface
complexes, which form the backbone of this modelling technique, are rarely
reported because of the intricacies of the experimental methods involved in their
determination. Therefore, the stability constants were predicted using a linear free
energy approach. In this regard, measured microbial metal-carboxyl stability
constants are related to stability constants for aqueous metal-organic acid anion
complexes that involve the same metal cation. If the correlation between these
two types of stability constants is valid, an unknown value for a metal-carboxyl
microbial surface stability constant can be estimated provided the stability
constant for the metal-organic acid anion aqueous complex is known. Whether
such a relationship could be constrained for a widely studied organic acid anion,
it would provide a means of estimating the microbial adsorption behavior of a
wide range of metal cations.
5.3.3 Aqueous complexation

Anaerobic digestion is rich in organic matter. The organic matter is composed of
biomass, humic substances and other organic molecules from the biodegradation,
such as VFAs, alcohol and acetate. Their formation is one of the major
biochemical reactions in AD. These organic substrates and humic substances act
as chelating agents affecting metal speciation. Thiol-containing organic matter, as
represented by cysteine, has also been reported to take part in aqueous
complexation with TEs as a strong functional group (Yekta et al., 2014a, b).
Additionally, in the course of AD, synthetic chelating agents are occasionally
added to improve the digester performance. Synthetic agents, such as EDTA,
influence (imparting both positive and negative effects) the bioavailability of TEs
in AD by forming stable soluble complexes. Thus, three groups of organic
species should be taken into consideration for organic complexation modelling:
(1) organic substrates in AD (for example, VFAs); (2) humic substances; and (3)
synthetic chelating agents (EDTA/NTA/EDDS). Here we discuss the general
methodology developed in (Willet & Rittmann, 2003) and adopted in (Maharaj
et al., 2019) to model TEs complexation in the ADM1 framework.
The formation and dissociation of TE complexes follows a kinetically controlled
approach. The complexation reaction involving organic ligands and TEs was used
here to illustrate the kinetic framework to model complexation reactions. New
state variables representing the organic ligand acid/base system and complexes
have to be taken into account to predict the pH of the system. The general
mechanism of complexation reactions and modelling can be written as follows:
k1
Mex + Ly [MeL]x+y (5.20)
k−1
[MeL]x+y Mex + Ly (5.21)

by guest
where k1 is the [MeL]x+y formation rate constant in [ML−3 T−1]; k−1 is the
dissociation constant in [T−1]. Equations (5.16) and (5.17) can be rewritten as:
K[MeL]x+y
Mex + Ly [MeL]x+y (5.22)
−3
where K[MeL]x+y is the equilibrium constant for [MeL] x+y
in [ML ]. The rate of
change of complex over time may be written as:
d[MeL]x+y
= k1 [Mex ][Ly ] − k−1 [MeL] x+y (5.23)
dt
where [Mex ], [Ly ], [MeL]x+y are the dynamic state variables for the free TE
concentration, organic ligand and TE-organic complex respectively. The
formation rate constant, k1 , for all the species and the dissociation rate constant,
k−1 , can be calculated from the stability constant by a simple relation as:
k1
K[MeL]x+y = (5.24)
k−1
5.4 MODELLING THE EFFECT OF TEs ON BIOKINETICS

5.4.1 Modelling the biouptake of TEs
It is widely accepted that metal geochemistry and environmental conditions in an
engineered system affect the uptake of TEs by microorganisms (Hudson, 1998;
Jansen, 2004; Sunda & Huntsman, 1998; Worms et al., 2006). The bioavailability
of metals in the case of AD has been recently studied (Gustavsson, 2012;
Jansen, 2004; Yekta et al., 2016). Bioavailability is thought to depend on: (i) the
internalization pathway; (ii) specificity of metals to the transporters; (iii)
physicochemistry of the bulk phase; (iv) size and nature of microorganisms;
(v) concentration of metal; and (vi) metal speciation.
To effectively affect the biochemical systems, the required metals have to be
transported from the bulk solution into the Cytosol (Figure 5.3). In this process
metals diffuse through the external medium to the surface of the organism, they
are subsequently adsorbed to the microbial surface and later internalized by the
corresponding mechanisms. These transport processes control the overall metal
bio-uptake rates (Jansen, 2004; Koster & Leeuwen, 2004; Worms et al., 2006).
The binding of metals to the surface of the microorganisms are thought to be
carried out by the extracellular polysaccharides (Bhaskar & Bhosle, 2006;
d’Abzac et al., 2010; Malik, 2004; Loaec et al., 1997). Therefore, the rate of
extracellular polysaccharide formation on the microorganisms might be linked to
the presence of a certain amount of a particular metal in the surrounding
environment (Aquino & Stuckey, 2004). The surface-bound metals are destined
to be internalized, accumulate on the surface or dissociate back to the bulk phase
(Worms et al., 2006). The internalization of metals depends largely on the
transport system (Hudson, 1998; Jansen, 2004; Worms et al., 2006). The

by guest
Figure 5.3 Various physico-chemical and transport mechanisms affecting TE

internalization in a microbial cell. L and TEn+ refer to ligand and free trace element
respectively. Subscripts int: internalized; bio: microbial; h: hydrophilic; L: lipophilic
(adapted from Worms et al., 2006).
hydrophobic nature of biological membranes limits the movement of molecules by

passive diffusion. Only neutral and non-polar molecules move across the membrane
by passive diffusion. Difference in concentration gradient across the membrane is
the driving force for passive diffusion of molecules across the membrane. Metal
species in AD are generally hydrophilic in nature and their transport is mediated
by specific metal-binding proteins (Worms et al., 2006). For example, metals are
transported as metal ions involving ATPases, natural resistance associated
microphage proteins and zinc-regulated or iron-regulated transporter. These two
steps of metal binding to the surface and metal internalization by specific
transporters have been expressed by Michaelis-Menten kinetics (Jansen, 2004).
Further, the mass balance for total TE has been established. Starting with the case
of growth limitation, metal concentration in the bulk phase can be related to
growth by Monod kinetics. As Monod kinetics assumes the external metal
concentration in the bulk phase, it might become erroneous to model internal
metal concentration. Droop model of relating internal metal concentration to the
growth renders an option here (Droop, 1983; Jansen, 2004). Microbial growth
rate was defined as:
dx
= mx (5.25)
dt
where, m is the growth rate constant, t is the time, and x is the biomass concentration.
The limitation of growth rate due to substrate and available metal ions was

by guest
represented as follows:
CMeOH (QCo − QCo,min) (QNi − QNi,min)
m = mmax (5.26)
CMeOH + KMeOH QCo QNi
where mmax is the maximum growth rate constant, CMeOH is the primary substrate
concentration, KMeOH is the substrate Monod constant, QCo is the cobalt content;
QCo,min is the minimum Co content, QNi is the Ni content and QNi,min is the
minimum Ni content. The competition of metal ions for the same transporter was
taken into account through the uptake flux rate as follows:
CCo2+ /KJM,Co2+
JCo,in = Jmax (5.27)
(CNi2+ /KJM,Ni2+ ) + (CCo2+ /KJM,Co2+ ) + 1
where Jmax is the maximum uptake flux; KJM,Co2+ is the Co Michaelis-Menten
constant, KJM,Ni2+ is the Ni Michaelis-Menten constant, CCo2+ is the free Co2+
concertation, and CNi2+ is the free Ni2+ concentration. The excretion of metal
ions was modelled as a first order process as follows:
JCo,eff = keff, Co (QCo − QCo,eff,min ) (5.28)
where keff,Co is the efflux rate constant and QCo,eff,min is a threshold value below
which no efflux takes place. The mass balances were established based on the
above-mentioned rates. The mass balance for metal content of the cells was
represented as follows:
dQCo
= JCo,in (t) − JCo,eff (t) − m(t)QCo (t) (5.29)
dt
The amount of total dissolved metal was accounted for with the following mass
balance:
dQCo,total
= −(JCo,in (t) − JCo,eff (t))x(t) (5.30)
dt
5.4.2 Dose-response modelling of TE effect in AD

Elements in trace amounts such as Fe, Zn, Ni, Co, Mo, Cu, Mn, Se, Mo, W, and
B considerably affect biogas production in anaerobic digesters by stimulating
and enhancing microbial activity (Choong et al., 2016; Federation, 2014; Feng
et al., 2010; Fermoso et al., 2009; Oleszkiewicz & Sharma, 1990; Romero-güiza
et al., 2016). Optimal levels of TEs should be maintained by adding externally
the appropriate micronutrients, especially when the feedstock used for AD is
deficient in these elements. It is a common industrial practice to supply these
elements externally to ensure that these will be in bioavailable form to cover
the nutritional needs of the microorganisms in the anaerobic environment.
However, increased concentrations of the same elements can act as inhibiting
or toxic factor and result in a reduced biogas production or complete reactor

by guest
failure (Feng et al., 2010; Hickey et al., 1989; Lin, 1992). Working within the
optimum concentration range is required to secure balancing between deficiency
and toxicity and ensure minimum operating cost for additives supply. This
behaviour is conceptually presented in Figure 5.4(a) for an essential TE. When an
element is considered ‘non-essential’ for microbial growth, the deficiency part of
Figure 5.4(a) is narrowed or supressed. In such cases, the effect of the element on
the microbial growth presented conceptually in Figure 5.4(b) can be observed.
Mathematically, this behaviour is approached by dose-response models. These
models are extensively used in pharmacokinetic (PK) studies (Zhao & Yang,
2014), where the effect of dosing a medicine to a sample population is monitored
by measuring some physiological response. The same concept can be applied for
modelling the effect of TEs on the biological activity in anaerobic digesters.
Figure 5.4 A conceptual dose-response curve for (a) an essential and (b) a
non-essential element.
In typical dose response studies only the part of the response up to the maximum
effect level is modelled (left hand side of Figure 5.4(a)). A sigmoidal curve usually
adequately describes this part of the curve. A number of mathematical expressions
simulate such a sigmoidal response. Selecting the appropriate model depends on the
following aspects.
Theoretical basis of the model: Very few sigmoidal models are derived from
a theoretical analysis of the dose-response effect. The majority of the models
are empirical equations or mathematical expressions sharing sigmoidal
curve characteristics.
Number of parameters in the model: The number of parameters affects the shape
(steepness) and position of the reflection point of the sigmoidal curve. It usually
ranges from two to five (excluding y0 and ymax values). Equations with fewer
parameters are preferred for simplicity. For the case of simple models with only
two parameters, the linearized form of the equation can be used to determine the

by guest
best-fit values simulating an experimental set of data. Generally this procedure is

carried out by nonlinear regression analysis.
Position of the reflection point: The position of reflection point of the sigmoidal
curve is also a significant criterion to select a dose response model for simulation.
The reflection point can be either fixed positioned at 50% of effective dose
(ED50) (e.g. Emax model) or parametrically adjustably above or below ED50.
The concept of dose-response modelling can be directly applied in modelling
the effect of TEs in anaerobic digesters. The effective concentration applied
for each TE should be mathematically correlated with the induced response in the
system. Therefore two critical parameters should be defined: (a) which is the
selected response and at which time should the system be sampled (observed y
values), and (b) which effective concentration of the TE produces a certain
level of response (measured x values). Ideally, direct effects such as
increased microbial activity related with higher observed growth rates and/or
increased yield coefficients should be quantified experimentally. Microbial
growth rate expressions can be modified appropriately by including terms for the
effect of TEs. When Monod kinetic expressions are used, these can be modified
as follows:
Si Sj
ri = mmax,i ··· . . . f (Ii ) . . . f (Ij ) . . . f ( pH)f (T) . . . f (TEi ) . . . f (TEj )
Ki + Si Kj + Sj
(5.31)
where ri is the specific growth rate of microorganisms in process i [ML−3

T−1]; mmax,i is the maximum specific growth rate of microorganisms in process i
[T−1]; Si . . . Sj is the concentration of limiting substrates i…j e.g. carbon source,
nitrogen source etc. [ML−3]; Ki . . . Kj is the half saturation constant for the
components i…j [ML−3]; f (Ii ) . . . f (Ij ) are the factors for inhibiting substances e.
g. free ammonia, butyrate (dimensionless); f ( pH) is the effect of pH on the
specific growth rate of microorganisms in process I; f (T) is the effect of
temperature on the specific growth rate of microorganisms in process i
(dimensionless); f (TEi ) . . . f (TEj ) is the effect of TEs i…j on the specific growth
rate of microorganisms in process i (dimensionless).
In a traditional modelling approach biomass yield (Y ) and related stoichiometric
coefficients are considered constant, although they may vary significantly during
the growth stages of the microbial biomass and the physicochemical conditions
in cell’s environment (e.g. TEs). When the yield is affected by the presence of
essential TEs then a similar approach can be used:
Y = Y0 f (TEi ) . . . f (TEj ) (5.32)
where Y is the observed yield coefficient in the presence of TEs; Y0, is the maximum
yield coefficient at the optimum level of TEs; f (TEi ) . . . f (TEj ) is the dose response
equations related with the TEs i…j.

by guest
Assuming that an element is essential for growth, any sigmoidal curve

describing the left hand side of Figure 5.4(a) can be used as a modifying
function. The model would result in non-growth conditions when the element is
absent. This corresponds to complete reactor failure due to severe element
deficiency. Conditions of mild deficiency are also modeled by equations (5.27) or
(5.28) at moderate element concentration. Finally, saturation is predicted when
the concentration of TEs are within the optimal range. Equations (5.27) and
(5.28) should be modified appropriately when the observed rates and/or yields
are not bound between 0 and 1 as in a ‘typical’ sigmoidal curve. For example
when the baseline effect is different than zero the appropriate f (TEj ) functions
should be used. Toxicity modelling is also possible by selecting appropriate
mathematical expression such as equations in Table 5.5.
Estimating the bio-kinetic parameters for an anaerobic digester, by fitting a
model to experimental data, assumes that the only limiting or inhibiting
substances that affect the process are these already included in the kinetic
expressions of the model (e.g. first part of equation 5.27). However under severe
or mild TEs deficiency (correspondingly toxicity) the bio-kinetic estimated
parameters might be significantly misleading as the system does not perform
under optimum conditions. The use of an updated revised model would result in
a different set of parameters that include also the effect of the trace elements in
the system.
Another significant aspect which should also be addressed is the selection of
the classes of the microbial consortia present in an anaerobic digester which are
mostly affected by the elements deficiency. It has been shown that the
microorganisms that are more sensitive to TEs fluctuation and deficiency are
those belonging to the group of methanogenic archaea (Glass & Orphan, 2012).
Therefore, the critical step in this modelling approach is to focus primarily on the
metabolic activity of this group by modifying the appropriate kinetic expression
and/or yield coefficients.
It is also important to define what is considered the effective concentration
(dose) of a TE in an anaerobic bioreactor resulting to a certain level of stimulating
response (or correspondingly toxicity effect). In pharmacokinetic studies when
a new medicine is tested in a sample population, a prescribed amount is
administered, while the resulted physiological response is monitored (e.g. changes
in blood pressure). Even if the drug is distributed differently in the human organs,
excreted, accumulated or metabolized, the total amount of the drug administered is
considered as the effective dose. However, in those studies factors such as age,
sex, race and so on, may significantly differentiate the results. Similarly, in the case
of anaerobic digesters, the total concentration of trace elements could be used as a
rough indication of the dose to be used in dose-response equations. However, it is
realized that a significant fraction of the total element in a bioreactor is in
non-readily bioavailable form and will not affect directly biogas production
(Thanh et al., 2015). Part of the element can be distributed according to the

by guest
by guest
Table 5.5 List of dose-response functions. 136
No. Name Dose-response Function General Trend
n
x
1 Hill equation y = ymax
an + x n
where
• ymax, a are parameters of the model
• n is the Hill coefficient

Dn
2 Emax equation y = E0 + Emax
+ Dn
EDn50
where
• Emax is maximum change in effect from base
line level
• E0 is baseline effect (or placebo effect in
pharmacokinetic), corresponding to the response

when the dose is zero (left asymptote parameter)
• ED50 is dose which achieves 50% of the
maximum response
• D is dose
by guest
3 Logarithmic y = ea−b/x
reciprocal where
equation
• a and b are parameters of the model
ymax ymax
4 Logistic y= =
1 + e−(a+bx) 1 + a′ e−bx
equation
where

• ymax, a or a′ and b are parameters of the model
(Continued)
137
by guest
Table 5.5 List of dose-response functions (Continued ). 138
No. Name Dose-response Function General Trend

ymax
5 Log-Logistic y=
1 + ae−b ln(x)
equation
where
• ymax, a and b are parameters of the model
−(a+bx) ′ −bx

6 Gompertz y = ymax e−e = ymax e−a e
equation
where
• ymax, a and b are parameters of the model.
by guest
c
7 Weibull y = ymax − ae−bx
equation
where
• ymax, a, b and c are parameters of the model
a
8 Richard’s y= d
equation (1 + e−(b+cx) )
where

• a, b, c and d are parameters of the model
(Continued)
139
by guest
Table 5.5 List of dose-response functions (Continued ).
No. Name Dose-response Function General Trend 140

ymax−y0
9 Equations y = y0 +
simulating EC50 a x b
1+ 1+
stimulation and x IC50
toxicity effects where
• y is the recorded response
• y0 is the baseline of the recorded response at
concentration levels approaching zero
• ymax is the maximum observed response
• EC50 is the concentration of trace element where
50% of the maximum effect is achieved
• IC50 is the concentration of the element where
50% inhibition (toxicity) is observed
• a and b are parameters that affect the slope of the
left hand side and right hand side of the
curve respectively

(x + a)
10 Equations y =
simulating b + c(x + a) + d(x + a)2
hermetic effects where
• a, b, c and d are parameters of the model
following processes (Callander & Barford, 1983a, b): reversible or irreversible

adsorption on the surface of solids (Section 5.3.2); precipitation to inorganic
minerals by abiotic reactions (e.g. hydroxide, carbonate, phosphate, sulfide
precipitates; Section 5.3.1); volatilization of the elements that produce inorganic or
organic volatile compounds (e.g. volatile selenium compounds); redox reactions
that affect the solubility of the element (e.g. Cr(III)/Cr(VI), Fe(II)/Fe(III), Se(VI)/
Se(IV)/ Se(0)/ Se(-II)); complexation by organic ligands (e.g. proteins, humic
acids, fulvic acids, yeast extract, EDTA, NTA, EDDS; Section 5.3.3)
(Gonzalez-Gil et al., 2003; Zhang et al., 2015) and species distribution according
to the element speciation (e.g. poly hydroxyl species) (Yekta et al., 2016).
Alternatively, as a rough estimate of the effective concentration (dose) it would
be appropriate to use the soluble fraction of the TE present in the reactor liquid
phase. However even under this convention, metal complexes with organic
macromolecules could still be in non-bioavailable form. Thus, it is of primary
concern to determine which chemical species of the TE are actually considered
directly bioavailable to enhanced biogas production and system stability (Yekta
et al., 2016).
From the mathematical point of view, inclusion of terms for TEs in an existing
model unavoidably increases model complexity. For each TE a new state variable
in the liquid phase should be introduced. In addition, a set of at least two new
parameters should be provided to include the appropriate sigmoidal shape
equation in the model. The complexity increases considerably when different
species of the TE should also be included as well as speciation modelling
equations to calculate the concentration of these species. These algebraic
equations corresponding to the equilibrium law for each reaction usually are
treated as instantaneous or fast dynamic equations. Furthermore, processes such
as adsorption, complexation or precipitation of the TEs require definition of
additional processes and their corresponding rates or equilibrium equations.
Providing coherent initial conditions for such a complex system is crucial for the
initialization of the integration algorithm. Additionally, the difference in time
scales between slow biological actions and fast dynamic or equilibrium abiotic
reactions results in system stiffness. This is the cause of solution failure in the
propagation of a solution algorithm.
5.5 CONCLUSION
This chapter reviewed the modelling aspects describing the main processes that
affect TE dynamics in AD environment. The consolidated knowledge from the
literature indicates that the effect and fate of TEs in AD system can be modelled
to predict the dynamic behavior of TEs in AD. However, there is a need to
improve the currently available AD models to account for precipitation,
adsorption, aquatic complexation and bio-uptake of TE, so as to quantify their
fate and effect. While problems may arise due to increase in model complexity,

by guest
the modelling approach should be compatible with the existing AD framework

(ADM1). Consequently, taking into account the multiplicity of processes that
affect TE dynamics in AD systems as well as the importance of these processes
in defining the biology and physicochemistry of the AD system, the need for a
TE module embedded within the ADM1 model has been recognized. This sets
the priorities in the efforts towards ‘mathematical model-based controlled TE
dosing’ in engineered AD systems.
REFERENCES
Ali M. I. and Schneider P. A. (2008). An approach of estimating struvite growth kinetic
incorporating thermodynamic and solution chemistry, kinetic and process description.
Chemical Engineering Science, 63(13), 3514–3525.
Allison J. D., Brown D. S. and Novo-Gradac K. J. (1991). MINTEQA2/PRODEFA2, a
Geochemical Assessment Model for Environmental Systems: Version 3.0 User’s
Manual. U. S. Environ. Prot. Agency EPA/600/3-91/021 115. USEPA, Washington,
DC.
Aquino S. F. and Stuckey D. C. (2004). Soluble microbial products formation in anaerobic
chemostats in the presence of toxic compounds. Water Research, 38, 255–266,
https://doi.org/10.1016/j.watres.2003.09.031
Banerjee R. and Ragsdale S. W. (2003). The many faces of vitamin B12. Annual Review of
Biochemistry, 72, 209–247, https://doi.org/10.1146/annurev.biochem.72.121801.
161828
Barat R., Montoya T., Seco A. and Ferrer J. (2011). Modelling biological and chemically
induced precipitation of calcium phosphate in enhanced biological phosphorus
removal systems. Water Research, 45, 3744–3752, https://doi.org/10.1016/j.watres.
2011.04.028
Barrera E. L., Spanjers H., Dewulf J., Romero O. and Rosa E. (2013). The sulfur chain in
biogas production from sulfate-rich liquid substrates: A review on dynamic modelling
with vinasse as model substrate. Journal of Chemical Technology and Biotechnology,
88, 1405–1420, https://doi.org/10.1002/jctb.4071
Barrera E. L., Spanjers H., Solon K., Amerlinck Y., Nopens I. and Dewulf J. (2015).
Modelling the anaerobic digestion of cane-molasses vinasse: Extension of the
Anaerobic Digestion Model No. 1 (ADM1) with sulfate reduction for a very high
strength and sulfate rich wastewater. Water Research, 71, 42–54, https://doi.org/10.
1016/j.watres.2014.12.026
Batstone D. J. and Keller J. (2003). Industrial applications of the IWA anaerobic digestion
model No.1 (ADM1). Water Science & Technology, 47, 199–206.
Batstone D. J., Keller J., Angelidaki I., Kalyuzhnyi S. V., Pavlostathis S. G., Rozzi A.,
Sanders W. T., Siegrist H. and Vavilin V. A. (2002). The IWA Anaerobic Digestion
Model No 1 (ADM1). Water Science & Technology, 45, 65–73, https://doi.org/10.
2166/wst.2008.678
Batstone D. J., Keller J. and Steyer J. P. (2006). A review of ADM1 extensions, applications,
and analysis: 2002–2005. Water Science & Technology, 54, 1–10, https://doi.org/10.
2166/wst.2006.520

by guest
Batstone D. J., Amerlinck Y., Ekama G., Goel R., Grau P., Johnson B., Kaya I., Steyer J. P.,
Tait S., Takacs I., Vanrolleghem P. A., Brouckaert C. J. and Volcke E. (2012). Towards a
generalized physicochemical framework. Water Science & Technology, 66, 1147–1161,
https://doi.org/10.2166/wst.2012.300
Bhaskar P. and Bhosle N. (2006). Bacterial extracellular polymeric substances (EPS):
A carrier of heavy metals in the marine food chain. Environment International, 32,
1–8.
Blumensaat F. and Keller J. (2005). Modelling of two-stage anaerobic digestion using the
IWA Anaerobic Digestion Model No. 1 (ADM1). Water Research, 39, 171–83,
https://doi.org/10.1016/j.watres.2004.07.024
Callander I. J. and Barford J. P. (1983a). Precipitation, chelation, and the availability of metals
as nutrients in anaerobic digestion. II. Applications. Biotechnology and Bioengineering,
25, 1959–1972, https://doi.org/10.1002/bit.260250806
Callander I. J. and Barford J. P. (1983b). Metals as Nutrients in Anaerobic Digestion.
Biotechnology and Bioengineering, 25, 1959–1972.
Chen Y., Cheng J. J. and Creamer K. S. (2008). Inhibition of anaerobic digestion process:
A review. Bioresource Technology, 99(10), 4044–4064, https://doi.org/10.1016/j.
biortech.2007.01.057.
Choong Y. Y., Ismail N., Abdullah A. Z. and Yhaya M. F. (2016). Impacts of trace element
supplementation on the performance of anaerobic digestion process: A critical review.
Bioresource Technology, 209, 369–379, https://doi.org/10.1016/j.biortech.2016.03.
028
Copp J. B., Jeppsson U. and Rosen C. (2003). Towards an ASM1 – ADM1 State Variable
Interface for Plant-Wide Wastewater Treatment Modelling. Proceeding Water
Environmental Federation, 498–510, https://doi.org/10.2175/193864703784641207
d’Abzac P., Bordas F., van Hullebusch E., Lens P. N. L. and Guibaud G. (2010). Metal
binding properties of extracellular polymeric substances extracted from anaerobic
granular sludges. Colloids Surfaces B Biointerfaces, 80, 161–168, https://doi.org/10.
1016/j.colsurfb.2010.05.043
Daughney C. J. and Fein J. B. (1998). The Effect of Ionic Strength on the Adsorption of H+,
Cd2+, Pb2+, and Cu2+ by Bacillus subtilis and Bacillus licheniformis: A Surface
Complexation Model. Journal of Colloid and Interface Science, 198, 53–77, https://
doi.org/10.1006/jcis.1997.5266
Dezham P., Rosenblum E. and Jenkins D. (1988). Digester gas control using iron salts.
Journal -Water Pollution Control Federation, 60(4), 514–517.
Donoso-Bravo A., Bandara W. M. K. R. T. W., Satoh H. and Ruiz-Filippi G. (2013). Explicit
temperature-based model for anaerobic digestion: Application in domestic wastewater
treatment in a UASB reactor. Bioresource Technology, 133, 437–442, https://doi.
org/10.1016/j.biortech.2013.01.174
Droop M. R. (1983). 25 Years of Algal Growth Kinetics A Personal View. Botanica Marina,
26, 99–112.
Ebigbo A., Phillips A., Gerlach R., Helmig R., Cunningham A. B., Class H. and Spangler L.
H. (2012). Darcy-scale modeling of microbially induced carbonate mineral precipitation
in sand columns. Water Resources Research, 48(7).
England J. L. (2013). Statistical physics of self-replication. Journal of Chemical Physics, 139,
121923, https://doi.org/10.1063/1.4818538

by guest
Federation W. E. (2014). Mineral nutrient requirements for high-rate methane fermentation of

acetate at low In: SRTM. Takashima and R. E. Speece (eds), Research Journal of the
Water Pollution Control Federation, 61(11/12), http://www.jst 61, 1645–1650.
Federovich V., Lens P. and Kalyuzhnyi S. (2003). Extension of Anaerobic Digestion Model
No. 1. Appl. Biochemistry & Biotechnology, 109, 33–45, https://doi.org/10.
1385/ABAB:109:1-3:33
Feng X. M., Karlsson A., Svensson B. H. and Bertilsson S. (2010). Impact of trace element
addition on biogas production from food industrial waste – Linking process to microbial
communities. FEMS Microbiology Ecology, 74, 226–240, https://doi.org/10.1111/j.
1574-6941.2010.00932.x
Fermoso F. G., Bartacek J., Jansen S. and Lens P. N. L. (2008). Metal supplementation to
UASB bioreactors: From cell-metal interactions to full-scale application. The Science
of the Total Environment, 407, 3652–3667, https://doi.org/10.1016/j.scitotenv.2008.
10.043
Fermoso F. G., Bartacek J., Jansen S. and Lens P. N. L. (2009). Metal supplementation to
UASB bioreactors: From cell-metal interactions to full-scale application. The Science
of the Total Environment, 407, 3652–3667, https://doi.org/10.1016/j.scitotenv.2008.
10.043
Fermoso F. G., Hullebusch E. D., Van Guibaud G., Collins G., Svensson B. H., Vink J. P. M.,
Esposito G. and Frunzo L. (2015). Fate of Trace Metals in Anaerobic Digestion. Trace
metal microbiology. In: Biogas Science and Technology, A. Gübitz, A. Bauer,
G. Bochmann, A. Gronauer and S. Weiss (eds), Springer, Berlin, https://doi.org/10.
1007/978-3-319-21993-6
Finazzo C., Harmer J., Bauer C., Jaun B., Duin E. C., Mahlert F., Goenrich M., Thauer R. K., Van
Doorslaer S. and Schweiger A. (2003). Coenzyme B induced coordination of coenzyme M
via its thiol group to Ni(I) of F430 in active methyl-coenzyme M reductase. Journal of
the American Chemical Society, 125, 4988–4989, https://doi.org/10.1021/ja0344314
Flores-Alsina X., Solon K., Mbamba C. K., Tait S., Gernaey K. V., Jeppsson U. and Batstone
D. J. (2016). Modelling phosphorus (P), sulfur (S) and iron (Fe) interactions for dynamic
simulations of anaerobic digestion processes. Water Research, 95, 370–382, https://doi.
org/10.1016/j.watres.2016.03.012
Flynn S. L., Szymanowski J. E. and Fein J. B. (2014). Modeling bacterial metal toxicity using
a surface complexation approach. Chemical Geology, 374, 110–116.
Galí A., Benabdallah T., Astals S. and Mata-Alvarez J. (2009). Modified version of ADM1
model for agro-waste application. Bioresource Technology, 100, 2783–2790, https://
doi.org/10.1016/j.biortech.2008.12.052
Glass J. B. and Orphan V. J. (2012). Trace metal requirements for microbial enzymes
involved in the production and consumption of methane and nitrous oxide. Front
Microbiology, 3, 1–20, https://doi.org/10.3389/fmicb.2012.00061
Gonzalez-Gil G. and Kleerebezem R. (1999). Effects of Nickel and Cobalt on Kinetics of
Methanol Conversion by Methanogenic Sludge as Assessed by On-Line CH 4
Monitoring Appl. Environmental Biology, 65, 1789–1793.
Gonzalez-Gil G., Jansen S., Zandvoort M. H. and Van Leeuwen H. P. (2003). Effect of
yeast extract on speciation and bioavailability of nickel and cobalt in anaerobic
bioreactors. Biotechnology and Bioengineering, 82, 134–142, https://doi.org/10.
1002/bit.10551

by guest
Gu Y., Zhang Y. and Zhou X. (2015). Effect of Ca(OH)2pretreatment on extruded rice straw
anaerobic digestion. Bioresource Technology, 196, 116–122.
Gustavsson J. (2012). Cobalt and Nickel Bioavailability for Biogas Formation. PhD thesis,
University of Linkoping, Sweden.
Hanhoun M., Montastruc L., Azzaro-pantel C., Biscans B., Frèche M. and Pibouleau L.
(2011). Temperature impact assessment on struvite solubility product: A
thermodynamic modelling approach. Chemical Engineering Journal, 167, 50–58,
https://doi.org/10.1016/j.cej.2010.12.001
Hauduc H., Smith S., Szabo A., Murthy S. and Tak I. (2015). A dynamic physicochemical
model for chemical phosphorus removal. Water Research, 73, 157–170.
Henze M., Gujer W., Mino T., Matsuo T., Wentzel M. C., Marais G. V. R. and
Van Loosdrecht M. C. M. (1999). Activated Sludge Model No.2d, ASM2d. Water
Science & Technology, 39, 165–182, https://doi.org/10.1016/S0273-1223(98)00829-4
Hickey R. F., Vanderwielen J. and Switzenbaum M. S. (1989). The effect of heavy metals on
methane production and hydrogen and carbon monoxide levels during batch anaerobic
sludge digestion. Water Research, 23, 207–218, https://doi.org/10.1016/0043-1354
(89)90045-6
Hille R., Van Foster T., Storey A. and Duncan J. (2004). Heavy metal precipitation by
sulphide and bicarbonate: evaluating methods to predict anaerobic digester overflow
performance. In: Mine Water 2004–Proceedings Int. Mine Water Assoc. Symp,
A. P. Jarvis, B. A. Dudgeon and P. L. Younger (eds), 2, 141–150.
Huang C., Lai J., Sun X., Li J., Shen J., Han W. and Wang L. (2016). Enhancing anaerobic
digestion of waste activated sludge by the combined use of NaOH and Mg(OH)2:
Performance evaluation and mechanism study. Bioresource Technology, 220, 601–608.
Hudson R. J. M. (1998). Which aqueous species control the rates of trace metal uptake by
aquatic biota? Observations and predictions of non-equilibrium effects. The Science of
the Total Environment, 219, 95–115, https://doi.org/10.1016/S0048-9697(98)00230-7
Jansen S. (2004). Speciation and bioavailability of cobalt and nickel in anaerobic wastewater
treatment. PhD thesis, Leerstoelgroep Fysische Chemie en Kolloidkunde, Wageningen
Univeristy, The Netherlands.
Jeen S. W., Amos R. T. and Blowes D. W. (2012). Modeling gas formation and mineral
precipitation in a granular iron column. Environmental Science & Technology, 46
(12), 6742–6749.
Jiang B. (2006). The effect of trace elements on the metabolism of methanogenic consortia.
PhD thesis, Wageningen University, The Netherlands.
Jimenez J., Latrille E., Harmand J., Robles A., Ferrer J., Gaida D., Wolf C., Mairet F., Bernard
O., Alcaraz-Gonzalez V., Mendez-Acosta H., Zitomer D., Totzke D., Spanjers H., Jacobi
F., Guwy A., Dinsdale R., Premier G., Mazhegrane S., Ruiz-Filippi G., Seco A., Ribeiro
T., Pauss A. and Steyer J. P. (2015). Instrumentation and control of anaerobic digestion
processes: A review and some research challenges. Reviews Environmental Science
Biotechnology, 14, 615–648, https://doi.org/10.1007/s11157-015-9382-6
Johnson B. R. and Shang Y. (2006). Applications and limitations of ADM 1 in municipal
wastewater solids treatment. Water Science & Technology, 54, 77–82, https://doi.
org/10.2166/wst.2006.528
Kalyuzhnyi S. and Fedorovich V. (1997). Integrated mathematical model of UASB reactor
for competition between sulphate reduction and methanogenesis. Water Science &
Technology, 36, 201–208, https://doi.org/10.1016/S0273-1223(97)00524-6

by guest
Kalyuzhnyi S. V. and Fedorovich V. V. (1998). Mathematical modelling of competition

between sulphate reduction and methanogenesis in anaerobic reactors. Bioresource
Kalyuzhnyi S., Fedorovich V., Lens P. and Pol L. H. (1998). Mathematical modelling as a
tool to study population dynamics between sulfate reducing and methanogenic
bacteria. Biodegradation, 9, 187–199.
Kauder J., Boes N., Pasel C. and Herbell J. D. (2007). Combining models ADM1 and ASM2d
in a sequencing batch reactor simulation. Chemical Engineering Technology, 30,
1100–1112, https://doi.org/10.1002/ceat.200700045
Kleerebezem R. and van Loosdrecht M. C. M. (2006). Critical analysis of some concepts
proposed in ADM1. Water Science & Technology, 54, 51, https://doi.org/10.
2166/wst.2006.525
Knobel A. N. and Lewis A. E. (2002). A mathematical model of a high sulphate wastewater
anaerobic treatment system. Water Research, 36, 257–265, https://doi.org/10.
1016/S0043-1354(01)00209-3
Koster W. and Van Leeuwen H. P. (2004). Physicochemical kinetics and transport at the
biointerface: Setting the stage. In: Iupac Series on Analytical and Physical Chemistry
of Environmental Systems, 9, 1–14.
Kong X., Wei Y., Xu S., Liu J., Li H., Liu Y. and Yu S. (2016). Inhibiting excessive
acidification using zero-valent iron in anaerobic digestion of food waste at high
organic load rates. Bioresource Technology, 211, 65–71, https://doi.org/10.1016/j.
biortech.2016.03.078
Konhauser K. (2007). Introduction to geomicrobiology. European Journal of Soil Science, 58
(5), 1215–1216, https://doi.org/10.1111/j.1365-2389.2007.00943_4.x
Koutsoukos P., Amjad Z., Tomson M. B. and Nancollas G. H. (1980). Crystallization of
calcium phosphates. A constant composition study. Journal of the American
Chemical Society, 102(5), 1553–1557.
Lawrence A. W. and McCarty P. L. (1965). The Role of Sulfide in Preventing Heavy Metal
Toxicity in Anaerobic Treatment. Journal Water Pollution Control Federation, 37,
392–406.
Lauwers J., Appels L., Thompson I. P., Degrève J., Impe J. F. and Van Dewil R. (2013).
Mathematical modelling of anaerobic digestion of biomass and waste: Power and
limitations. Progress in Energy and Combustion Science, 39, 383–402, https://doi.
org/10.1016/j.pecs.2013.03.003
Lin C.-Y. (1992). Effect of heavy metals on volatile fatty acid degradation in anaerobic
digestion. Water Research, 26, 177–183, https://doi.org/10.1016/0043-1354(92)
90217-r
Lizarralde I., Fernández-Arévalo T., Brouckaert C., Vanrolleghem P., Ikumi D. S., Ekama
G. A., Ayesa E. and Grau P. (2015). A new general methodology for incorporating
physico-chemical transformations into multi-phase wastewater treatment process
models. Water Research, 74, 239–256, https://doi.org/10.1016/j.watres.2015.01.031
Loaec M., Olier R. and Guezennec J. (1997). uptake of lead, cadmium and zinc by bacterial
exopolysaccharide. Water Research, 31, 1171–1997.
Lopez-Vazquez C. M., Mithaiwala M., Moussa M. S., van Loosdrecht M. C. M. and
Brdjanovic D. (2013). Coupling ASM3 and ADM1 for wastewater treatment process
optimisation and biogas production in a developing country: Case-study Surat, India.

by guest
Journal Water, Sanitation Hygiene Development, 3, 12, https://doi.org/10.

2166/washdev.2013.017
MacConnell C. B. and Collins H. P. (2009). Utilization of re-processed anaerobically digested
fiber from dairy manure as a container media substrate. Acta Horticulturae, 819,
279–286.
Maharaj B. C., Mattei M. R., Frunzo L., van Hullebusch E. D. and Esposito G. (2018). ADM1
digestion processes. Bioresource Technology, 267, 666–676, https://doi.org/10.
1016/j.biortech.2018.06.099
Malik A. (2004). Metal bioremediation through growing cells. Environment International,
30, 261–278, https://doi.org/10.1016/j.envint.2003.08.001
Maurer M. and Boller M. (1999). Modelling of phosphorus precipitation in wastewater
treatment plants with enhanced biological phosphorus removal. Water Science &
Maurer M., Abramovich D., Siegrist H. and Gujer W. (1999). Kinetics of biologically induced
phosphorus precipitation in waste-water treatment. Water Research, 33, 484–493,
https://doi.org/10.1016/S0043-1354(98)00221-8
Mazumder T. K., Nishio N., Fukuzaki S. and Nagai S. (1987). Production of extracellular
vitamin B-12 compounds from methanol by Methanosarcina barkeri. Applied
Microbiology and Biotechnology, 26, 511–516, https://doi.org/10.1007/BF00253023
Mbamba C. K., Batstone D. J., Flores-Alsina X. and Tait S. (2015a). A generalised chemical
precipitation modelling approach in wastewater treatment applied to calcite. Water
Research, 68, 342–353.
Mbamba C. K., Tait S., Flores-Alsina X. and Batstone D. J. (2015b). A systematic study of
multiple minerals precipitation modelling in wastewater treatment. Water Research, 85,
359–370.
Morel F. and Morgan J. (1972). Numerical method for computing equilibriums in aqueous
chemical systems. Environmental Science & Technology, 6(1), 58–67.
Morse J. W. and Arakaki T. (1993). Adsorption and coprecipitation of divalent metals with
mackinawite (FeS). Geochimica et Cosmochimica Acta, 57(15), 3635–3640.
Mu H., Chen Y. and Xiao N. (2011). Effects of metal oxide nanoparticles (TiO2, Al2O3,
SiO2and ZnO) on waste activated sludge anaerobic digestion. Bioresource
Technology, 102, 10305–10311.
Mudhoo A. and Kumar S. (2013). Effects of heavy metals as stress factors on anaerobic
digestion processes and biogas production from biomass. International Journal
Environmental Science Technology, 10, 1383–1398, https://doi.org/10.
1007/s13762-012-0167-y
Musvoto E. V., Wentzel M. C., Loewenthal R. E. and Ekama G. A. (2000a). Integrated
chemical-physical processes modelling – I. Development of a kinetic-based model for
mixed weak acid/base systems. Water Research, 34, 1857–1867, https://doi.org/10.
1016/S0043-1354(99)00334-6
Musvoto E. V, Wentzel M. C. M. and Ekama G. a. M. (2000b). Integrated Chemical Physical
Processes Modelling -II. Simulating Aeration Treatment of Anaerobic Digester
Supernatants. Water Research, 34, 1868–1880.
Nopens I., Batstone D. J., Copp J. B., Jeppsson U., Volcke E., Alex J. and Vanrolleghem P. a.
(2009). An ASM/ADM model interface for dynamic plant-wide simulation. Water
Research, 43, 1913–1923, https://doi.org/10.1016/j.watres.2009.01.012

by guest
Ohlinger K. N., Young T. M. and Schroeder E. D. (1998). Predicting struvite formation in

digestion. Water Research, 32, 3607–3614, https://doi.org/10.1016/S0043-1354(98)
00123-7
Oleszkiewicz J. A. and Sharma V. K. (1990). Stimulation and inhibition of anaerobic
processes by heavy metals-A review. Biology Wastes, 31, 45–67, https://doi.org/10.
1016/0269-7483(90)90043-R
Oude Elferink S. J. W. H., Visser A., Hulshoff Pol L. W. and Stams A. J. M. (1994). Sulfate
reduction in methanogenic bioreactors. FEMS Microbiology Reviews, 15, 119–136,
https://doi.org/10.1016/0168-6445(94)90108-2
Papirio S., Frunzo L., Mattei M. R. and Ferraro A. (2017). Sustainable Heavy Metal
Remediation, Springer, Berlin, https://doi.org/10.1007/978-3-319-61146-4
Parker W. J. and Wu G.-H. (2006). Modifying ADM1 to include formation and emission of
odourants. Water Science & Technology, 54, 111–117, https://doi.org/10.2166/wst.
2006.532
Paulo L. M., Stams A. J. M. and Sousa D. Z. (2015). Methanogens, sulphate and heavy
metals: A complex system. Reviwes Environmetal Science Biology, 14, 537–553.
Peiris B. R. H., Rathnasiri P. G., Johansen J. E., Kuhn A. and Bakke R. (2006). ADM1
simulations of hydrogen production. Water Science & Technology, 53, 129–137,
https://doi.org/10.2166/wst.2006.243
Peng S. and Colosi L. M. (2016). Anaerobic digestion of algae biomass to produce energy
during wastewater treatment. Water Environment Research, 88(1), 29–39.
Poinapen J. and Ekama G. A. (2010). Biological sulphate reduction with primary sewage
sludge in an upflow anaerobic sludge bed reactor - Part 6: Development of a kinetic
model for BSR. Water Sa, 36(3), 203–214.
Preis W. and Gamsjäger H. (2001). Thermodynamic investigation of phase equilibria in metal
carbonate–water–carbon dioxide systems. Monatshefte fuer Chemie/Chemical Monthly,
132(11), 1327–1346.
Qiao L. and Ho G. (1996). The effect of clay amendment on speciation of heavy metals in
sewage sludge. Water Science and Technology, 34(7-8), 413–420.
Rahaman M. S., Mavinic D. S., Meikleham A. and Ellis N. (2014). Modelling phosphorus
removal and recovery from anaerobic digester supernatant through struvite crystallization
in a fluidized bed reactor. Water Research, 51, 1–10, https://doi.org/10.1016/j.watres.
2013.11.048
Reichert P. (1994). AQUASIM – A tool for simulation and data analysis of aquatic systems.
Water Science & Technology, 30(2), 21–30.
Ristow N. E., Whittington-Jones K., Corbett C., Rose P. and Hansford G. S. (2002).
Modelling of a recycling sludge bed reactor using AQUASIM (vol 27, pg 445, 2001).
Water Sa, 28, 111–120.
Rittmann B. E., Banaszak J. E., VanBriesen J. M. and Reed D. T. (2002). Mathematical
modelling of precipitation and dissolution reactions in microbiological systems.
Biodegradation, 13, 239–250, https://doi.org/10.1023/A:1021225321263
Romero-güiza M. S., Vila J., Mata-alvarez J., Chimenos J. M. and Astals S. (2016). The role
of additives on anaerobic digestion: A review. Renewable and Sustainable Energy
Reviews, 58, 1486–1499, https://doi.org/10.1016/j.rser.2015.12.094

by guest
Rosen C., Vrecko D., Gernaey K. V., Pons M. N. and Jeppsson U. (2006). Implementing
ADM1 for plant-wide benchmark simulations in Matlab/Simulink. Water Science &
Technology, 54, 11, https://doi.org/10.2166/wst.2006.521
Roussel J. (2012). Metal behaviour in anaerobic sludge digesters supplemented with trace
nutrients. PhD Thesis, Univ. Birmingham, United Kingdom.
Roussel J. and Carliell-Marquet C. (2016). Significance of vivianite precipitation on the
mobility of iron in anaerobically digested sludge. Frontiers in Environmental Science,
4, 60.
Schwarz A. O. and Rittmann B. E. (2007). A biogeochemical framework for metal
detoxification in sulfidic systems. Biodegradation, 18, 675–692, https://doi.org/10.
1007/s10532-007-9101-2
Scott W. D., Wrigley T. J., Webb K. M. and Science E. (1991). A computer model of struvite
solution chemistry. Talanta, 38, 889–895.
Shakeri S., Gustavsson J., Svensson B. H. and Skyllberg U. (2012). Sulfur K-edge XANES
and acid volatile sulfide analyses of changes in chemical speciation of S and Fe during
sequential extraction of trace metals in anoxic sludge from biogas reactors. Talanta, 89,
470–477, https://doi.org/10.1016/j.talanta.2011.12.065
Shaojun Z. and Mucci A. (1993). Calcite precipitation in seawater using a constant addition
technique: A new overall reaction kinetic expression. Geochimica et Cosmochimica
Acta, 57(7), 1409–1417.
Smith S., Takács I., Murthy S., Daigger G. T. and Szabó A. (2008). Phosphate Complexation
Model and Its Implications for Chemical Phosphorus Removal. Water Environment
Research, 80, 428–438, https://doi.org/10.2175/106143008X268443
Solon K. (2015). IWA Anaerobic Digestion Model No. 1 extended with Phosphorus
and Sulfur Literature Review, Division of Industrial Electrical Engineering and
Automation Faculty of Engineering, Lund University, 1–16.
Solon K., Flores-Alsina X., Mbamba C. K., Volcke E. I. P., Tait S., Batstone D., Gernaey
K. V. and Jeppsson U. (2015). Effects of ionic strength and ion pairing on
(plant-wide) modelling of anaerobic digestion. Water Research, 70, 235–245, https://
doi.org/10.1016/j.watres.2014.11.035
Stumm W. and Morgan J. J. (1996). Aquatic Chemistry, 3rd ed. John Wiley and Sons, Inc,
New York.
Stupperich E. and Kräutler B. (1988). Pseudo vitamin B12 or 5-hydroxybenzimidazolyl-
cobamide are the corrinoids found in methanogenic bacteria. Archives of
Microbiology, 149, 268–271, https://doi.org/10.1007/BF00422016
Sunda W. G. and Huntsman S. A. (1998). Interactions among Cu2+, Zn2+, and Mn2+ in
controlling cellular Mn, Zn, and growth rate in the coastal alga Chlamydomonas.
Limnol. Oceanogr, 43, 1055–1064, https://doi.org/10.4319/lo.1998.43.6.1055
Szabó A., Takács I., Murthy S., Daigger G. T., Licskó I. and Smith S. (2008). Significance of
Design and Operational Variables in Chemical Phosphorus Removal. Water
Environment Research, 80, 407–416, https://doi.org/10.2175/106143008X268498
Tait S., Clarke W. P., Keller J. and Batstone D. J. (2009). Removal of sulfate from
high-strength wastewater by crystallisation. Water Research, 43, 762–772, https://
doi.org/10.1016/j.watres.2008.11.008
Thanh P., Ketheesan B., Zhou Y. and Stuckey D. C. (2015). Trace metal speciation and
bioavailability in anaerobic digestion: A review. Biotechnology Advances, 34,
122–136, https://doi.org/10.1016/j.biotechadv.2015.12.006

by guest
Thauer R.K. (1998). Biochemistry of methanogenesis: A tribute to marjory stephenson: 1998

marjory stephenson prize lecture. Microbiology, 144(9), 2377–2406.
Uemura S. (2010). Mineral Requirements for Mesophilic and Thermophilic Anaerobic
Digestion of Organic Solid Waste. International Journal of Environmental Research,
4, 33–40.
Uludag-Demirer S. and Othman M. (2009). Removal of ammonium and phosphate from the
supernatant of anaerobically digested waste activated sludge by chemical precipitation.
Bioresource technology, 100(13), 3236–3244.
Van Briesen J. M. and Rittmann B. E. (1999). Modeling speciation effects on biodegradation
in mixed metal/chelate systems. Biodegradation, 10(5), 315–330.
van Hullebusch E. D., Guibaud G., Simon S., Lenz M., Shakeri Yekta S., Fermoso F. G., Jain
R., Duester L., Roussel J., Guillon E., Skyllberg U., Almeida C. M. R., Pechaud Y.,
Garuti M., Frunzo L., Esposito G., Carliell-Marquet C., Ortner M. and Collins G.
(2016). Methodological approaches for fractionation and speciation to estimate trace
element bioavailability in engineered anaerobic digestion ecosystems: An overview.
Critical. Reviews Environmental Science and Technology, 46, 1324–1366, https://
doi.org/10.1080/10643389.2016.1235943
Van Langerak E. P. A., Beekmans M. M. H., Beun J. J., Hamelers H. V. M. and Lettinga G.
(1999). Influence of phosphate and iron on the extent of calcium carbonate precipitation
during anaerobic digestion. Journal of Chemical Technology & Biotechnology:
International Research in Process, Environmental & Clean Technology, 74(11),
1030–1036.
Van Rensburg P., Musvoto E. V., Wentzel M. C. and Ekama G. A. (2003). Modelling
multiple mineral precipitation in anaerobic digester liquor. Water Research, 37,
3087–3097, https://doi.org/10.1016/S0043-1354(03)00173-8
Vavilin V. A., Vasiliev V. B., Rytov S. V. and Ponomarev A. V. (1995). Modelling ammonia
and hydrogen sulfide inhibition in anaerobic digestion. Water Research, 29, 827–835,
https://doi.org/10.1016/0043-1354(94)00200-Q
Wang R., Li Y., Chen W., Zou J. and Chen Y. (2016). Phosphate release involving PAOs
activity during anaerobic fermentation of EBPR sludge and the extension of ADM1.
Chemical Engineering Journal, 287, 436–447, https://doi.org/10.1016/j.cej.2015.10.
110
Warren L. A. and Haack E. A. (2001). Biogeochemical controls on metal behaviour in
freshwater environments. Earth-Science Review, 54, 261–320, https://doi.org/10.
1016/S0012-8252(01)00032-0
Willet A. I. and Rittmann B. E. (2003). Slow complexation kinetics for ferric iron and EDTA
complexes make EDTA non-biodegradable. Biodegradation, 14, 105–121, https://doi.
org/10.1023/A:1024074329955
Worms I., Simon D. F., Hassler C. S. and Wilkinson K. J. (2006). Bioavailability of trace
metals to aquatic microorganisms: Importance of chemical, biological and physical
processes on biouptake. Biochimie, 88, 1721–1731, https://doi.org/10.1016/j.biochi.
2006.09.008
Wrigley T. J., Scoit W. D. and Webb K. M. (1992). An improved computer model of struvite.
Talanta, 39, 1597–1603.

by guest
Xu F., Li Y. and Wang Z. W. (2015). Mathematical modelling of solid-state anaerobic

digestion. Progress in Energy and Combustion Science, 51, 49–66, https://doi.
org/10.1016/j.pecs.2015.09.001
Ye Z. L., Chen S. H., Wang S. M., Lin L. F., Yan Y. J., Zhang Z. J. and Chen J. S. (2010).
Phosphorus recovery from synthetic swine wastewater by chemical precipitation using
response surface methodology. Journal of Hazardous Materials, 176(1-3), 1083–1088.
Yekta S. S., Lindmark A., Skyllberg U., Danielsson Å. and Svensson B. H. (2014a).
Importance of reduced sulfur for the equilibrium chemistry and kinetics of Fe(II), Co
(II) and Ni(II) supplemented to semi-continuous stirred tank biogas reactors fed with
stillage. Journal of Hazardous Materials, 269, 83–88, https://doi.org/10.1016/j.
jhazmat.2014.01.051
Yekta S. S., Skyllberg U., Danielsson Å., Björn A. and Svensson B. H. (2016). Chemical
speciation of sulfur and metals in biogas reactors—Implications for cobalt and nickel
bio-uptake processes. Journal of Hazardous Materials, 324, 110–116, https://doi.
org/10.1016/j.jhazmat.2015.12.058
Yekta S. S., Svensson B. H., Björn A. and Skyllberg U. (2014b). Thermodynamic modelling
of iron and trace metal solubility and speciation under sulfidic and ferruginous
conditions in full scale continuous stirred tank biogas reactors. Applied Geochemistry,
47, 61–73, https://doi.org/10.1016/j.apgeochem.2014.05.001
Zaher U. and Chen S. (2006). Interfacing the IWA Anaerobic Digestion Model No.1 (ADM1)
with Manure and Solid Waste Characteristics. Proceeding Water Environmental
Federation, 2006(9), 3162–3175, https://doi.org/10.2175/193864706783751726
Zaher U., Grau P., Benedetti L., Ayesa E. and Vanrolleghem P. (2007). Transformers for
interfacing anaerobic digestion models to pre- and post-treatment processes in a
plant-wide modelling context. Environmental Modelling Software, 22, 40–58, https://
doi.org/10.1016/j.envsoft.2005.11.002
Zandvoort B. M. H., Hullebusch E. D. Van, Fermoso F. G. and Lens P. N. L. (2006). Trace
Metals in Anaerobic Granular Sludge Reactors: Bioavailability and Dosing Strategies.
Engineering Life Science, 6, 293–301, https://doi.org/10.1002/elsc.200620129
Zhang T., Bowers K. E., Harrison J. H. and Chen S. (2010). Releasing phosphorus from
calcium for struvite fertilizer production from anaerobically digested dairy effluent.
Water Environment Research, 82(1), 34–42.
Zhang W., Zhang L. and Li A. (2015). Enhanced anaerobic digestion of food waste by trace
metal elements supplementation and reduced metals dosage by green chelating agent [S,
S]-EDDS via improving metals bioavailability. Water Research, 84, 266–277, https://
doi.org/10.1016/j.watres.2015.07.010
Zhang Y., Piccard S. and Zhou W. (2015). Improved ADM1 model for anaerobic digestion
process considering physico-chemical reactions. Bioresource Technology, 196,
279–289, https://doi.org/10.1016/j.biortech.2015.07.065
Zhao W. E. and Yang H. (2014). Statistical Methods in Drug Combination Studies, CRC
Press, Boca Raton, FL.

by guest
APPENDIX
Notation
KA/B,H2S Acid-base kinetic coefficient (m3 kmol−1 d−1)
KH,H2S Henry’s law coefficient (kmol−3 bar−1)
KI,H2S Inhibition coefficient by undissociated H2S (kmol m−3)
K100 Concentration of H2S or pH at which the uptake rate is decreased 100
times (kmol m−3)
K2 Concentration of H2S or pH at which the uptake rate is decreased 2 times
(kmol m−3)
Ka,H2S Acid-base kinetic coefficient (m3 kmol−1 )
kdec,i First order decay rate of species i (d−1)
k la Gas-liquid mass-transfer coefficient (d−1)
km,i Monod maximum specific uptake rate (d−1)
kp2 Competitive product inhibition coefficient (kg m−3)
KS,j Half saturation value (kg COD_Sm−3)
KSSO4 Half saturation value for sulfate (kg SO4 m−3)
Pgas,H2S Hydrogen sulfide partial pressure (bar)
Sj Concentration of soluble organic components (kg COD m−3)
Xi Concentration of particulate component i (kg COD m−3)
IpH pH inhibition function
IH 2 s Sulfide inhibition function
rA/B Acid-base kinetic rate (kmol m−3 d−1)
rgas Liquid-gas kinetic gas transfer rate (kmol m−3 d−1)
rdecay,i Kinetic rate of bacterial decay (kg COD m−3 d−1)
rgrowth,i Kinetic rate of microbial growth (kg COD m−3 d−1)

by guest
Bioresource Technology 267 (2018) 666–676
Contents lists available at ScienceDirect
Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech
ADM1 based mathematical model of trace element precipitation/dissolution T

in anaerobic digestion processes
⁎
Bikash Chandra Maharaja,b, , Maria Rosaria Matteib, Luigi Frunzob, Eric D. van Hullebuschc,d,
Giovanni Espositoa
a
University of Cassino and the Southern Lazio, Department of Civil and Mechanical Engineering, Cassino, Italy
b
University of Naples Federico II, Department of Mathematics and Applications “Renato Caccioppoli”, Naples, Italy
c
Université Paris-Est, Laboratoire Géomatériaux et Environnement (LGE), EA 4508, UPEM, 77454 Marne-la-Vallée, France
d
IHE Delft Institute for Water Education, Department of Environmental Engineering and Water Technology, The Netherlands
A R T I C LE I N FO A B S T R A C T
Keywords: Due to the complex biogeochemistry of trace elements (TEs, e.g. Fe, Ni and Co) in anaerobic digestion processes,
Anaerobic digestion their role and fate is poorly understood. Challenging, time consuming and low detection limits of analytical
Trace elements procedures necessitate recruitment of mathematical models. A dynamic mathematical model based on anaerobic
Precipitation digestion model no.1 (ADM1) has been proposed to simulate the effect of TEs. New chemical equilibrium as-
Mathematical modeling
sociation/dissociation and precipitation/dissolution reactions have been implemented to determine TE bioa-
ADM1
vailability and their effect on anaerobic digestion. The model considers interactions with inorganic carbonate
(HCO3− and CO32−), phosphate (PO43−, HPO42−, H2PO4−) and sulfide (HS− and S2−). The effect of deficiency,
activation, inhibition and toxicity of TEs on the biochemical processes has been modelled based on a dose-
response type inhibition function. The new model can predict: the dynamics of TEs (among carbonate, sulfide
and phosphate); the starvation of TEs; and the effect of initial sulfur-phosphorus ratio in an in-silico batch
anaerobic system.
1. Introduction the solid phase of AD. Such chemical extraction methods (standalone or
combination of BCR sequential extraction, Tessier sequential extrac-
Anaerobic digestion (AD) is a well established multistep process, in tion, Simultaneously Extracted Metals/Acid-Volatile Sulfide (SEM-
which a consortium of microorganisms, in absence of oxygen, degrades AVS), S K-edge X-ray absorption spectroscopy) are easier to implement.
complex organic matter to produce biogas (CH4, CO2, H2 and H2S) and However, maintenance of anoxic condition while sampling, failure to
digestate (Dai et al., 2017). Although AD is a robust technology, the determine TE bioavailable concentration and unavailability of routine
multiplicity of operational parameters makes intricate to implement an method are major disadvantages (van Hullebusch et al., 2016). The TEs
optimized, controlled and stable process (Jimenez et al., 2015), parti- fate in AD bioreactors is influenced by the chemical reactions leading to
cularly in the case of AD of municipal solid waste (Esposito et al., 2011; the formation of precipitates with S and P compounds as well as com-
Xu et al., 2015). During the last years, there has been an increased plexes with: 1) organic substrates of AD (VFA and thiol containing
attention towards developing optimized and stable processes in AD. To amino acids); 2) humic substances; and if available, 3) synthetic che-
a major extent, the role of inorganic components (trace elements (TEs), lating agents (EDTA or NTA). TEs also adsorb on the particulate com-
sulfur (S) and phosphorus (P)) and abiotic processes (precipitation, ponents constituting the anaerobic system. Fe is a structural component
complexation and surface complexation) have received considerable of the majority of enzymes in AD pathways. Also, it serves as a redox
attention across the research community (Glass and Orphan, 2012; carrier. Ni is a part of coenzyme F430 ring structure of methyl reductase
Ketheesan et al., 2016; Thanh et al., 2015, 2017a). Recently, attempts which facilitates reductions of methyl co-enzyme M to methane during
have been made to predict bioavailability of TEs in anaerobic digestion methanogenesis. Co is essential to both acetoclastic methanogens and
system (Cai et al., 2018; Shakeri et al., 2012; Thanh et al., 2017b; Zhang hydrogenotrophic methanogens. It is an integral part of coronoid
et al. 2015a). Fe, Ni, and Co are in the center of these studies. Most of structure of vitamin B12 which binds to coenzyme methylase. These
them recruit chemical extraction methods to predict bioavailability in findings have been illustrated in various experimental studies (Glass
⁎
Corresponding author at: Department of Civil and Mechanical Engineering, University of Cassino and Southern Lazio, Via di Basio 43, 03043 Cassino (FR), Italy.
E-mail address: [email protected] (B.C. Maharaj).
Received 22 April 2018; Received in revised form 24 June 2018; Accepted 28 June 2018
Available online 03 July 2018
0960-8524/ © 2018 Elsevier Ltd. All rights reserved.
B.C. Maharaj et al. Bioresource Technology 267 (2018) 666–676
and Orphan, 2012), which have confirmed the fragility imparted by TEs waste water treatment process models (Lizarralde et al., 2015). Al-
to the methanogenic metabolism and thus to the stability and efficiency though some studies included Fe in the physicochemical model de-
of the whole AD process. scription, there has been a gap in modelling the effect of TE on mi-
Due to the significant role played by TEs in AD systems, the wide crobial kinetics, both in terms of inhibition and stimulation as well as
range of interactions affecting TE bioavailability, the experimental and the effect of multimineral precipitation on TE bioavailability. There-
operational shortcomings, developing a mathematical model able to fore, a dynamic mathematical model based on the anaerobic digestion
consider the complex dynamics of TEs in anaerobic environments is model no.1 (ADM1) approach has been proposed to simulate the TEs
relevant for both academia and industries. Such a model may support precipitation and its effect on anaerobic digestion in terms of total
experimental studies by individuating important process parameters methane production. The main objective of this work is to develop an
and conditions which affect TE bioavailability. On a specific note, ADM1 based full kinetic mathematical model to: 1) mechanistically
chemical precipitation represents a significant process in defining the describe all the main biochemical and physicochemical processes re-
TE bioavailability, due to the strong interactions with other physico- quired for precipitation/dissolution and uptake of TE; 2) implement a
chemical reactions in AD such as the weak acid base system and ion TE dose-response effect on methane production; 3) investigate into
complexation (Musvoto et al., 2000b; Solon et al., 2015; van speciation of TE, in particular the fractionation of TE among sulfide,
Hullebusch et al., 2016). Anaerobic environments are typically char- carbonate, phosphate and dissolved species and its relation to pH and
acterized by the presence of abundant inorganic ligands, such as car- TE dose. To this aim, new chemical equilibrium dissociation, ion
bonate, phosphate and sulfide, which are assumed to control the che- pairing and precipitation reactions have been defined and added to
mical speciation of TEs by the formation of poorly soluble precipitates ADM1. Further, hydrolysis, acidogenesis, acetogenesis, methanogenesis
(Callander and Barford, 1983; Yekta et al., 2016). Carbonate, phosphate and acid-base equilibrium have been combined to TE precipitation/
and sulfide interact also with the light metals usually contained in the dissolution and the effects on total methane production have been
digestion substrates (e.g. calcium (Ca), magnesium (Mg), potassium evaluated.
(K)), leading to the formation of significant quantities of struvite, hy-
droxyapatite and calcite. Based on extensive experimental activity 2. Model development and simulation
(Barrera et al., 2015), sulfide concentration has been recognized as a
primary factor in the definition of TEs bioavailability as it regulates the Before proceeding further, it is important to introduce few defini-
formation of metal sulfides, which are characterized by a very low so- tions and the terminology used for the discussion of the numerical re-
lubility product (Thanh et al., 2015). A model able to take into account sults. These are as follows:
chemical precipitation therefore automatically requires representation
of other physicochemical reactions, such as the equilibria character- 1) Component: The basic entities of the reactions that take place in the
izing specific ionic forms of inorganic C, S and P compounds as well as system.
additional gas transfer processes. 2) Species: Two or more components combine to form a species. So,
The benchmark model ADM1 developed by the IWA Task Group on components can combine to give species but the combination of two
Anaerobic Digestion (Batstone et al., 2002), represents the currently species does not give rise to any other species. Chemical species
most comprehensive framework of the AD process as it has been suc- considered as an outcome of precipitation reactions are assumed to
cessfully implemented to study the interactions between various op- be solid.
erational parameters, to understand the underlying mechanistic prin- 3) Chemical system: It consists of all the components that interact with
ciples, to estimate system behaviour in varied scenarios, to overall each other through precipitation and acid/base reactions. For ex-
improve the process performance in terms of a desired output and to ample, sulfur chemical system consists of S2−, HS−, H2S, Fe2+,
recruit a model predictive control system based on reduced version of Ni2+ and Co2+.
the model. Although the ADM1 predicts the dynamic concentration of 4) Percentage fraction: It is the amount of a particular species formed
biochemical components fairly well, its predictions are limited due to compared to the total species for the same component. It may have
insufficient consideration of physicochemical processes (Batstone et al., any values between 0 and 100. For example, the fraction of FeCO3 is
2012; Ekama et al., 2006) (e.g. ADM1 ignored TEs dynamics including the amount of FeCO3 formed in the system at a given condition.
the liquid-solid precipitation reactions). The major reason of side lining
such processes is attributed to the wide range of components to be 2.1. Basic model structure
taken into account (Zhang et al. 2015a). However, over the last years,
there have been attempts to implement additional physicochemical A full structured kinetic framework based on ADM1 (Batstone et al.,
processes in AD and waste water treatment process models. ADM1 2002) has been used to model the effect of TEs precipitation on AD
based models for sulfur rich waste streams have been defined and va- performance. The model includes 4 principal types of processes with a
lidated (Barrera et al., 2015; Federovich et al., 2003). Sulfate reduction varying time range: protonation/deprotonation being the fastest pro-
processes have been additionally implemented, thereby increasing the cesses, biochemical processes being the slowest, precipitation/dissolu-
complexity and also the number of variables of the model. Non-ideal- tion and gas transfer with time range intermediate to the above pro-
ities arising due to ionic strength of the system has been incorporated in cesses. Acid base reactions in the system have been implemented in the
ADM1 and the necessity to include physicochemical processes in stan- form of protonation/deprotonation reactions with varying pKa values
dard biochemical frameworks has been concluded (Solon et al., 2015). for the various acid base pairs considered. Protonation/deprotonation
In a major extension of ADM1, interactions of phosphorus, sulfur and are implemented as differential equations to avoid model stiffness
iron have been taken into consideration to model and validate multiple arising from the consideration of a mixed differential algebraic ap-
mineral precipitation reactions (Flores-Alsina et al., 2016). In this proach. Kinetic expression for protonation/deprotonation reactions are
study, multiple mineral precipitation has been estimated by saturation based on the standard expression described in ADM1. Further, pH has
index calculation which is based on ion activity correction (Mbamba been calculated on the basis of the charge balance generating from the
et al., 2015). New precipitation and gas transfer processes have been acidic and basic components considered. Precipitation and dissolution
linked to biochemical framework of ADM1 to investigate the effects of reactions for inorganic components have been considered. Precipitation
soluble Ca, Mg, inorganic phosphorus and nitrogen on anaerobic di- occurs at supersaturation of the reacting species and involves the
gestion (Zhang et al., 2015a). Ion association-dissociation, precipita- transfer/loss of mass from liquid phase to the solid phase which is based
tion-dissolution and liquid gas transfers also have been described to on the kinetic model described by Koutsoukos et al. (1980). This mass
establish a framework for incorporating physicochemical processes in loss from liquid phase reverses when the system is undersaturated to
667
2.2. Biochemical framework
AD is the biological conversion of organic substrate into biogas

through a series of biochemical reactions. The first step is disintegration
of composite particulate matter into carbohydrates, proteins and lipids.
The subsequent step involves the hydrolysis of carbohydrates, proteins
and lipids into monosaccharide, amino acids and LCFA. Thereupon,
these soluble compounds are converted into VFAs in the acidogenesis
step and further into acetic acid and H2 in the acetogenesis step. Finally,
methane is formed from acetic acid and H2 during the methanogenesis
stage.
Since TEs have been reported as micronutrients in AD systems and
are constituents of co-factors in enzyme systems, their effect on bio-
chemical processes has been explicitly taken into account by introdu-
Fig. 1. Schematic representation of the main physico-chemical processes im- cing an additional term (I Me2+) in the inhibition expressions as follows:
plemented in the proposed model.
I1 = IpH ·IIN _lim·IMe2 +, (3)
accommodate the changing equilibrium. Apart from the standard gas I2 = IpH ·IIN _lim·IH2·IMe2 +, (4)
stripping processes considered in ADM1, additionally H2S stripping has
been considered. Gas and liquid phase reach an equilibrium point with I3 = IpH ·IIN _lim·IH2·INH3·IMe2 +, (5)
respect to each other which has been represented by the Henry’s law. In where I1 ⋯ n = f (SI ,1 ⋯ n ) are the established inhibition functions used in
the biochemical module, a consortium of microorganisms degrades and ADM1 framework (Batstone et al., 2002) which include an additional
assimilates organic waste in the five distinct steps of disintegration, non-competitive biostatic inhibition function, I Me2+. The latter keeps in
hydrolysis, acidogenesis, acetogenesis and methanogenesis to produce track the effect of TEs on the biochemical rates and acts as a growth
methane, carbon dioxide and hydrogen. In specific, phosphorus, sulfur limiting factor. I Me2+ is expressed as:
and TEs have been introduced as new products of disintegration step.
The microbial uptake of TEs and the TEs inhibition on microbial ac- a1·Me 2 + + a2
IMe2 + = ,
tivities have been defined as well. (Me )2 + b1·Me 2 +
2 + + b2 (6)
Biochemical processes have been considered as irreversible pro- where a1, a2 , b1, b2 are assumed constants which can been adjusted to
cesses while protonation/deprotonation, precipitation/dissolution and obtain a desirable optimum dose-response function at a particular TE
gas stripping are reversible processes, as shown by appropriate arrow concentration, and Me 2+ denotes the concentration of bioavailable TEs
marks in Fig. 1. species within the bulk liquid. The function takes into account both the
The developed model tracks the dynamics of the components both stimulation and inhibiting action of TEs on biochemical rate depending
in the liquid and gas phase. For each component considered in the li- on the concentration range.
quid phase (i.e. microbial species, dissolved compounds, precipitates, Complex organic matter has been also considered as a source for
inert), the mass balance can be written as: phosphorus, sulfur and TEs which arise in the system from the disin-
tegration step. Therefore, phosphorus has been supposed to be released
dVliq Sliq, i m1 m2
dt
= qin Sin, i−qout Sliq, in + Vliq (∑ ρ α
j=1 bio, j i, j + ∑ j=1 ρprec,j βi,j as HPO42−, which represents the most abundant form in the pH range
of 6–14. Similarly, the released sulfur has been considered in the form
m2
+ ∑ j=1 ρdissol,j γi,j−ρT ,i ), i = 1, …, n , 1
(1)
of HS- which has been depicted as the abundant form in the pH range of
6–12. Furthermore, uptake of TEs has been linked to the acetate and
hydrogen uptake. The stoichiometry has been updated accordingly.
where Vliq is the reactor volume, Sliq, i is the liquid phase state variable,
qin and qout are the inlet and outlet flow rates, ρbio, j αi, j , ρprec, j βi, j , ρdissol, j γi, j
and ρT , i , denote the process rate for biochemical transformations, pre- 2.3. Physicochemical framework
cipitation, dissolution and gas transfer for component i, respectively, n1
is the number of liquid phase components considered in the model, m1 is 2.3.1. Liquid-gas processes
the number of the implemented biochemical transformations (including The gas-transfer for H2, CH4, CO2 and H2S has been taken into ac-
the acid-base reactions) and m2 is the number of precipitation/dis- count. The amount of mass transferred from liquid phase to gas phase
solution reactions considered. The mass balance equations for the has been predicted by considering the equilibrium between liquid phase
components in the gas phase, needed to predict the energy production and gas phase. The equations of liquid to gas mass transfer have been
potential of the system and accounting for the liquid to gas mass included in the proposed model using the following rate equation as
transfer, have been included by following the approach proposed in the prescribed in ADM1.
ADM1. The general form of the equations can be written as follows: ρT , i = kL a (Sliq, i−KH , i pgas, i ), (7)
dSgas, i Sgas, i qgas Vliq where the terms kl a , Sliq, i , KH , i and pgas, i denote the volumetric gas
=− + ρT , i , i = 1, …, n1,
dt Vgas Vgas (2) transfer coefficient, the dynamic state variable for soluble gas, the
Henry’s constant and summation of partial pressure exerted by in-
where Sgas, i is the gas phase state variable, qgas is the gas flow rate, Vgas is dividual biogas components (H2, CO2, CH4 and H2S), respectively.
the volume of reactor headspace. Note that only for the liquid species
hydrogen (H2), methane (CH4), carbon dioxide (CO2) and hydrogen 2.3.2. Liquid-liquid processes
sulfide (H2S), the corresponding components in the gas phase have been The original ADM1 incorporated 6 acid-base reactions which de-
considered. The expression of the rates in Eqs. (1), and (2) are discussed scribed the acid/base equilibria of acetic acid/acetate, propionic acid/
in the following subsections. propionate, butyric acid/butyrate, valeric acid/valerate, dissolved
carbon dioxide/bicarbonate and ammonium/ammonia. To explicitly
consider the effect of TEs precipitation on AD dynamics, 6 additional
668
acid/base pairs have been implemented in the developed model. They soluble components involved in the reaction and the Ksp value of the
are related to the bicarbonate/carbonate, phosphoric acid/dihydrogen corresponding precipitate. More precisely, supersaturation of reactants
phosphate, dihydrogen phosphoric acid/hydrogenphosphate, hydro- in the aqueous phase determines whether precipitation or dissolution
genphosphoric acid/phosphate, hydrogen sulfide/bisulfide and bi- take place. The general precipitation-dissolution equilibrium reaction is
sulfide/sulfide acid pairs. The kinetic acid/base reaction rates are de- represented as follows:
scribed by the following equation:
xM v + + yAv − ↔ Mx Ay (11)
ρA / B = kA / B (Sliq . A−·Sliq . H +−K a·Sliq . HA), (8)
where x and y are the number of cations and anions and v+ and v− are
where kA / B and K a are the rate coefficient for the base to acid reaction the valences. Precipitation occurs only when the solution is super-
and the acid association and dissociation constant respectively. saturated with participating ionic species. The reverse reaction, dis-
Additionally, the charge balance has been modified to take into solution, takes place when the system is undersaturated.
account the effects of additional components as:
SAC− SPRO− SBU − SVA− 2.3.3.1. Precipitation. The adopted precipitation model is a kinetic
θ = SCAT + SNH4+− − − − −SHCO3− + 2·Sca2 + + 2·SMg2 +
64 112 160 208 model (Koutsoukos et al., 1980) which takes into account the
+ 2·SNi2 + + 2·SCo2 + + 2·SFe2 +−S H2 PO4−−2·SHPO42 −−3·SPO43 −−SHS−−2·S S 2 − important steps involved in the precipitation reaction. The model is
an improvement from first order models and takes into consideration
−SAN (9) the intricate processes (development of supersaturation, nucleation,
Eq. (9) can be explicated in terms of S H+ as follows: and growth) in the precipitation of a mineral phase. The concentration
based kinetic rate of the precipitation process has been modelled as
θ 1 2 following:
S H+ = − + θ + 4Kw
2 2 (10)
+ + − −
ρprec = kr′, Mv +Av − [[[M m ]v · [Aa ]v ]1/ v −(Ksp
′ , Mv +Av − )1/ v]n , (12)
where θ is the net charge in the system resulting from all the acid/bases
considered in the model, S H+ is the variable for hydrogen ion con- m+ a−
where, [M ] and [A ] denote the concentrations of cations and
centration and KW is the ionic product of water.
′ , Mv +Av − is the so-
anions, kr′, Mv +Av − is the precipitation rate constant, Ksp
lubility product, v+ and v− are the total number of cationic and anionic
2.3.3. Liquid-solid processes
charges respectively and v = v+ + v−, n is a constant for crystallization
Solid precipitation processes are the major liquid–solid processes
of sparingly soluble salts. The precipitation processes are reversible in
that were not included in the original ADM1. The presence and absence
the model.
of the precipitates could bring a variation in the physicochemistry of
the reactor which may alter the pH, thereby affecting bio-chemical
2.3.3.2. Dissolution. Likewise, when the system is undersaturated the
processes. To improve the model’s ability to describe the non-biologi-
reverse reaction takes place leading to the dissolution of the insoluble
cally mediated processes during AD, precipitation-dissolution reactions
mineral precipitate into soluble ions. In undersaturated systems, the
have been introduced in the extended ADM1.
dissolution reaction can be represented as:
The precipitation processes have been modelled by following a well-
established approach and the individual precipitation reactions, taken + + − −
ρdiss = −kr′, Mv +Av − [[[M m ]v ·[Aa ]v ]1/ v −(Ksp
′ , Mv +Av − )1/ v]n
into consideration in this work, are the most likely to occurr for TEs in 1
AD systems (Musvoto et al., 2000b; Roussel, 2012; Zhang et al., 2015b). × Kprec
1+
Emphasis has been given to simulate the solid precipitation involving Mv +Av − (13)
Ni, Fe and Co, and three major anionic components i.e. inorganic
where Kprec is an assumed constant. At undersaturation, dissolution
carbon, inorganic sulfur and inorganic phosphorus. In addition, Ca and
takes place till equilibrium is reached or until the mineral precipitate in
Mg have been considered as well. The precipitation reactions taken into
the reactor becomes zero.
consideration are depicted in Table 1. The solid precipitates formed are
CaCO3, MgCO3, NiCO3, CoCO3, FeCO3, Ca3(PO4)2, Ni3(PO4)2,
2.4. Model parameters and initial conditions
Fe3(PO4)2, Co3(PO4)2, NiS, CoS, FeS and NH4MgPO4·6H2O.
The precipitation kinetics rate depends on the concentration of the
An in-silico batch anaerobic digester of working volume 0.75 l (head
space volume 0.25 l) was adopted for numerical simulations. The
Table 1
List of the precipitation/dissolution reactions considered in the model.
temperature of the reactor was set at 35 °C. The initial conditions for
each specific simulation scenario were selected on the basis of experi-
# Reaction pKsp Reference ence and literature and are reported in Table 2. The TEs concentration
range was selected from literature (Thanh et al., 2015). The initial
(l) + CO3(l) ⇋ CaCO3
Ca2+ 2−
1 (s) 8.48 (Plummer and
Busenberg, 1982) values of TEs were set at different values based on the experiment and
(l) + CO3(l) ⇋ MgCO3 (s)
Mg2+ 2−
2 7.46 (Smith et al., 2004) effect under study. For scenario 2 (starvation), the initial TE con-
(l) + CO3(l) ⇋ NiCO3 (s)
Ni2+ 2−
3 11.2 (Smith et al., 2004) centrations as well as the fractions of TEs produced from disintegration
(l) + CO3(l) ⇋ FeCO3 (s)
Fe2+ 2−
4 10.59 (Preis and Gamsjäger,
were set to zero. The TE concentrations in the biomass (fermented in-
2001)
5 (l) + CO3(l) ⇋ CoCO3 (s)
Co2+ 2−
11.2 (Smith et al., 2004) oculum) have not been explicitly considered. The initial values for
6 Ca2+
(l) + PO 4(l) ⇋ Ca3(PO4)2 (s)
3−
28.4 (Smith et al., 2004) ammonia and bicarbonate were adjusted at certain value so as to run
(l) + PO4(l) ⇋ Ni3(PO4)2 (s)
Ni2+ 3−
7 31 (Smith et al., 2004) the reactor around neutral pH. The initial amount of phosphorus and
(l) + + PO4(l) ⇋ Co3(PO4)2
Co2+ 3−
8 34 (Smith et al., 2004)
(s)
sulfur were taken from literature (Musvoto et al., 2000b; Sharma et al.,
(l) + PO4(l) ⇋ Fe3(PO4)2 (s)
Fe2+ 3−
9 37.76 (Smith et al., 2004)
2014). The stoichiometric and kinetic parameters for the biochemical
(l) + S(l) ⇋ FeS (s)
Fe2+ 2−
10 18 (Callander and Barford,
1983) compartment were adopted according to ADM1 for mesophilic systems.
(l) + S(l) ⇋ NiS
Ni2+ 2−
11 (s) 20.7 (Callander and Barford, The model parameters related to the precipitation rates, the TE bio-
1983) uptake and TE dose-response functions were newly introduced in this
(l) + S(l) ⇋ CoS
Co2+ 2−
12 (s) 20.3 (Callander and Barford,
study. The values for the precipitate saturation constants (Ksp) as well as
1983)
13 (l) + NH4(l) + PO4(l) ⇋ MgNH4PO4
Mg2+ + 3−
(s) 13.26 (Ohlinger et al., 1998) the acid/base equilibrium constants were collected from various
sources (Table 1).
669
Table 2 complexation and sorption etc. Therefore, precipitation reactions with

Initial concentration of various dynamic state variables considered in the model suitable initial concentrations of the reacting species were considered in
(* Refer to Table 3). the Visual MINTEQ equilibrium solver. The ionic strength of the system
State No Variable Scenario #1 – Scenario #2 – Scenario #3 – Unit was set to zero for all the predictions. In order to compare model results
Speciation Starvation Variable S/P with MINTEQ, the simulation time was increased so as to get a steady
state for all the variables considered in the model. The model outputs
1 Ssu 0 0 0 g COD L−1
were compared with the Visual MINTEQ predictions, confirming that
2 Saa 0 0 0 g COD L−1
3 Sfa 0 0 0 g COD L−1 the model accurately predicts the formation of precipitates and the pH
4 Sva 0 0 0 g COD L−1 dynamics. This comparison holds good for all the 13 precipitation re-
5 Sbu 0 0 0 g COD L−1 actions taken into consideration.
6 Spro 0 0 0 g COD L−1
7 Sac 0 0 0 g COD L−1
8 Sh2 0 0 0 g COD L−1
3.2. Exploration by simulation: effect of trace elements on anaerobic
9 Sch4 0 0 0 g COD L−1 digestion
10 Sco2 0 0 0 g COD L−1
11 Snh3 0 0 0 g COD L−1 The model capability of covering different ranges of input values of
12 SI 0 0 0 g COD L−1
different parameters represents a sort of priority. Moreover, an accurate
13 Xc 2 2 1 g COD L−1 model should also be able to simulate the effect of variation of opera-
14 Xch 0 0 0 g COD L−1
tional parameters, which can strongly affect the reactor performances
15 Xpr 0 0 0 g COD L−1
16 Xli 0 0 0 g COD L−1 (Mattei et al., 2015). For this reason, three different simulation sce-
17 Xsu 0.12 0.12 0.012 g COD L−1 narios were considered. The first scenario examined the effect that
18 Xaa 0.12 0.12 0.012 g COD L−1 various initial TE concentrations exert on TE speciation and on turn on
19 Xfa 0.12 0.12 0.012 g COD L−1 methane production. The second scenario was aimed at investigating
20 Xc4 0.12 0.12 0.012 g COD L−1
21 Xpro 0.12 0.12 0.012 g COD L−1
the role of TE starvation on the AD dynamics while the third scenario
22 Xac 0.12 0.12 0.012 g COD L−1 referred to the effects that a variable initial sulfur/phosphorus ratio
23 Xh2 0.12 0.12 0.012 g COD L−1 exerts on TE speciation and methanogenic activity. For each scenario,
24 XI 0 0 0 g COD L−1 various simulation runs were considered as it will be specified in the
31 Shva 0 0 0 g COD L−1 following sections.
32 Shbu 0 0 0 g COD L−1
33 Shpro 0 0 0 g COD L−1 3.2.1. Scenario 1: TE speciation
34 Shac 0 0 0 g COD L−1
35 Shco3 0 0 0 M
Numerical simulations were run to investigate the effect of different
36 Snh4 0.006 0.006 0.001 M initial TE concentrations on the performances of an anaerobic digester
operated in batch conditions. Sixteen simulation RUNS were con-
37 Sco3 0.005 0.005 0.0025 M
38 SCa 0.002 0.002 0.002 M sidered: the initial concentrations of Fe, Ni and Co were varied in the
40 SMg 0.0025 0.0025 0.0025 M ranges10−6 M to 10−3 M, 10−7 M to 10−4 M and 10−9 M to 10−6 M,
42 SNi Variable* Variable* 0.00001 M respectively. The specific initial data for the numerical simulations are
44 SCo Variable* Variable* 0.0000001 M
provided in Tables 2 and 3. The simulation results reported in Fig. 2
46 SFe Variable* Variable* 0.0001 M
51 Spo4 0.006 0.006 Variable* M demonstrate that, out of all the chemical systems considered in the
58 Ss2 0.001 0.001 Variable* M model, i.e. sulfide, phosphate and carbonate, sulfide precipitates pre-
dominate in the solid phase of the anaerobic batch digester. Indeed, in
15 out of 16 various simulation RUNS considered, more than 80% of TE
2.5. Model implementation precipitated as sulfide compounds (FeS, NiS and CoS). Also, the ten-
dency to form sulfide precipitate has been found higher at lower initial
The implementation of the developed mathematical model has in- concentration of TEs. A gradual increase in the concentration of car-
volved: i) the update of the biochemical framework to include the effect bonate precipitates (FeCO3, NiCO3 and CoCO3) can be observed when
of TEs on microbial kinetics as well as the release of sulfur, phosphorus increasing the initial TE concentration. It is interesting to note here
and TEs during the disintegration step; ii) the introduction of new acid- that, with increase in TE dose amount, the fraction of dissolved TE
base reactions; iii) the inclusion of solid-liquid processes; iv) the col- remains more or less constant for a given TE concentration. Under this
lection of the parameters values governing the precipitation processes. circumstance, there is an evident transfer of mass among the carbon,
In particular, these values have been collected from suitable literature phosphorus and sulfur chemical system. For example, in simulation
and scientific reports (Musvoto et al., 2000b; Roussel, 2012). RUNs 14, 15 and 16 the fraction of TEs in carbon and sulfur precipitates
The system of ordinary differential equations constituting the ex- changes. Further, the sulfur fraction decreases as the amount of TE dose
tended-ADM1 model has been implemented in an original code and has increases across all the three TE fractions. Equally important is the
been solved using the algorithm ODE 15s, a multistep, variable-order solubility of TEs. Out of the three TEs considered, Co at higher con-
solver on MATLAB® platform. Numerical simulations of specific sce- centration is slightly more soluble as compared to corresponding Fe and
narios have been performed to test model reliability. Ni concentrations. On the other hand, at lower concentration the dis-
solved fraction of Co is comparatively larger than the corresponding
3. Results fractions of Fe and Ni. Eventually, traces of CoCO3 formation start from
RUN 7.
3.1. Model verification In addition to compiling speciation data from the above mentioned
simulations (RUN 1–16), attempts have been made to assemble the
In order to verify the model accuracy and prediction ability, the effect of TE concentration on methane production as shown in Fig. 2.
precipitation sub module predictions were compared with the ones The following pattern of methane production has been observed during
obtained through Visual MINTEQ 3.1 (Gustafsson, 2018, https:// RUN 1–16. With increase in TE dose amount there is an increase in
vminteq.lwr.kth.se/download/). The latter is a freeware chemical methane production due to the increase in free TE concentration in the
equilibrium software that calculates chemical equilibrium composition system. The optimum values for the dose-response function of Fe, Ni
of natural waters by considering speciation, solubility, organic and Co have been set to 10−4 M, 10−5 M, and 10−7 M, respectively.
670
Table 3
Variable initial concentration (M) of TEs, S and P used in the model simulations.
Speciation Simulations Starvation Simulations Variable S/P Simulations
RUN Fe Ni Co RUN Fe Ni Co RUN S P
−6 −7 −9 −4 −5 −7 −4
1 5 × 10 5 × 10 5 × 10 No 1 × 10 1 × 10 1 × 10 S1 5 × 10 4.75 × 10−3
2 7 × 10−6 7 × 10−7 7 × 10−9 Ni-Fe 0.00 0.00 1 × 10−7 S2 6 × 10−4 5.00 × 10−3
3 9 × 10−6 9 × 10−7 9 × 10−9 Ni-Co 1 × 10−4 0.00 0.00 S3 7 × 10−4 5.25 × 10−3
4 1 × 10−5 1 × 10−6 1 × 10−8 Co-Fe 0.00 1 × 10−6 0.00 S4 8 × 10−4 5.50 × 10−3
5 3 × 10−5 3 × 10−6 3 × 10−8 Full 0.00 0.00 0.00 S5 9 × 10−4 5.75 × 10−3
6 5 × 10−5 5 × 10−6 5 × 10−8 S6 1 × 10−3 6.00 × 10−3
7 7 × 10−5 7 × 10−6 7 × 10−8 S7 1.1 × 10−3 6.25 × 10−3
8 9 × 10−5 9 × 10−6 9 × 10−8 S8 1.2 × 10−3 6.50 × 10−3
9 1 × 10−4 1 × 10−5 1 × 10−7 S9 1.3 × 10−3 6.75 × 10−3
10 3 × 10−4 3 × 10−5 3 × 10−7 S10 1.4 × 10−3 7.00 × 10−3
11 5 × 10−4 5 × 10−5 5 × 10−7 S11 1.5 × 10−3 7.25 × 10−3
12 7 × 10−4 7 × 10−5 7 × 10−7
13 9 × 10−4 9 × 10−5 9 × 10−7
14 1 × 10−3 1 × 10−4 1 × 10−6
15 3 × 10−3 3 × 10−4 3 × 10−6
16 5 × 10−3 5 × 10−4 5 × 10−6
Provided that, production of methane is rapidly stimulated when the TE availability of a particular TE and if this could be reflected in the cu-
concentration is on the lower end of the optimal value. On the other mulative methane production. Starvation effect was examined by set-
hand, when the TE concentration exceeds the above mentioned optimal ting initial concentration of the TEs under study at zero. For example,
value the production rate of methane reduces. With further increase in for Ni-Fe starvation, Co initial concentration was set at optimum value,
TE dose, a lag in production of methane has been observed particularly while initial concentration of Ni and Fe as well as their disintegration
when reaching an initial input Fe: Ni: Co = 3.0 × 10−3 M: fraction were set at zero. Such a strategy was adopted for all the other
3.0 × 10−4 M: 3.0 × 10−6 M (RUN 15). Similarly, concentration of cases. TEs content of the biomass (inocula) has not been considered
VFAs in the system was affected by the stimulation/inhibition effect of explicitly as the complex organic matter and TEs supplementation have
TEs on methane production. In case of inhibition, VFAs accumulated in been adopted as the major source of TEs in the anaerobic digester. More
the liquid phase. However, in the case of stimulation, there was com- precisely, five simulation runs were investigated. These include, i) No
paratively lesser accumulation of VFAs in the liquid phase (data not starvation (all the TE initial concentrations were set at the optimum
shown). value); ii) Ni-Fe Starvation (initial concentration of Co was set at op-
timum value whereas Ni and Fe are set at zero); iii) Ni-Co Starvation
(initial concentration of Fe is set at the optimum value whereas Ni and
3.2.2. Scenario 2: TE starvation
Co are set at zero); iv) Co-Fe Starvation (initial concentration Ni was set
In the second series of simulations (Fig. 3) an attempt was made to
at the optimum value whereas Co and Fe are set at zero) and; v)
check if the model could represent accurately the unavailability and
Fig. 2. TEs speciation in an anaerobic batch reactor under different initial TEs concentration. The TEs are considered to be in 4 fractions: sulfide, phosphate,
carbonate, and dissolved TE. Each bar represents a specific simulation run and the corresponding cumulative methane production is represented as cumulative
methane production.
671
fraction of sulfur precipitate was found to be predominant for all the

initial sulfur/phosphorus ratios. Moreover, as the amount of sulfur and
phosphorus initial concentration increases there is a decrease in the
fraction of sulfide precipitates. At the same time, pH profile tends to
decrease as reported in Fig. 5. The pH trend for RUN 1 was observed to
be higher than the corresponding values of RUN11. Likewise, the cu-
mulative methane production changed as well. Cumulative methane
production increased with increase in initial concentration of sulfur,
phosphorus in parallel to the increase in amount of free TE fraction in
the system.
Finally, after completing simulations based on sulfur/phosphorus
ratio, dynamic speciation of elements (Fe, Ni, Co, Ca and Mg) in the
system was obtained for RUN S6 and it has been presented in Fig. 6. The
primary aim of this exercise was to demonstrate the dynamic prediction
Fig. 3. Cumulative methane production under different starvation conditions. capability of the model. So, a suitable simulation (RUN S6) was se-
Five cases have been considered: for the Ni-Fe starvation case, Ni and Fe have lected. The model was applied to simulate time-varying speciation of
been set at an initial concentration value of zero, while Co was set at the op- TEs among carbonate, sulfide and phosphorus species. The TEs showed
timum value. This has been repeated for Ni-Co and Co-Fe. The initial con- a similar trend in increase of sulfide precipitation (FeS, NiS and CoS)
centration values have been set close to zero for “All-Starvation”. In case of “No- with time. The decrease in TEs concentration in the batch AD system is
Starvation” the optimum concentration of Fe, Ni and Co (i.e. 10−4 M, 10−5 M
largely dependent on the formation of sulfide precipitates. Phosphate
and 10−7 M, respectively) have been set as initial concentration values.
and carbonate precipitates are present in trace amount. Similarly, light
metals, Ca2+ and Mg2+ speciation was simulated for 100 days. CaCO3
Complete Starvation (initial concentrations of all TEs were set at zero). and Ca3(PO)4 are the major precipitates. The process of precipitation as
The exact values for the initial concentration of TEs are reported in well as dissolution is evident in case of Ca. MgCO3 is the major pre-
Table 3. As shown in Fig. 3, the change in the initial concentration of cipitate for Mg, which formed in the beginning of the simulation
TEs strongly affects the methane production. It is found that, with (∼2.498 × 10−3 M), within few hours of simulation time. However,
changing the starvation TEs, the total methane production also dissolution took place slowly and by end of 20 days MgCO3 got fully
changed. When there is no starvation of TE on the batch system the dissolved.
cumulative methane production after 100 days is 2.47 × 10−2 M. On
the other hand, when there is complete starvation of all the TEs cu- 4. Discussions and perspectives
mulative methane production decreased by 8 times to 3 × 10−3 M.
Likewise, cumulative methane production was affected when one of the The model results presented in this study demonstrate the relation
metals is supplied in optimum concentration and the rest of two TEs are between the physicochemical and biochemical processes in AD. The
set to a value close to 0. Changing the initial concentration of Ni-Fe to observations presented above highlights the effects that TEs precipita-
zero resulted in a cumulative production of methane of 1.8 × 10−2 M. tion/dissolution and bio-uptake have on change in pH and on overall
Similarly, changing the initial input of Ni-Co and Co-Fe to zero resulted production of methane. Further, a set of simulations indicated the
in a reduced cumulative production of 1.34 × 10−2 M and predictive capacity of the model in terms of TEs starvation and effect of
1.97 × 10−2 M. In all the cases, a reduction in methane production is change in sulfur and phosphorus initial concentration. Finally, the
accompanied by the accumulation of acetate in the liquid phase (data change in TE speciation has been represented as an illustration of dy-
not shown). namic speciation prediction. In the following section, we have discussed
the appropriateness of a dynamic mathematical model as a tool to as-
3.2.3. Scenario 3: Effect of sulfur and phosphorus sess TE speciation in AD.
The combined effect of sulfur and phosphorus on the precipitation The primary objective of a process modelling study is to determine
of TEs and methane production was investigated. The simulations were which factors or processes should be taken into consideration and to
carried out in the presence of varying amount of sulfur and phosphorus what extent. Objectives such as understanding TE speciation in AD
as shown in Figs. 4 and 5. The deliberate difference in initial con- systems through process models or elucidating the interaction of var-
centrations (Tables 2 and 3) as compared to the speciation simulations ious inorganic species in an AD system requires complex model struc-
was attempted to record effect of sulfur and phosphorus precipitation of ture that can extend system boundaries to include solid-liquid-gas
TE. Total sulfur (mainly as HS−) and phosphorus (mainly as HPO42−) phase. For instance, the original ADM1 has been extended to include
initial concentration were varied from 5.0 × 10−4 M of S and the interactions among phosphorus, sulfur and iron. Concurrently, their
4.75 × 10−3 M of P to 1.5 × 10−3 M of S and 7.25 × 10−4 M of P. This potential effects on total biogas production (CO2, CH4, H2 and H2S)
specific range was considered based on the results obtained in the have been demonstrated (Flores-Alsina et al., 2016). Similarly, dis-
previous simulations (Simulation Scenario 1) and keeping in mind the crepancies in inorganic nitrogen and carbon balances have been re-
working pH range of a batch anaerobic digester. TE initial concentra- solved and an improved physicochemical framework has been in-
tions were kept constant (Fe: Ni: Co: 10−4: 10−5: 10−7). Total methane corporated to describe solid phase precipitation reactions (Zhang et al.,
production, pH and acetic acid production profiles were recorded. 2015b). The model has been validated with the experimental work for
Sulfur and phosphorus react with Fe, Ni and Co to form, FeS and precipitation of Ca and Mg in a batch reactor (Musvoto et al., 2000a).
Fe3(PO4)2; NiS and Ni3(PO4)3; CoS and Co3(PO4)2, respectively. As Although such efforts are important in terms of process modelling and
shown in Fig. 4, with increase in initial concentration of sulfur and formulating new model structure, they are characterized by the lack of
phosphorus to 9.0 × 10−4 M and 5.75 × 10−3 M, respectively, there is an accurate model description for predicting TEs dynamics. Similarly,
negligible change in the fractionation of TE. This pattern is observed procedural shortcomings of experimental methods (Thanh et al., 2015;
across all the TE speciation in that particular range. More precisely, van Hullebusch et al., 2016) and resolution limits of analytical tech-
there is formation of carbonate and phosphate of Fe, Ni and Co in the niques have made estimation of TEs bioavailability intricate. Never-
initial runs (RUNS 1, 2, 3, 4 and 5). The fractions of carbonate and theless, in some cases, steady state chemical equilibrium modelling
phosphate are very small (less than 1%) and they are not visible in the have been employed as well (Aquino and Stuckey, 2007; Yekta et al.,
graphical representations of the simulation results. In contrast, the 2014). Under these circumstances, formulating an ADM1 based
672
Fig. 4. TEs speciation and cumulative methane production under different initial sulfur/phosphorus ratios. The TEs are considered to be in 4 fractions: sulfide,
phosphate, carbonate, and dissolved fraction. Each bar represents a specific simulation run and the corresponding cumulative methane production is represented as
cumulative methane production.
Indeed, a TE has opposite effect in small dose than in large dose, which
results in the stimulation of methane production at low concentrations
and inhibition for higher concentrations (Fermoso et al., 2008). In ad-
dition, simulation results demonstrate that the model was able to cap-
ture changes in the pH of the system that is indirectly due to change in
reactor chemical composition and TE precipitation at a given point of
reactor simulation. That is, demonstrating the importance of physico-
chemical processes in AD system.
A detail analysis of the speciation simulations establishes the like-
lihood of TEs to precipitate, in various chemical forms, in a given set of
initial conditions for a batch AD system. The mineral precipitates, FeS,
CoS and NiS, constitute the major mineral fraction found in the anae-
robic system as confirmed by experimental studies (Fermoso et al.,
Fig. 5. pH trend under different initial sulfur/phosphorus ratios. 2009; Morse and Luther, 1999; Roussel, 2012), Fe forms insoluble salts
with sulfur. Solubility product values of Ni and Co are close but are
higher than iron. In another instance, 99.8% of iron and zinc has been
dynamic model to determine TEs speciation may be useful and sup-
demonstrated to be precipitated as MeS by using sulfate reducing
portive to experimental studies.
bacteria. In the same study X-ray diffraction analysis revealed that FeS,
In order to elucidate TEs bioavailability and hence define an op-
FeS2 and ZnS are the predominant compounds (Kaksonen et al., 2003).
timum dosing strategy, the time varying aspect of TE speciation should
Precipitation of sulfide with iron has been extensively investigated
be explicitly emphasized. Moreover, one of the preeminent issues ap-
(Gustavsson, 2012; Liu et al., 2015; Roussel and Carliell-Marquet, 2016;
pearing in estimation of TE in AD system is the process conditions
Yekta et al., 2017, 2016). The two main precipitates that have been
(Thanh et al., 2017a). Estimating and correlating TEs speciation at
observed are pyrite and amorphous FeS in anaerobic granular sludge
different initial concentration of dose, using a dynamic model, as de-
reactors (van der Veen et al., 2007). According to Pourbaix diagram,
monstrated in this work, would capture the fractionation of TEs across
pyrite may be the main iron sulfide mineral found in anaerobic en-
the solid-liquid-gas phase of the reactor, in particular in the solid phase.
vironments (Roussel, 2012). However, amorphous FeS is found pre-
Tracking the dynamic evolution of TEs might contribute to understand
dominantly before the start of pyritization (Nielsen et al., 2005) process
and design a TE dosing strategy for stable reactor operation.
which has been agreeably adapted in the model and illustrated by the
This study demonstrates that differences in CH4 production can be
model simulations. Other iron sulfide minerals might have been in-
linked to the speciation of TEs and the change in reactor environment in
cluded in the model by considering an additional oxidation-reduction
terms of pH. Above all, depending upon anaerobic digestion condition
module. This has been temporarily overlooked to define a re-
and composition there is equilibrium between the liquid and solid
presentative yet simple model able to catch and describe the major
phases (Le-Chatelier’s principle). Thus, addition of TEs in anaerobic
physicochemical processes occurring in a real anaerobic batch system.
digester may disturb this equilibrium and affect CH4 production as
TE-carbonates are the next major class of minerals found in the si-
highlighted in this study. TEs have been considered to have a Hormesis
mulation results. Although metal carbonates have low solubility pro-
effect on AD (Bartacek et al., 2008). Hormesis is a term used to describe
duct values as compared to metal sulfide and metal phosphate, they
favourable biological response to low exposure to stressors/inhibitors.
should not form in AD systems. However, it has been demonstrated that
673
Fig. 6. Dynamic speciation: Change in speciation of TEs (including Ca and Mg) over 100 days under optimal TE condition captured at RUN no. 6 of variable S/P
simulations.
carbonates could precipitate with metals as second or third precipitate TEs has been also considered as a direct input into the system.
in the case of sulfide limitation (Callander and Barford, 1983). In our 2) The redox state of the system may be investigated to improve the
case, the formation of metal carbonate precipitates might be related to TEs speciation prediction.
the pH range attained for most of the simulation studies which favours 3) The effect of TE diffusion phenomenon across biofilms may be ad-
metal carbonate precipitation. In water-carbonate chemical system, the ditionally studied to assess the role of speciation in anaerobic sys-
first ionization constant for carbonic acid is 6.4 and the second ioni- tems.
zation constant is 10.3. This implies that bicarbonate is the pre- 4) Other physico-chemical processes (organic complexation and bio-
dominant carbonate species for a pH range of 7–8. Although carbonate sorption) might be studied to consolidate the understanding of
is the dominant species around pH 11–12, carbonate minerals start speciation and fluxes of TEs operating in AD.
forming in small quantity from a pH range of 6–8. This explains the
small amount of carbonate precipitates formed in the system.
5. Conclusions
The modelling effort required to strengthen the concept that me-
thane production potential is a function of TE speciation, keeping in
A general approach to model TEs precipitation and its effect on
mind the usage of model to recruit a control strategy, may require
methane production, has been implemented to extend the original
additional model complexity in terms of significant processes and
ADM1. An original tailor code to solve the aqueous chemistry and
components. Nevertheless, the goal of this study is to introduce a model
precipitation reactions has been used and verified by chemical equili-
framework based on ADM1 which can simulate the effect of TEs on AD.
brium calculations. The extended ADM1 is able to quantify the amount
This model can be used to formulate a TE dosing scheme or a reduced
and chemical forms (i.e. carbonate, sulfide and phosphate) of TEs
version of this can be used as a model predictive control of TE dosing in
present at a particular point of time in an AD reactor. Sulfide was found
AD. At this stage, with unavailability of proper experimental data to
as major precipitate. Further, unavailability or availability of TEs and
calibrate and validate such a model, it is necessary to individuate and
initial sulfur-phosphorus ratio affect the overall methane production in
define physicochemical processes which can affect the speciation of TEs
anaerobic batch reactor.
in anaerobic digesters. The use of such dynamic model based on ADM1
to predict the speciation of TEs in anaerobic digestion system is still
under development and some limitations and gap areas in this study Acknowledgements
can be summarized below:
This study was carried out in the framework of Marie Skłodowska-
1) The source of trace elements is not clear. The present model for- Curie European Joint Doctorate (EJD) in Advanced Biological Waste-
mulation considers TEs to be produced as a disintegration product To-Energy Technologies (ABWET) funded by the Horizon 2020
during the first non-enzymatic degradation step of AD. Moreover, Program (grant agreement 643071). The authors thank COST
Action1302 (European Network on Ecological Roles of Trace Metals in
674
Anaerobic Biotechnologies) meetings for insightful sessions on trace 1021/ja00525a015.

elements speciation in anaerobic digestion. The authors acknowledge Liu, Y., Wang, Q., Zhang, Y., Ni, B.J., 2015. Zero valent iron significantly enhances me-
thane production from waste activated sludge by improving biochemical methane
meaningful discussions held at University of Naples Federico II with Dr. potential rather than hydrolysis rate. Sci. Rep. 5, 1–6. https://doi.org/10.1038/
Ludovico Pontoni and Dr. Vincenzo Luongo. srep08263.
Lizarralde, I., Fernández-Arévalo, T., Brouckaert, C., Vanrolleghem, P., Ikumi, D.S.,
Ekama, G.A., Ayesa, E., Grau, P., 2015. A new general methodology for incorporating
Appendix A. Supplementary data physico-chemical transformations into multi-phase wastewater treatment process
models. Water Res. 74, 239–256. https://doi.org/10.1016/j.watres.2015.01.031.
Supplementary data associated with this article can be found, in the Mattei, M.R., Frunzo, L., D’Acunto, B., Esposito, G., Pirozzi, F., 2015. Modelling microbial
population dynamics in multispecies biofilms including Anammox bacteria. Ecol.
online version, at https://doi.org/10.1016/j.biortech.2018.06.099. Modell. 304, 44–58. https://doi.org/10.1016/j.ecolmodel.2015.02.007.
Mbamba, K., Christian, Tait, S., Flores-Alsina, X., Batstone, D.J., 2015. A systematic study
References of multiple minerals precipitation modelling in wastewater treatment. Water Res. 85,
359–370. https://doi.org/10.1016/j.watres.2015.08.041.
Morse, J.W., Luther, G.W., 1999. Chemical influence on trace metalsulphide interactions
Aquino, S.F., Stuckey, D.C., 2007. Bioavailability and toxicity of metal nutrients during in anoxic sediments. Geochim. Cosmochim. Acta 63https://doi.org/10.1016/S0016-
anaerobic digestion. J. Environ. Eng. 133, 28–35. https://doi.org/10.1061/(ASCE) 7037(99)00258-6. 3378 3378.
0733-9372(2007) 133:1(28). Musvoto, E.V., Ekama, G.A., Wentzel, M.C., Loewenthal, R.E., 2000a. Extension and ap-
Barrera, E.L., Spanjers, H., Solon, K., Amerlinck, Y., Nopens, I., Dewulf, J., 2015. plication of the three-phase weak acid/base kinetic model to the aeration treatment
Modeling the anaerobic digestion of cane-molasses vinasse: extension of the of anaerobic digester liquors. Water SA 26, 417–438.
Anaerobic Digestion Model No. 1 (ADM1) with sulfate reduction for a very high Musvoto, E.V., Wentzel, M.C., Loewenthal, R.E., Ekama, G.A., 2000b. Integrated che-
strength and sulfate rich wastewater. Water Res. 71, 42–54. https://doi.org/10. mical-physical processes modelling – I. Development of a kinetic-based model for
1016/j.watres.2014.12.026. mixed weak acid/base systems. Water Res. 34, 1857–1867. https://doi.org/10.1016/
Bartacek, J., Fermoso, F.G., Lens, P.N.L., 2008. Cobalt toxicity in anaerobic granular S0043-1354(99)00334-6.
sludge: influence of chemical speciation. J. Ind. Microbiol. Biotechnol. 35, Nielsen, A.H., Lens, P., Vollertsen, J., Hvitved-Jacobsen, T., 2005. Sulfide-iron interac-
1465–1474. https://doi.org/10.1007/s10295-008-0448-0. tions in domestic wastewater from a gravity sewer. Water Res. 39, 2747–2755.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., https://doi.org/10.1016/j.watres.2005.04.048.
Sanders, W.T., Siegrist, H., Vavilin, V.A., 2002. The IWA Anaerobic Digestion Model Ohlinger, K.N., Young, T.M., Schroeder, E.D., 1998. Predicting struvite formation in di-
No 1 (ADM1). Water Sci. Technol. 45, 65–73. https://doi.org/10.2166/wst.2008. gestion. Water Res. 32, 3607–3614. https://doi.org/10.1016/S0043-1354(98)
678. 00123-7.
Batstone, D.J., Amerlinck, Y., Ekama, G., Goel, R., Grau, P., Johnson, B., Kaya, I., Steyer, Plummer, L.N., Busenberg, E., 1982. The solubilities of calcite, aragonite and vaterite in
J.P., Tait, S., Takacs, I., Vanrolleghem, P.A., Brouckaert, C.J., Volcke, E., 2012. CO2-H2O solutions between 0 and 90 °C, and an evaluation of aqueous model for the
Towards a generalized physicochemical framework. Water Sci. Technol. 66, system CaCO3-CO2-H2O. Geochim. Cosmochim. Acta 46, 1011–1040.
1147–1161. https://doi.org/10.2166/wst.2012.300. Preis, W., Gamsjäger, H., 2001. Thermodynamic investigation of phase equilibria in metal
Cai, Y., Zhao, X., Zhao, Y., Wang, H., Yuan, X., Zhu, W., Cui, Z., Wang, X., 2018. carbonate-water-carbon dioxide systems. Monatsh. Chem. 132, 1327–1346. https://
Bioresource Technology Optimization of Fe 2 + supplement in anaerobic digestion doi.org/10.1007/s007060170020.
accounting for the. Bioresour. Technol. 250, 163–170. https://doi.org/10.1016/j. Roussel, J., Carliell-Marquet, C., 2016. Significance of vivianite precipitation on the
biortech.2017.07.151. mobility of iron in anaerobically digested sludge. Front. Environ. Sci. 4, 1–12.
Callander, I.J., Barford, J.P., 1983. Precipitation, chelation, and the availability of metals https://doi.org/10.3389/fenvs.2016.00060.
as nutrients in anaerobic digestion. II. Applications. Biotechnol. Bioeng. 25, Roussel, J., 2012. Metal behaviour in anaerobic sludge digesters supplemented with trace
1959–1972. https://doi.org/10.1002/bit.260250806. nutrients. PhD Thesis, Univ. Birmingham, United Kingdom.
Dai, X., Hu, C., Zhang, D., Dai, L., Duan, N., 2017. Impact of a high ammonia-ammonium- Shakeri, S., Gustavsson, J., Svensson, B.H., Skyllberg, U., 2012. Talanta Sulfur K-edge
pH system on methane-producing archaea and sulfate-reducing bacteria in meso- XANES and acid volatile sulfide analyses of changes in chemical speciation of S and
philic anaerobic digestion. Bioresour. Technol. 245, 598–605. https://doi.org/10. Fe during sequential extraction of trace metals in anoxic sludge from biogas reactors.
1016/j.biortech.2017.08.208. Talanta 89, 470–477. https://doi.org/10.1016/j.talanta.2011.12.065.
Ekama, G.A., Wentzel, M.C., Loewenthal, R.E., 2006. Integrated chemical-physical pro- Sharma, K., Derlon, N., Hu, S., Yuan, Z., 2014. Modeling the pH effect on sulfidogenesis in
cesses kinetic modelling of multiple mineral precipitation problems. Water Sci. anaerobic sewer biofilm. Water Res. 49, 175–185. https://doi.org/10.1016/j.watres.
Technol. 53, 65–73. https://doi.org/10.2166/wst.2006.407. 2013.11.019.
Esposito, G., Frunzo, L., Panico, A., Pirozzi, F., 2011. Modelling the effect of the OLR and Smith, R.M., Martell, A.E., Motekaitis, R.J., 2004. NIST critically selected stability con-
OFMSW particle size on the performances of an anaerobic co-digestion reactor. stants of metal complexes database. NIST Standard Reference Database 46, Version 8.
Process Biochem. 46, 557–565. https://doi.org/10.1016/j.procbio.2010.10.010. 0. National Institute of Standard and Technology, Gaithersburg, MD, USA.
Federovich, V., Lens, P., Kalyuzhnyi, S., 2003. Extension of Anaerobic Digestion Model Solon, K., Flores-Alsina, X., Mbamba, C.K., Volcke, E.I.P., Tait, S., Batstone, D., Gernaey,
No. 1. Appl. Biochem. Biotechnol. 109, 33–45. https://doi.org/10.1385/ABAB:109:1- K.V., Jeppsson, U., 2015. Effects of ionic strength and ion pairing on (plant-wide)
3:33. modelling of anaerobic digestion. Water Res. 70, 235–245. https://doi.org/10.1016/
Fermoso, F.G., Bartacek, J., Jansen, S., Lens, P.N.L., 2008. Metal supplementation to j.watres.2014.11.035.
UASB bioreactors: from cell-metal interactions to full-scale application. Sci. Total Thanh, et al., 2015. Trace metal speciation and bioavailability in anaerobic digestion: a
Environ. 407, 3652–3667. https://doi.org/10.1016/j.scitotenv.2008.10.043. review. Biotechnol. Adv. 34, 122–136. https://doi.org/10.1016/j.biotechadv.2015.
Fermoso, F.G., Bartacek, J., Jansen, S., Lens, P.N.L., 2009. Metal supplementation to 12.006.
UASB bioreactors: from cell-metal interactions to full-scale application. Sci. Total Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2017a. Effect of operating conditions on
Environ. 407, 3652–3667. https://doi.org/10.1016/j.scitotenv.2008.10.043. speciation and bioavailability of trace metals in submerged anaerobic membrane
Flores-Alsina, X., Solon, K., Kazadi Mbamba, C., Tait, S., Gernaey, K.V., Jeppsson, U., bioreactors. Bioresour. Technol. 243, 810–819. https://doi.org/10.1016/j.biortech.
Batstone, D.J., 2016. Modelling phosphorus (P), sulfur (S) and iron (Fe) interactions 2017.07.040.
for dynamic simulations of anaerobic digestion processes. Water Res. 95, 370–382. Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2017b. Effect of Ethylenediamine-N, N’-
https://doi.org/10.1016/j.watres.2016.03.012. disuccinic acid (EDDS) on the speciation and bioavailability of Fe 2 + in the presence
Glass, J.B., Orphan, V.J., 2012. Trace metal requirements for microbial enzymes involved of sulfide in anaerobic digestion. Bioresour. Technol. 229, 169–179. https://doi.org/
in the production and consumption of methane and nitrous oxide. Front. Microbiol. 3, 10.1016/j.biortech.2016.12.113.
1–20. https://doi.org/10.3389/fmicb.2012.00061. van der Veen, A., Fermoso, F.G., Lens, P.N.L., 2007. Bonding from analysis of metals and
Gustafsson, J.P., (https://vminteq.lwr.kth.se/download/), 2018. Visual Minteq 3.1. sulfur fractionation in methanol-grown anaerobic granular sludge. Eng. Life Sci. 7,
Gustavsson, J., 2012. Cobalt and Nickel Bioavailability for Biogas Formation. PhD 480–489. https://doi.org/10.1002/elsc.200720208.
Thesis. University of Linkoping, Sweden. van Hullebusch, E.D., Guibaud, G., Simon, S., Lenz, M., Shakeri Yekta, S., Fermoso, F.G.,
Jimenez, J., Latrille, E., Harmand, J., Robles, A., Ferrer, J., Gaida, D., Wolf, C., Mairet, F., Jain, R., Duester, L., Roussel, J., Guillon, E., Skyllberg, U., Almeida, C.M.R., Pechaud,
Bernard, O., Alcaraz-Gonzalez, V., Mendez-Acosta, H., Zitomer, D., Totzke, D., Y., Garuti, M., Frunzo, L., Esposito, G., Carliell-Marquet, C., Ortner, M., Collins, G.,
Spanjers, H., Jacobi, F., Guwy, A., Dinsdale, R., Premier, G., Mazhegrane, S., Ruiz- 2016. Methodological approaches for fractionation and speciation to estimate trace
Filippi, G., Seco, A., Ribeiro, T., Pauss, A., Steyer, J.P., 2015. Instrumentation and element bioavailability in engineered anaerobic digestion ecosystems: an overview.
control of anaerobic digestion processes: a review and some research challenges. Rev. Crit. Rev. Environ. Sci. Technol. 46, 1324–1366. https://doi.org/https://doi.org/10.
Environ. Sci. Biotechnol. 14, 615–648. https://doi.org/10.1007/s11157-015-9382-6. 1080/10643389.2016.1235943.
Kaksonen, A.H., Riekkola-Vanhanen, M.L., Puhakka, J.A., 2003. Optimization of metal Xu, F., Li, Y., Wang, Z.W., 2015. Mathematical modeling of solid-state anaerobic diges-
sulphide precipitation in fluidized-bed treatment of acidic wastewater. Water Res. 37, tion. Prog. Energy Combust. Sci. 51, 49–66. https://doi.org/10.1016/j.pecs.2015.09.
255–266. https://doi.org/10.1016/S0043-1354(02)00267-1. 001.
Ketheesan, B., Minh, P., Stuckey, D.C., 2016. Bioresource Technology Iron deficiency and Yekta, S.S., Svensson, B.H., Björn, A., Skyllberg, U., 2014. Thermodynamic modeling of
bioavailability in anaerobic batch and submerged membrane bioreactors (SAMBR) iron and trace metal solubility and speciation under sulfidic and ferruginous condi-
during organic shock loads. Bioresour. Technol. 211, 136–145. https://doi.org/10. tions in full scale continuous stirred tank biogas reactors. Appl. Geochem. 47, 61–73.
1016/j.biortech.2016.03.082. https://doi.org/10.1016/j.apgeochem.2014.05.001.
Koutsoukos, P., Amjad, Z., Tomson, M.B., Nancollas, G.H., 1980. Calcium phosphates. A Yekta, S.S., Skyllberg, U., Danielsson, Å., Björn, A., Svensson, B.H., 2016. Chemical
constant composition study. J. Am. Chem. Soc. 102, 1553–1557. https://doi.org/10. speciation of sulfur and metals in biogas reactors—implications for cobalt and nickel
675
bio-uptake processes. J. Hazard. Mater. 324, 110–116. https://doi.org/10.1016/j. Zhang, Y., Piccard, S., Zhou, W., 2015b. Improved ADM1 model for anaerobic digestion
jhazmat.2015.12.058. process considering physico-chemical reactions. Bioresour. Technol. 196, 279–289.
Yekta, S.S., Ziels, R.M., Björn, A., Skyllberg, U., Ejlertsson, J., Karlsson, A., Svedlund, M., https://doi.org/10.1016/j.biortech.2015.07.065.
Willén, M., Svensson, B.H., 2017. Importance of sulfide interaction with iron as Zhang, W., Zhang, L., Li, A., 2015a. Enhanced anaerobic digestion of food waste by trace
regulator of the microbial community in biogas reactors and its effect on methano- metal elements supplementation and reduced metals dosage by green chelating agent
genesis, volatile fatty acids turnover, and syntrophic long-chain fatty acids de- [S, S]-EDDS via improving metals bioavailability. Water Res. 84, 266–277. https://
gradation. J. Biosci. Bioeng. 123, 597–605. https://doi.org/10.1016/j.jbiosc.2016. doi.org/10.1016/j.watres.2015.07.010.
12.003.
676
Table A.1: List of variables used in the model.
# Description
1 Dynamic state variable for soluble sugars
2 Dynamic state variable for soluble amino acids
3 Dynamic state variable for soluble fatty acids
4 Dynamic state variable for soluble valerate
5 Dynamic state variable for soluble butyrate
6 Dynamic state variable for soluble propionate
7 Dynamic state variable for soluble acetate
8 Dynamic state variable for soluble hydrogen
9 Dynamic state variable for soluble methane
10 Dynamic state variable for soluble carbon dioxide
11 Dynamic state variable for soluble ammonia
12 Dynamic state variable for soluble inerts
13 Dynamic state variable for particulate composite
14 Dynamic state variable for particulate carbohydrates
15 Dynamic state variable for particulate protein
16 Dynamic state variable for particulate lipids
17 Dynamic state variable for particulate sugar degraders
18 Dynamic state variable for particulate amino acid degraders
19 Dynamic state variable for particulate fatty acid degraders
20 Dynamic state variable for particulate valerate and butyrate degraders
21 Dynamic state variable for particulate propionate degraders
22 Dynamic state variable for particulate acetate degraders
23 Dynamic state variable for particulate hydrogen degraders
24 Dynamic state variable for particulate inerts
25 Dynamic state variable for gas phase hydrogen
26 Dynamic state variable for gas phase methane
27 Dynamic state variable for gas phase carbon dioxide
31 Dynamic state variable for soluble valeric acid
32 Dynamic state variable for soluble butyric acid
33 Dynamic state variable for soluble propionic acid
34 Dynamic state variable for soluble acetic acid
35 Dynamic state variable for soluble bicarbonate ion
36 Dynamic state variable for soluble ammonium ion
37 Dynamic state variable for soluble carbonate ion
38 Dynamic state variable for soluble divalent calcium ion
39 Dynamic state variable for calcium carbonate (CaCO3) precipitate
40 Dynamic state variable for soluble divalent magnesium ion
41 Dynamic state variable for magnesium carbonate (MgCO3) precipitate
42 Dynamic state variable for soluble divalent nickel ion
43 Dynamic state variable for nickel (II) carbonate (NiCO3) precipitate
44 Dynamic state variable for soluble divalent cobalt ion
45 Dynamic state variable for cobalt (II) carbonate (CoCO3) precipitate
46 Dynamic state variable for soluble divalent iron ion
47 Dynamic state variable for iron (II) carbonate (FeCO3) precipitate
48 Dynamic state variable for soluble H3PO4- ion
49 Dynamic state variable for soluble H2PO4- ions
50 Dynamic state variable for soluble HPO42-ions
51 Dynamic state variable for soluble PO43- ions
52 Dynamic state variable for Nickel (II) phosphate (Ni3(PO4)2) precipitate
53 Dynamic state variable for Cobalt (II) phosphate (Co3(PO4)2) precipitate
54 Dynamic state variable for Iron (II) phosphate (Fe3(PO4)2) precipitate
55 Dynamic state variable for Calcium (II) (Ca3(PO4)2) precipitate
56 Dynamic state variable for soluble H2S
57 Dynamic state variable for soluble HS- ions
58 Dynamic state variable for soluble S2- ions
59 Dynamic state variable for Nickel sulfide (NiS) precipitate
60 Dynamic state variable for Cobalt sulfide (CoS) precipitate
61 Dynamic state variable for Iron sulfide (FeS) precipitate
62 Dynamic state variable for Struvite (NH4MgPO4) precipitate
63 Dynamic state variable for gas phase H2S
Table A.2: Biochemical rate coefficients for soluble components.
Component 1 2 3 4 5 6 7 8 9 10 11 12
Process Ssu Saa Sfa Sva Sbu Spro Sac Sh2 Sch4 Sco2 Snh3 SI
1 Disintegration
2 Hydrolysis of carbohydrates
3 Hydrolysis of proteins
4 Hydrolysis of lipids
5 Uptake of sugars ∑
6 Uptake of amino acids ∑
7 Uptake of LCFA ( ) ( ) ( )
8 Uptake of Valerate
9 Uptake of Butyrate
10 Uptake of propionate ( ) ( ) ∑ ( )
11 Uptake of acetate ∑
12 Uptake of Hydrogen ∑
13 Decay of Xsu
14 Decay of Xaa
15 Decay of Xfa
16 Decay of Xc4
17 Decay of Xpro
18 Decay of Xac
19 Decay of XH2
Table A.3: Biochemical rate coefficients for particulate components.
Component 13 14 15 16 17 18 19 20 21 22 23 24
Process Xc Xch Xpr Xli Xsu Xaa Xfa Xc4 Xpro Xac Xh2 XI
1 Disintegration
5 Uptake of sugars
6 Uptake of amino acids
7 Uptake of LCFA
10 Uptake of propionate
11 Uptake of acetate
12 Uptake of Hydrogen
13 Decay of Xsu
14 Decay of Xaa
15 Decay of Xfa
16 Decay of Xc4
17 Decay of Xpro
18 Decay of Xac
19 Decay of XH2
Table A.4: Biochemical rate coefficients for TEs uptake.
Component 13 42 44 46 50 57 22
Process Xc SNi2+ SCo2+ SFe2+ Shpo4 Shs Xac
1 Disintegration
11 Uptake of Acetate
Table A.5: Kinetic rate equation for biochemical processes
Process Kinetic rate
1 Disintegration
5 Uptake of sugars
6 Uptake of amino acids
7 Uptake of LCFA
10 Uptake of propionate
11 Uptake of acetate
13 Decay of Xsu
14 Decay of Xaa
15 Decay of Xfa
16 Decay of Xc4
17 Decay of Xpro
18 Decay of Xac
19 Decay of XH2
Table A.6: Stoichiometric matrix for acid-base reactions.
Component 4 5 6 7 10 11 31 32 33 34 35 36 37 48 49 50 51 56 57 58
Process Sva Sbu Spro Sac Sco2 Snh3 Shva Shbu Shpro Shac Shco3 Snh4 Sco3 Sh3po4 Sh2po4 Shpo4 Spo4 Sh2s Shs Ss2
A1 Valerate acid-base
A2 Butyrate acid-base
A3 Propionate acid-base
A4 Acetate acid-base
A5 HCO3- acid-base
A6 NH4+ acid-base
A7 CO32- acid-base
A8 H2PO4acid-base
A9 HPO42- acid-base
A10 PO43- acid-base
A11 HS- acid-base
A12 S2- acid-base
Table A.7: Kinetic rate equations for acid-base processes.
A1 Valerate acid-base ⁄ ( )
A2 Butyrate acid-base ( )
⁄
A3 Propionate acid-base ( )
⁄
A4 Acetate acid-base ( )
⁄
A5 HCO3- acid-base ⁄ ( )
A6 NH4+ acid-base ⁄ ( )
2-
A7 CO3 acid-base ⁄ ( )
A8 H2PO4- acid-base ⁄
( )
A9 HPO42- acid-base ⁄
( )
3-
A10 PO4 acid-base ⁄ ( )
-
A11 HS acid-base ⁄
( )
2-
A12 S acid-base ⁄ ( )
Table A.8: Stoichiometric matrix for mineral precipitation reactions.
Component 35 37 51 58 38 40 42 44 46 39 41 43 45 47 52 53 54 55 59 60 61 62
Process Snh4 Sco3 Spo4 Ss2 Sca2+ SMg2+ SNi2+ SCo2+ SFe2+ X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13
P1 CaCO3 (s) Precipitation
P2 MgCO3(s) Precipitation
P3 NiCO3(s) Precipitation
P4 CoCO3(s) Precipitation
P5 FeCO3(s) Precipitation
P6 Ca3(PO4)2(s) Precipitation ⁄
P7 Ni3(PO4)2 (s) Precipitation ⁄
P8 Co3(PO4)2 (s) Precipitation ⁄
P9 Fe3(PO4)2(s) Precipitation ⁄
P10 FeS (s) Precipitation
P11 NiS (s) Precipitation
P12 CoS (s) Precipitation
P13 MgNH4PO4(S) Precipitation
Table A.9: Kinetic rates for precipitation processes.
⁄
⁄
P1 CaCO3 (s) Precipitation [( ) ]
MgCO3(s) Precipitation ⁄
⁄
P2 [( ) ]
NiCO3(s) Precipitation ⁄
⁄
P3 [( ) ]
CoCO3(s) Precipitation ⁄
⁄
P4 [( ) ]
P5 FeCO3(s) Precipitation ⁄
⁄
[( ) ]
P6 Ca3(PO4)2(s) Precipitation ⁄ ⁄ ⁄
* +
P7 Ni3(PO4)2 (s) Precipitation ⁄ ⁄ ⁄
* +
P8 Co3(PO4)2 (s) Precipitation ⁄ ⁄ ⁄
* +
P9 Fe3(PO4)2(s) Precipitation ⁄ ⁄ ⁄
* +
P10 FeS (s) Precipitation ⁄ ⁄
* +
P11 NiS (s) Precipitation ⁄ ⁄
* +
P12 CoS (s) Precipitation ⁄ ⁄
* +
P13 NH4MgPO4(S) Precipitation ⁄
⁄
[( ) ]
Table A.10a: Input for MINTEQ
Run Ni2+ Fe2+ Co2+ NH4+ CO32-

1 - - - - -
2 - + + + +
4 + + - + -
5 + - + - -
6 + + - + +
7 - + + - -
8 - + - + -
9 - + + - +
10 + - + - +
11 + + + + +
Table A.10b: Comparison between model simulation and MINTEQ results.

Minteq Simulation
pH NiCO3 CoCO3 FeCO3 pH NiCO3 CoCO3 FeCO3
-6 -6 -4 -4 -4
8.629 9.5×10 9.66×10 9.9×10 8.61 0.11×10 0.10×10 0.001
8.220 9.49×10-6 9.96×10-5 1.9×10-3 8.215 0.98×10-5 0.99×10-4 0.002
9.947 9.93×10-5 9.48×10-6 1.99×10-3 9.95 0.10×10-3 0.10×10-4 0.002
8.790 9.95×10-5 9.97×10-6 9.99×10-4 8.77 0.10×10-3 0.10×10-3 0.001
-5 -6 -3 -3 -4
8.220 9.94×10 9.68×10 1.99×10 8.21 0.11×10 0.12×10 0.002
-5 -5 -3 -5 -3
9.412 9.56×10 9.96×10 1.99×10 9.41 0.99×10 0.10×10 0.002
-6 -6 -3 -4 -4
9.910 9.38×10 9.55×10 1.99×10 9.91 0.12×10 0.14×10 0.002
-5 -3 -6 -4
6.181 ----NA-- 6.17×10 1.86×10 6.18 0.35×10 0.30×10 0.0018
-5 -5 -4 -4 -4
6.117 5.21×10 5.70×10 8.52×10 6.12 0.59×10 0.25×10 0.86*10-3
8.271 9.95×10-5 9.97×10-6 1.99×10-3 8.27 0.10×10-3 0.10×10-3 0.002
Table A.11: Parameter values used in various sub-modules of the proposed model.
Parameter Value Unit Parameter Value Unit Parameter Value Unit Parameter Value Unit
Vliq 1.0 L f_xI.xc 0.25 - Vgas 0.25 L Ksp_FeCO3 10.59 -
kdis 0.5 d-1 f_ch.xc 0.20 - Pgaslim 0 Pr_FeCO3 102 M-1d-1
Khydch 10 d-1 f_pr.xc 0.20 - KA 108 M-1d-1 Dis_FeCO3 102 M-1d-1
khyd_pr 10 d-1 f_li.xc 0.25 - Ka 10-4.84 M Ksp_Ni3PO4_2 31 -
khyd_li 10 d-1 f_hs.xc 0.0006 - KB 108 M-1d-1 Pr_Ni3PO4_2 102 M-1d-1
km_su 30 d-1 f_hpo4.xc 0.00006 - Kb 10-4.82 M Dis_Ni3PO4_2 102 M-1d-1
Ks_su 0.5 g COD L-1 f_fa.li 0.950 - KPRO 108 M-1d-1 Ksp_Fe3PO4_2 37.76 -
km_aa 50 d-1 f_bu.su 0.130 - Kpro 10-4.88 M Pr_Fe3PO4_2 102 M-1d-1
Ks_aa 0.3 g COD L-1 f_pro.su 0.270 - KAC 108 M-1d-1 Dis_Fe3PO4_2 102 M-1d-1
km_fa 6 d-1 f_ac.su 0.410 - Kac 1.8*10-5 M Ksp_Co3PO4_2 34 -
Ks_fa 0.4 g COD L-1 f_h2.su 0.190 - KCO 108 M-1d-1 Pr_Co3PO4_2 102 M-1d-1
km_c4 20 d-1 f_va.aa 0.230 - Kco 10-6.35 M Dis_Co3PO4_2 102 M-1d-1
Ks_c4 0.2 g COD L-1 f_bu.aa 0.260 - KNI 108 M-1d-1 Ksp_Ca3PO4_2 28.4 -
km_pro 13 d-1 f_pro.aa 0.050 - kni 5.6*10-10 M Pr_Ca3PO4_2 102 M-1d-1
Ks_pro 0.1 g COD L-1 f_ac.aa 0.400 - KCO3 108 M-1d-1 Dis_Ca3PO4_2 102 M-1d-1
km_ac 8 d-1 f_h2.aa 0.060 - Kco3 4.8*10-11 M Ksp_NiS 20.7 -
Ks_ac 0.15 g COD L-1 C_su 0.03130 mole C (g COD)-1 kh3po4 6.9*10-3 M Pr_NiS 102 M-1d-1
km_h2 35 d-1 C_aa 0.030 mole C (g COD)-1 KH3PO4 108 M-1d-1 Dis_NiS 102 M-1d-1
Ks_h2 7*10-6 g COD L-1 C_bu 0.0250 mole C (g COD)-1 khpo4 6.2*10-8 M Ksp_CoS 20.3 -
kdec_Xsu 0.02 d-1 C_pro 0.02680 mole C (g COD)-1 KHPO4 108 M-1d-1 Pr_CoS 102 M-1d-1
kdec_Xaa 0.02 d-1 C_ac 0.03130 mole C (g COD)-1 kpo4 4.8*10-13 M Dis_CoS 102 M-1d-1
kdec_Xfa 0.02 d-1 C_va 0.0240 mole C (g COD)-1 KPO4 108 M-1d-1 Ksp_FeS 18 -
kdec_Xc4 0.02 d-1 C_bac 0.03130 mole C (g COD)-1 Pr_CaCO3 102 M-1d-1 Pr_FeS 102 M-1d-1
kdec_Xpro 0.02 d-1 C_ch4 0.01560 mole C (g COD)-1 Ksp_CaCO3 8.48 - Dis_FeS 102 M-1d-1
kdec_Xac 0.02 d-1 N_bac 0.08/14 mole N (g COD)-1 Dis_CaCO3 102 M-1d-1 Ksp_MgNH4PO4 13.26 -
kdec_Xh2 0.02 d-1 N_aa 0.0070 mole N (g COD)-1 Pr_MgCO3 102 M-1d-1 Pr_MgNH4PO4 102 M-1d-1
Y_su 0.100 - kla_h2 3.84 d-1 Ksp_MgCO3 7.46 - Dis_MgNH4PO4 102 M-1d-1
Y_aa 0.080 - kla_met 2.16 d-1 Dis_MgCO3 102 M-1d-1 pH_LL1 4.0 -
Y_fa 0.060 - kla_co2 60 d-1 Ksp_NiCO3 11.2 - pH_UL1 5.5 -
Y_c4 0.060 - KH_h2 0.00078 kmole bar-1 Pr_NiCO3 102 M-1d-1 pH_LL2 6.0 -
Y_pro 0.040 - KH_ch4 0.00116 kmole bar-1 Dis_NiCO3 102 M-1d-1 pH_UL2 7.0 -
Y_ac 0.050 - KH_co2 0.035 kmole bar-1 Ksp_CoCO3 11.2 pH_LL3 5.0 -
Y_h2 0.060 - T 308.16 K Pr_CoCO3 102 pH_UL3 6.0 -
f_sI.xc 0.10 - R 0.08314 bar kmole-1 K-1 Dis_CoCO3 102
Bioresource Technology 276 (2019) 253–259
Contents lists available at ScienceDirect
Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech
ADM1 based mathematical model of trace element complexation in T

anaerobic digestion processes
⁎
Bikash Chandra Maharaja,b,d, , Maria Rosaria Matteib, Luigi Frunzob, Eric D. van Hullebuschc,d,e,
Giovanni Espositof
a
b
University of Naples Federico II, Department of Mathematics and Applications “Renato Caccioppoli”, Naples, Italy
c
d
IHE Delft Institute for Water Education, Department of Environmental Engineering and Water Technology, Delft, the Netherlands
e
Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Université Paris Diderot, UMR 7154, CNRS, F-75005 Paris, France
f
University of Naples Federico II, Department of Civil, Architectural and Environmental Engineering, Naples, Italy
G R A P H I C A L A B S T R A C T
A R T I C LE I N FO A B S T R A C T
Keywords: In this study, a new model based on anaerobic digestion model no.1 (ADM1) approach has been proposed to
Anaerobic digestion simulate trace elements (TEs) complexation, precipitation and their effect on the anaerobic batch methane
Trace elements production. TEs complexation reactions with VFAs and EDTA have been incorporated in an extended ADM1
Complexation model which considers TE precipitation/dissolution reactions as well as biodegradation processes. The kinetic
Mathematical modeling
model tracks the dynamics of 90 state variables which constitute the components of the proposed anaerobic
ADM1
digestion (AD) model. The incorporation of the complexation reactions required the definition of new inorganic
components (EDTA species) and new complexation process rates in the ADM1 framework. The charge balance
was modified accordingly to consider the effects of the additional components. The new model is able to predict:
a) the effect of TE-EDTA/VFA complexation on methane production, and b) the effect of the initial calcium and
magnesium concentrations on process performance.
1. Introduction organic matter comprises of biomass, organics from the biodegradation,

such as volatile fatty acids (VFAs), alcohol and acetate. The production
Anaerobic digestion is a bioprocess rich in organic matter. The of these organic ligands from the biodegrdation of organic matter is
⁎
Corresponding author at: University of Cassino and the Southern Lazio, Department of Civil and Mechanical Engineering, Cassino, Italy.
E-mail address: [email protected] (B.C. Maharaj).
Received 11 October 2018; Received in revised form 18 December 2018; Accepted 20 December 2018
Available online 21 December 2018
0960-8524/ © 2018 Elsevier Ltd. All rights reserved.
believed to be one of the major chemical reactions in AD. These organic in the proposed model are the predominant ones occurring in an
ligands act as chelating agent which might influence the TE speciation. anaerobic digestor. The complexes result from the reaction between TE
Additionally, in course of AD, synthetic chelating agents are occasion- cations and the synthetic organic ligand EDTA (Ethylenediamine tetra
ally added to improve the digester performance. Synthetic agents such acetic acid) as well as the volatile fatty acids.
as ethyleneaminetetraacetic acid (EDTA), Nitrilotriacetic acid (NTA)
and Ethylenediamine-N,N′-disuccinic acid (EDDS) influence the bioa- 2.1.1. TE-EDTA complexation
vailability of TEs in AD by forming stable soluble complexes. Thus, the The formation and dissociation of TE complexes constitute the main
two major groups of organic species which influence the TE partitioning feature of the proposed model and are considered to be kinetically
in AD are: 1) organic substrates (VFAs); and 2) synthetic agents (EDTA controlled. The complexation reaction involving the TE complex
or NTA). Complexation reactions diverge the TEs flux from the solid [MeEDTA]2− is used here to illustrate the kinetic framework adopted in
phase precipitation to the liquid phase complexation (Aquino and the complexation module of the proposed model. New state variables
Stuckey, 2007). This transfer of TE flux increases the bioavailability of constituting the EDTA acid/base system and complexes have been in-
TEs and hence reduces the TE requirements in the anaerobic digester troduced in the complexation module. The approach used is based on a
(Thanh et al., 2017a). previous study (Willet and Rittmann, 2003). The complexation reac-
Despite a significant amount of research efforts being dedicated tions considered are as follows:
(Tsapekos et al., 2018; van Hullebusch et al., 2016), there is no con- k1
solidated understanding of the dynamic partitioning of TE species in AD Me 2 + + EDTA4 − → [MeEDTA]2 − (1)
systems among different organic and inorganic fractions (Fermoso
k−1
et al., 2015). Given the available experimental techniques and proce- [MeEDTA]2 − → Me 2 + + EDTA4 − (2)
dures, a direct determination of the entire speciation pattern of TEs is
The Eqs. (1) and (2) can be rewritten as:
difficult to accomplish (van Hullebusch et al., 2016). In this regard, it is
logical to address the research gaps by a modeling approach. Thus, a K
[MeEDTA]−2
Me 2 + + EDTA4 − ↔ [MeEDTA]2 − (3)
mathematical model based on ADM1 may be recruited to study the
effect of TE-complexation reactions on the dynamics and hence the The rate equation for the mechanism in Eq. (3) can be written as:
bioavailability of TEs in an AD system. Although there are some ex-
ρcomp = Cx (k1 S EDTA4 − ∙S Me2 + − k−1 ∙S[MeEDTA]2 −) (4)
perimental studies focusing on the effect of TE-complexes EDTA/NTA/
EDDS on methane production (Schiavon Maia et al., 2017; Thanh et al., where, ρcomp is the complexation process rate, Cx is the kinetic rate
2017a,b; Vintiloiu et al., 2013), there has been no attempt to in- constant, S EDTA4 −, S Me2 + , S[MeEDTA]2 − are the dynamic state variables for
corporate the TE bioavailability and TE-complexation processes in AD EDTA4−, TEs and TE complex formed. The overall formation constant
mathematical models. To date, the mathematical models incorporating of a TE-EDTA complex is set at a particular value and the reverse rate
physicochemical processes are limited to the precipitation reactions constant is calculated from the values of the equilibrium constant,
(Flores-Alsina et al., 2016; Maharaj et al., 2018; Mbamba et al., 2015; otherwise known as stability constant (Willet and Rittmann, 2003). The
Zhang et al., 2015a, b). formation rate constants, k1, for all the species have been assumed to be
The main objective of this work is to develop an ADM1 based 10 × 1010 and the dissociation rate constants, k−1, have been calculated
mathematical model to: 1) mechanistically describe all the main bio- from the stability constants.
chemical and physicochemical processes required for EDTA and VFAs
complexation in addition to precipitation/dissolution and uptake of TEs 2.1.2. Effect of TEs on microbial metabolism
in AD; 2) implement a modified TE dose-response function which takes The effect of TEs on biochemical processes has been explicitly taken
into account soluble TEs (complexed and free) in liquid phase in AD; 3) into account by introducing an additional non-competitive biostatic
investigate into the complexation of TE with EDTA and VFAs, in par- inhibition term (I Me2 + ) in the inhibition expressions used in the ADM1
ticular the fractionation of TEs among precipitates and dissolved com- framework (Batstone et al., 2002):
plexes in relation to the change in initial concentrations of calcium and
2+
magnesium, and TE dose in AD. a1 ∙ (Me 2 + + C Me ) + a2
I Me2 + = 2+ 2+
,
(Me 2 + + C Me )2 + b1 ∙ (Me 2 + + C Me ) + b2 (5)
2. Model development and simulation
where, a1, a2 , b1, b2 are assumed constants which can been adjusted to
2.1. Mathematical model obtain a desirable optimum dose-response function at a particular free
2+
TE Me2+ and complexed TE C Me concentration. Monod-type non-
In this study, a new model based on anaerobic digestion model no.1 competitive inhibition functions, extensively used in ADM1, can only be
(ADM1) approach has been proposed to simulate the TE complexation, used for a unitary effect. Such a function cannot emulate stimulation
precipitation and its effect on AD in terms of total methane production. and inhibition regimes simultaneously. However, the function (Eq. (5))
In this regard, the biodegradation processes, including hydrolysis, considers both the stimulation and inhibiting action of TEs (hormesis)
acidogenesis, acetogenesis, methanogenesis, and the acid-base equili- on the biochemical rates depending on the concentration range. Both
brium reactions have been coupled to the TE complexation, precipita- complexed TEs and free TEs have been considered as bioavailable
tion and redissolution processes and the effects on total methane pro- fraction of TEs.
duction have been evaluated. A full kinetic framework has been used to
implement the biological and physicochemical model processes. 2.2. Model simulations
Overall, the model tracks the dynamics of 90 state variables which
constitute the components of the proposed AD model. The incorpora- The prediction capability of a model can be evaluated by its ability
tion of the precipitation and complexation reactions in the physico- to cover a wide range of input values. More importantly, a robust model
chemical module leads to the definition of new inorganic components should illustrate the capacity to simulate a variation in the critical
in the ADM1 framework. In particular, the precipitation, redissolution operating parameters. Due to this reason, two simulation scenarios have
and complexation reactions considered in this model are the most likely been considered. The first simulation set refers to a variable EDTA
formed in AD systems. Precipitation and redissolution take place at the dosage with optimal (SCENARIO 1(A)) and suboptimal (SCENARIO
solid/liquid interphase and have been modeled according to Maharaj 1(b)) TEs initial concentrations in the reactor. Ca, Mg and EDTA initial
et al. (2018). The organic complexation reactions (Table 1) considered concentration variation has been considered in the second set of
254
Table 1
List of the complexation reactions considered in the model.
# Reaction Stability Constant (Log K) Reference
(l) + EDTA(l) ⇋ [Ca-EDTA]

Ca2+ 4− 2−
1 (l) 10.7 Callander and Barford (1983)
Mg2+(l) + EDTA(l) ⇋ [Mg-EDTA]
4− 2−
2 (l) 8.7 Callander and Barford (1983)
3 Ni(l) + EDTA(l) ⇋ [Ni-EDTA]
2+ 4− 2−
(l) 20 Smith et al. (2004)
(l) + EDTA(l) ⇋ [Co-EDTA]
Co2+ 4− 2−
4 (l) 18 Smith et al. (2004)
(l) + EDTA(l) ⇋ [Fe-EDTA]
Fe2+ 4− 2−
5 (l) 16 Smith et al. (2004)
−
6 Ca(l) + Va(l) ⇋ [Ca-Va] (l)
2+ +
0.3 Smith et al. (2004)
−
(l) + Bu(l) ⇋ [Ca-Bu]
Ca2+ +
7 (l) 0.94 Smith et al. (2004)
−
(l) + Pro(l) ⇋ [Ca-Pro]
Ca2+ +
8 (l) 0.93 Smith et al. (2004)
−
9 Ca(l) + Ac(l) ⇋ [Ca-Ac] (l)
2+ +
1.18 Smith et al. (2004)
−
10 Mg(l) + Bu(l) ⇋ [Mg-Bu] (l)
2+ +
0.96 Smith et al. (2004)
−
Mg2+(l) + Pro(l) ⇋ [Mg-Pro]
+
11 (l) 0.97 Smith et al. (2004)
−
Mg2+(l) + Ac(l) ⇋ [Mg-Ac]
+
12 (l) 1.26 Smith et al. (2004)
−
13 Ni(l) + Bu(l) ⇋ [Ni-Bu] (l)
2+ +
0.691 Smith et al. (2004)
−
(l) + Pro(l) ⇋ [Ni-Pro]
Ni2+ +
14 (l) 1.19 Smith et al. (2004)
−
(l) + Ac(l) ⇋ [Ni-Ac]
Ni2+ +
15 (l) 1.44 Smith et al. (2004)
−
16 Co(l) + Bu(l) ⇋ [Co-Bu] (l)
2+ +
0.591 Smith et al. (2004)
−
(l) + Pro(l) ⇋ [Co-Pro]
Co2+ +
17 (l) 1.13 Smith et al. (2004)
−
18 Co2+
(l) + Ac (l) ⇋ [Co-Ac] +
(l) 1.38 Smith et al. (2004)
−
(l) + Ac(l) ⇋ [Fe-Ac]
Fe2+ +
19 (l) 1.40 Smith et al. (2004)
Table 2
Initial concentrations (M) of EDTA, Ca and Mg used in the model simulations.
Scenario 1(A) Scenario 1(B) Scenario 2
EDTA with optimal TEs EDTA with sub-optimal TEs Variable Ca-Mg Simulations
RUN EDTA RUN EDTA RUN EDTA Ca Mg
1 0.00 1 0.00 CM1 0.00 1 × 10−3 1 × 10−3

2 1 × 10−5 2 1 × 10−6 CM2 0.00 2 × 10−3 2 × 10−3
3 2 × 10−5 3 2 × 10−6 CM3 0.00 3 × 10−3 3 × 10−3
4 3 × 10−5 4 3 × 10−6 CM4 0.00 4 × 10−3 4 × 10−3
5 4 × 10−5 5 4 × 10−6 CM9 5.0 × 10−5 1 × 10−3 1 × 10−3
6 5 × 10−5 6 5 × 10−6 CM10 5.0 × 10−5 2 × 10−3 2 × 10−3
7 6 × 10−5 7 6 × 10−6 CM11 5.0 × 10−5 3 × 10−3 3 × 10−3
8 7 × 10−5 8 7 × 10−6 CM12 5.0 × 10−5 4 × 10−3 4 × 10−3
9 8 × 10−5 9 8 × 10−6 CM17 1 × 10−4 1 × 10−3 1 × 10−3
10 9 × 10−5 10 9 × 10−6 CM18 1 × 10−4 2 × 10−3 2 × 10−3
11 1 × 10−4 11 1 × 10−5 CM19 1 × 10−4 3 × 10−3 3 × 10−3
CM20 1 × 10−4 4 × 10−3 4 × 10−3
simulations (SCENARIO 2). The initial conditions for the particular the order (most dominant TE sulfide precipitate first): Fe > Co > Ni.
scenarios were selected on the basis of experience and literature and However, with the increase in the EDTA concentration, the amount of
have been reported in Table 2. The TEs concentration range was se- sulfide precipitate decreased considerably for all the TEs. Across all the
lected from literature (Thanh et al., 2015). The initial amount of cal- simulation RUNs the tendency to form sulfide precipitate was higher in
cium and magnesium were taken from literature (Musvoto et al., 2000). the case of Fe. For Co, the amount of sulfide precipitate was com-
The values for the stability constant K were collected from various paratively lower than FeS. The amount of CoS drastically decreased
sources (Table 1). The system of ordinary differential equations con- with increase in EDTA dosing. Ni was found to be the most bioavailable
stituting the extended-ADM1 model has been implemented in an ori- TE, even in RUN 1 (Figs. 1a and 1b). With the maximum dosage of
ginal code and has been solved using the algorithm ODE 15s, on MA- EDTA (RUN 11; Figs. 1a and 1b) the amount of sulfide precipitate for Ni
TLAB® platform. The predictions of the complexation module have been and Co was very low or negligible. However, FeS was found in the
verified with the chemical equilibrium calculations (Visual Minteq 3.5). system even at the maximum EDTA concentration. With the increase in
EDTA concentration, a gradual decrease in carbonate precipitates
(FeCO3, NiCO3 and CoCO3) was also observed. The formation of
3. Results
phosphate precipitates was very low and it was not evident in the
fractionation schemes. The fraction of free TE is higher when EDTA is
3.1. Scenario 1: Addition of EDTA with TEs in anaerobic digestion
added with sub-optimal TEs (Fig. 1b) in comparison to EDTA added
with optimal TEs (Fig. 1a). It is interesting to note that the rate of mass
Model simulations were carried out to study the effect of initial
transfer from sulfide precipitates to complexes is higher when EDTA is
EDTA concentrations on the production of biogas in an AD batch system
added to the reactor with optimal concentration of TEs. Indirectly, the
(Figs. 1a and 1b). For each scenario (Scenario 1(A) and 1(B)), 11 si-
solubility of TEs increased with the addition of EDTA as a large fraction
mulation RUNs were analysed. EDTA was dosed with optimal TEs in
of TEs transformed into [TE-EDTA]2−. The solubility of Fe was ob-
Scenario 1(A). In Scenario 1(B) EDTA was added with sub-optimal
served to be less as compared to Ni and Co in both Fig. 1a and b. In RUN
TEs. The specific values for each model RUN have been provided in
11 (Figs. 1a and b), with the highest EDTA dose, almost the entire Ni
Table 2. The numerical study reported in Figs. 1a and 1b demonstrated
and Co transformed to soluble fraction ([Ni-EDTA]2− and [Co-
that, out of all the three chemical systems (sulfide, carbonate and
EDTA]2−), while for Fe considerable amount remained as FeS in the
phosphate), sulfide dominated the batch system when EDTA was not
reactor. With the increase in initial EDTA concentration with optimum
dosed (RUN 1; Figs. 1a and 1b). Affinity for sulfide precipitation is in
255
Fe Speciation Ni Speciation
100 100
NiS(s)
80 80
NiCO3(s)
% Fraction Fe
% Fraction Ni
FeS(s) Ni3(PO4)2(s)
60 60
FeCO (s) 2+
3 Ni (l)
Fe (PO ) (s) +
3 4 2 NiBu (l)
40 40
2+
Fe (l) NiPro+(l)
+ +
FeAc (l) NiAc (l)
20 20
2- 2-
[FeEDTA] (l) [NiEDTA] (l)
0 0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
Co Speciation CH4 production

0.025
100
Control
CoS(s) 1
0.02
80 CoCO (s) 2
3
Co (PO ) (s) 3
% Fraction Co
3 4 2
60
0.015 4
Co2+(l)
Moles
5
CoBu+(l)
40 0.01 6
+
CoPro (l) 7
+
CoAc (l) 8
20 2- 0.005
[CoEDTA] (l) 9
10
0 0
1 2 3 4 5 6 7 8 9 10 11 0 10 20 30 40 50 60 70 80 90 100
Time (d)
Fig. 1a. TEs speciation in an anaerobic batch reactor under different initial EDTA concentrations with optimal TE dosage (SCENARIO 1(A)). The TEs are considered
to be in fractions: sulfide, phosphate, carbonate in solid phase; TE-EDTA, TE-VFAs (Bu, Pro and Ac) and dissolved TE in liquid phase. Each bar corresponds to a
specific simulation run at time t = 100 days, and the corresponding cumulative methane production is presented. Optimal TE concentrations: Fe = 1.0 × 10−4 M;
Ni = 1.0 × 10−5 M; and Co = 1.0 × 10−7 M.
amount of TEs Scenario 1(A), the methane production did not increase Scenario 1(A).
significantly in the anaerobic batch digester (Fig. 1a). However, the
addition of EDTA, with sub-optimal TEs Scenario 1(B), significantly 3.2. Scenario 2: Effect of calcium and magnesium initial concentration on
affected the methane production (Fig. 1b). The difference in methane TE-EDTA complexation in anaerobic digestion
production between RUN1 (without EDTA) and RUN 11 (with max-
imum EDTA dosage) is about 1 × 10−2 M of methane in Scenario 1(B). Fig. 2 depicts the effect of calcium, magnesium and EDTA initial
Such an increase in the rate of methane production was not observed in concentration on TE-EDTA complexation, TE precipitation and methane
Fe Speciation Ni Speciation
100 100
NiS(s)
80 80
NiCO (s)
3
% Fraction Fe
% Fraction Ni
Ni3(PO4)2(s)
60 FeS(s) 60
2+
FeCO (s) Ni (l)
3
40 Fe (PO ) (s) 40 NiBu+(l)

3 4 2
2+ NiPro+(l)
Fe (l)
+
20 + 20 NiAc (l)
FeAc (l)
2-
2- [NiEDTA] (l)
[FeEDTA] (l)
0 0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
Co Speciation CH4 production

0.02
100
Control
CoS(s) 1
80 CoCO (s) 0.015 2
3
Co (PO ) (s) 3
% Fraction Co
3 4 2
60 2+
4
Moles
Co (l)
0.01 5
+
CoBu (l) 6
40 +
CoPro (l) 7
CoAc+(l) 0.005 8
20
[CoEDTA]2-(l) 9
10
0 0
1 2 3 4 5 6 7 8 9 10 11 0 10 20 30 40 50 60 70 80 90 100
Time (d)
Fig. 1b. TEs speciation in an anaerobic batch reactor under different initial EDTA concentrations with sub-optimal TE dosage (SCENARIO 1(B). The TEs are
considered to be in fractions: sulfide, phosphate, carbonate in solid phase; TE-EDTA, TE-VFAs (Bu, Pro and Ac) and dissolved TE in liquid phase. Each bar corresponds
to a specific simulation run at time t = 100 days, and the corresponding cumulative methane production is presented. Optimal TE concentrations:
Fe = 1.0 × 10−5 M; Ni = 1.0 × 10−6 M; and Co = 1.0 × 10−8 M.
256
EDTA=0.00M EDTA=0.00M EDTA=0.00M

100 100 100
80 80 80
% Fraction Co
% Fraction Fe
% Fraction Ni
60 60 60
40 40 40
20 20 20
0 0 0
1 2 3 4 1 2 3 4 1 2 3 4

100 100 100
80 80 80
% Fraction Co
% Fraction Fe
% Fraction Ni
60 60 60
40 40 40
20 20 20
0 0 0
1 2 3 4 1 2 3 4 1 2 3 4

100 100 100
80 80 80
% Fraction Co
% Fraction Fe
% Fraction Ni
60 60 60
40 40 40
20 20 20
0 0 0
1 2 3 4 1 2 3 4 1 2 3 4
TES(s) TECO (s) TE (PO ) (s) 2+

TE (l)
+
TE-Bu (l) TE-Pro (l)
+ +
TE-Ac (l)
2-
[TE-EDTA] (l)
3 3 42
Fig. 2. TEs speciation under different initial EDTA, Ca and Mg concentrations. The TEs are considered to be in fractions: sulfide, phosphate, carbonate in solid phase;
TE-EDTA, TE-VFAs (Bu, Pro and Ac) and dissolved TE in liquid phase. Each row represents the TE speciation for Fe, Ni and Co when a particular concentration of
EDTA was added. Each stacked bar within a particular EDTA concentration and for a given TE, corresponds to a varying Ca and Mg initial concentration
(1 = 1.0 × 10−3 M; 2 = 2.0 × 10−3 M; 3 = 3.0 × 10−3 M; and 4 = 4.0 × 10−3 M of Ca and Mg).
production in a set of numerical simulations. The simulations were 4. Discussion

carried out for four initial concentrations of calcium and magnesium
and four EDTA regimes at optimal values of TEs (Table 2). The specific In this study, a mathematical model based on ADM1 for TE com-
range for calcium and magnesium initial concentration was selected plexation has been proposed, verified and simulated with suitable
based on previous experience. As shown in Fig. 2, an increase in EDTA scenarios. The numerical simulations of the proposed model presented
initial concentration resulted in an increase in TE-complexation. When above highlights the effect of TE-EDTA and TE-VFA complexation
no EDTA was in system, the addition of calcium and magnesium led to processes on the cumulative methane production. The numerical results
the formation of more carbonate minerals (EDTA = 0.00 M) as com- proved that the model was able to evaluate the stimulating capacity of
pared to the simulations where EDTA was already present. With further EDTA under various TEs dosage conditions in an anaerobic digester.
increase in EDTA dosage (EDTA = 2.5 × 10−5 and 10 × 10−5 M), Further, a series of simulations demonstrated the competing nature of
sulfide precipitates quantity decreased. However, with the addition of calcium and magnesium with respect to TEs for available EDTA, and
calcium and magnesium, the amount of calcium and magnesium pre- their effect on TE precipitation and cumulative methane production. In
cipitates increased and the complexation of TEs decreased across all the the following paragraphs we have discussed the qualitative prediction
concentrations of EDTA investigated. Conversely, the concentration of capability of the proposed model by comparing the simulation results
[Ca-EDTA]2− and [Mg-EDTA]2− increased with increasing the calcium, with experimental studies reported in the literature.
magnesium and EDTA addition (data not shown). With addition of The major finding of our simulation studies pertains to the sensi-
calcium and magnesium, the total cumulative methane production was tivity of microbial activity to EDTA dosing (Figs. 1a and b). The pro-
not significantly affected by the initial concentration of calcium and posed model captures the stimulation of biodegradation by EDTA when
magnesium. However, the rate of methane production decreased with sub-optimal concentration of TEs is present in the system. In contrast,
the addition of calcium and magnesium (data not shown). In summary, when optimal concentration of TEs is dosed into the system, chelating
the addition of EDTA increased the formation of EDTA complexes. effect of EDTA on stimulating biodegradation is negligible. Recently
Whereas, addition of calcium and magnesium increased the calcium Thanh et al. (2017a) reported on a Fe2+ supplementation study carried
and magnesium precipitates and decreased the sulfur precipitate. out in presence of EDDS (biodegradable structural compound analogue
to EDTA). Addition of EDDS at a molar ratio of 1:1 to Fe2+ improved
the production of methane as EDDS affected the bioavailability of Fe2+
257
by competing with sulfide. However, the study did not report on the magnesium concentration added one or both at a time. It showed the
bioavailability of Ni and Co when EDDS was added. The study focussed presence of calcium and magnesium in the liquid phase as well as the
on Fe due to its dominant presence as sulfide precipitate in AD systems. formation of precipitates. A higher initial ratio of calcium and magne-
Similar to this study, the addition of EDDS was carried out in presence sium resulted in a slower production of ammonia gas. This indicates the
of Fe + Co + Mo + Ni in anaerobic batch and semi-continuous reactor effect of liquid to solid on the liquid to gas transfer processes. The
(Zhang et al., 2015a, b). The study showed a 50% reduction of optimal authors did not explicitly mention the impact of Ca-Mg on other com-
TEs dosage due to the addition of EDDS. Such an effect of chelating ponents of biogas i.e. CH4, CO2 and H2. However, it is clear from our
agent has also been observed in our model simulations (Fig. 1b), where simulation study (data not shown) that the decrease in methane pro-
the addition of EDTA with suboptimal concentration of TEs improved duction is in parallel to the increase in total amount of precipitates.
the rate of methane production. Conversely, Fermoso et al. (2008) re- These relations may also be attributed to the pH induced process in-
ported that dosing a higher amount of EDTA into an anaerobic system hibition of the in-silico batch reactor.
may result in the formation of TE-complexes (Fermoso et al., 2008)
(here [Co-EDTA]2−), thus lowering the toxicity level because of the 5. Conclusions
strong binding of EDTA with Co. Thus, bioavailability of Co decreases
which results in reduced methane production. In the same study, when A general approach to model the effect of TEs complexation on
small amount of [Co-EDTA]2− was dosed, there was improvement in methane production has been implemented on an extended ADM1
methanogenic activity. Overall, there are multiple and contesting points model. Numerical simulations showed that: 1) the addition of EDTA is
of view on the role of chelating agents, (EDTA in particular) in AD effective with suboptimal concentration of TEs rather than optimal
systems. In a study, the methanogenic activity of an anaerobic system concentration of TEs; and 2) the variation of EDTA, Ca and Mg initial
decreased as a consequence of adding higher concentration (1 mM) of concentrations affect TE bioavailability and methane production in an
EDTA (Aquino and Stuckey, 2007). On other hand, studies have been anaerobic batch reactor. An original aqueous chemistry module of TE-
reported on stimulating nature of EDTA (Vintiloiu et al., 2013) at a EDTA & TE-VFA complexation has been verified by chemical equili-
concentration of 50 mg kg−1 of substrate (∼0.16 mM). Nevertheless, brium calculations. The model is able to quantify the chemical form
the research is focusing nowadays on finding alternate green substitute (free, precipitated and complexed) of TEs present in the system.
of EDTA such as EDDS (as reported above) and has so far attributed a
stimulatory effect on EDDS. In such a scenario, a mathematical model Acknowledgements
(which can be modified for any specific use) as presented in this study
might be used for scrutinizing chelating agents by evaluating their ef- This study was supported by the Marie Skłodowska-Curie European
fect on TE bioavailability and methane production in a limited time Joint Doctorate-ABWET, funded by the Horizon 2020 program [Grant
frame. However, the proposed model formulation is not absolute and agreement 643071]. The authors thank: COST Action 1302 (European
complete and needs further improvements. For example, the proposed Network on Ecological Roles of Trace Metals in Anaerobic
model does not capture the discrepancies in TE inhibition arising due to Biotechnologies), Dr. Ludovico Pontoni and Dr. Vincenzo Luongo for
the mode of reactor operation, viz. batch and continuous. The amount fruitful discussions.
of TE required to overcome TE limitation in AD systems is less for
continuous mode of operation as compared to batch mode. This is due Appendix A. Supplementary data
to the formation of large amount of biological ligands during the con-
tinuous mode of operation (Gonzalez-gil and Kleerebezem, 1999; Thanh Supplementary data to this article can be found online at https://
et al., 2015). doi.org/10.1016/j.biortech.2018.12.064.
[Ca-EDTA]2− and [Mg-EDTA]2− complex formation was also in-
vestigated (Fig. 2). The increase in initial calcium and magnesium leads References
to carbonate precipitates formation in the AD system. With the increase
in dosing of calcium and magnesium, the rate of methane production Aquino, S.F., Stuckey, D.C., 2007. Bioavailability and toxicity of metal nutrients during
becomes lower. It seems reasonable that the decrease in the rate of anaerobic digestion. J. Environ. Eng. 133, 28–35. https://doi.org/10.1061/(ASCE)
0733-9372(2007) 133:1(28).
methane production is largely due to the pH change of the system. This Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A.,
is evident from the fact that with the addition of higher amount of Ca- Sanders, W.T., Siegrist, H., Vavilin, V.A., 2002. The IWA Anaerobic Digestion Model
Mg in the system with no EDTA (Fig. 2), there is a sharp decrease in No 1 (ADM1). Water Sci. Technol. 45, 65–73. https://doi.org/10.2166/wst.2008.
678.
methane production rates (for 4 × 10−3 M Ca-Mg) and this correlates Callander, I.J., Barford, J.P., 1983. Precipitation, chelation, and the availability of metals
with a pH increase of the system (data not shown). The effect of calcium as nutrients in anaerobic digestion. II. Applications. Biotechnol. Bioeng. 25,
and magnesium addition on TE-EDTA complexation was also studied. 1959–1972. https://doi.org/10.1002/bit.260250806.
Fermoso, F.G., Bartacek, J., Ching, L., Lens, P., 2008. Supplementation of cobalt to UASB
Another possible reason for the decrease in methane production rate is reactors by pulse dosing: CoCl2 versus CoEDTA2− pulses. Biochem. Eng. J. 42,
the less availability of carbonates (a source of carbon) due to the in- 111–119. https://doi.org/10.1016/j.bej.2008.06.005.
crease in carbonate precipitation. The results show that the amount of Fermoso, F.G., Hullebusch, E.D Van, Guibaud, G., Collins, G., Svensson, B.H., Vink,
J.P.M., Esposito, G., Frunzo, L., 2015. Fate of trace metals in anaerobic digestion. In:
TE complexes increases when higher concentration of EDTA is added.
Gubitz, G., Bauer, A., Bochmann, G., Gronauer, A., Weiss, S. (Eds.), Biogas Science
On the other hand, the amount of precipitate (sum total of carbonate, and Technology. Advances in Biochemical Engineering/Biotechnology, pp. 171–195
phosphate, sulfide precipitate) increased when more calcium and https://doi.org/10.1007/978-3-319-21993-6.
magnesium were added. Recently (Thanh et al., 2017a), reported that Flores-Alsina, X., Solon, K., Kazadi Mbamba, C., Tait, S., Gernaey, K.V., Jeppsson, U.,
Batstone, D.J., 2016. Modelling phosphorus (P), sulfur (S) and iron (Fe) interactions
the addition of calcium and magnesium up to 4 × 10−3 M improved the for dynamic simulations of anaerobic digestion processes. Water Res. 95, 370–382.
methane production rate. However, the study did not report on the https://doi.org/10.1016/j.watres.2016.03.012.
effect of addition of calcium and magnesium on TE-EDTA complexation Gonzalez-gil, G., Kleerebezem, R., 1999. Effects of Nickel and Cobalt on Kinetics of
Methanol Conversion by Methanogenic Sludge as Assessed by On-Line CH 4
as it focused on the addition of the three metal groups: Ca-Mg, Fe-Co- Monitoring Effects of Nickel and Cobalt on Kinetics of Methanol Conversion by
Ni, and Cu-Zn-Mn, to an AD system. It was observed that the addition of Methanogenic Sludge as Assessed by On-Line CH 4 Monitoring 65, 1789–1793.
Ca-Mg improves the methane production rate as compared to other Maharaj, B.C., Mattei, M.R., Frunzo, L., van Hullebusch, E.D., Esposito, G., 2018. ADM1
metal groups. In our case, the decrease in the rate of methane pro- digestion processes. Bioresour. Technol. 267, 666–676. https://doi.org/10.1016/j.
duction correlates to a higher precipitation rate and hence to the de- biortech.2018.06.099.
crease in bioavailable TEs in the bioreactor (as explained above). An- Mbamba, K., Christian, Tait, S., Flores-Alsina, X., Batstone, D.J., 2015. A systematic study
of multiple minerals precipitation modelling in wastewater treatment. Water Res. 85,
other study (Zhang et al., 2015a, b) focused on the calcium and
258
359–370. https://doi.org/10.1016/j.watres.2015.08.041. improve the methane yield of sewage sludge. Bioresour. Technol. 270, 732–737.
Wentzel, Musvoto, M.C.M., Ekama, G.A.M., 2000. Integrated chemical-physical processes https://doi.org/10.1016/j.biortech.2018.09.136.
modelling-II. Simulating aeration treatment of anaerobic digester supernatants. van Hullebusch, E.D., Guibaud, G., Simon, S., Lenz, M., Shakeri Yekta, S., Fermoso, F.G.,
Water Res. 34, 1868–1880. https://doi.org/10.1016/S0043-1354(99)00335-8. Jain, R., Duester, L., Roussel, J., Guillon, E., Skyllberg, U., Almeida, C.M.R., Pechaud,
Schiavon Maia, D.C., Niklevicz, R.R., Arioli, R., Frare, L.M., Arroyo, P.A., Gimenes, M.L., Y., Garuti, M., Frunzo, L., Esposito, G., Carliell-Marquet, C., Ortner, M., Collins, G.,
Pereira, N.C., 2017. Removal of H2S and CO2 from biogas in bench scale and the pilot 2016. Methodological approaches for fractionation and speciation to estimate trace
scale using a regenerable Fe-EDTA solution. Renewable Energy 109, 188–194. element bioavailability in engineered anaerobic digestion ecosystems: an overview.
https://doi.org/10.1016/j.renene.2017.03.023. Crit. Rev. Environ. Sci. Technol. 46, 1324–1366. https://doi.org/10.1080/10643389.
Smith, R.M., Martell, A.E., Motekaitis, R.J., 2004. NIST critically selected stability con- 2016.1235943.
stants of metal complexes database. NIST Standard Reference Database 46. Retrieved Vintiloiu, A., Boxriker, M., Lemmer, A., Oechsner, H., Jungbluth, T., Mathies, E.,
from: < https://www.nist.gov/document-14894 > . Ramhold, D., 2013. Effect of ethylenediaminetetraacetic acid (EDTA) on the bioa-
Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2015. Trace metal speciation and vailability of trace elements during anaerobic digestion. Chem. Eng. J. 223, 436–441.
bioavailability in anaerobic digestion: a review. Biotechnol. Adv. 34, 122–136. https://doi.org/10.1016/j.cej.2013.02.104.
https://doi.org/10.1016/j.biotechadv.2015.12.006. Willet, A.I., Rittmann, B.E., 2003. Slow complexation kinetics for ferric iron and EDTA
Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2017a. Effect of ethylenediamine-N, N’- complexes make EDTA non-biodegradable. Biodegradation 14, 105–121. https://doi.
disuccinic acid (EDDS) on the speciation and bioavailability of Fe2+ in the presence org/10.1023/A:1024074329955.
of sulfide in anaerobic digestion. Bioresour. Technol. 229, 169–179. https://doi.org/ Zhang, W., Zhang, L., Li, A., 2015a. Enhanced anaerobic digestion of food waste by trace
10.1016/j.biortech.2016.12.113. metal elements supplementation and reduced metals dosage by green chelating agent
Thanh, P.M., Ketheesan, B., Stuckey, D.C., Zhou, Y., 2017b. Dosing of Ethylenediamine-N, [S, S]-EDDS via improving metals bioavailability. Water Res. 84, 266–277. https://
N″-disuccinic acid (EDDS) to improve the bioavailability of Fe2+ in the presence of doi.org/10.1016/j.watres.2015.07.010.
sulfide in a submerged anaerobic membrane bioreactor. Chem. Eng. J. 330, 175–182. Zhang, Y., Piccard, S., Zhou, W., 2015b. Improved ADM1 model for anaerobic digestion
https://doi.org/10.1016/j.cej.2017.07.158. process considering physico-chemical reactions. Bioresour. Technol. 196, 279–289.
Tsapekos, P., Alvarado-Morales, M., Tong, J., Angelidaki, I., 2018. Nickel spiking to https://doi.org/10.1016/j.biortech.2015.07.065.
259
ADM1 based mathematical model of trace element complexation in anaerobic digestion
processes
Bikash Chandra Maharaja,b,d*, Maria Rosaria Matteib, Luigi Frunzob, Eric D. van Hullebuschc,d,e and Giovanni Esposito a
a
b
University of Naples Federico II, Department of Mathematics and Applications "Renato Caccioppoli", Naples, Italy
c
d
IHE Delft Institute for Water Education, Department of Environmental Engineering and Water Technology, Delft, The Netherlands
e
Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Université Paris Diderot, UMR 7154, CNRS, F-75005 Paris, France
*Corresponding Author; E-mail: [email protected]
# Description
S1 General form of the proposed model
S2 The additional main features of the proposed model
S3 Stoichiometric matrix for EDTA protonation/deprotonation reactions.
S4 Stoichiometric matrix for VFA complexation reactions.
S5 Stoichiometric matrix for EDTA complexation reactions.
S6 Model Verification
S7 TE fractionation obtained from simulation studies.
S8 Parameters for the ADM1 based TE complexation model.
S9 Initial condition for Model Simulations

S1. General form of the proposed model
dVliq Si
= qin Sin,i − qout Si +Vliq (δi ρA,i (t, 𝐒) − ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 )) +
dt
m1 m2
+ Vliq (∑ αi,j ρbio,j (t, 𝐒, 𝐗) + ∑ βi,j ρcmplx,j (t, 𝐒)) +

(1)
j=1 j=1
m3 m3

j=1 j=1
i = 1, … , n1 , t > 0,
m1
dVliq X,i
dt (2)
j=1
i = n1 + 1, … , n2 , t > 0,
dSgas,i Sgas,i qgas

dt Vgas
(3)
i = 1, … , n1 , t > 0,
dVliq Xprec,i in
= qin Xprec,i − qout Xprec,i + Vliq (𝛼̅𝑖 ρprec,i (t, 𝐒, 𝐗 𝐩 ) + 𝛽𝑖̅ ρdissol,i (t, 𝐒, 𝐗 𝐩 )),
dt (4)
i = 1, … , m3 , t > 0,
where:
n1 is the number of soluble components;
n2 − n1 is the number of particulate components, including the microbial species and the inert material;
m1 is the number of biochemical processes considered;
m2 is the number of complexation processes considered;

m3 is the number of precipitation/dissolution processes considered;
αi,j is the stoichiometric coefficient of species ion biochemical process j;
βi,j is the stoichiometric coefficient of species i on complexation process j;
γi,j is the stoichiometric coefficient of species i on precipitation/dissolution process j;
δi,j is the stoichiometric coefficient for acid-base reaction for ith on biochemical process j;
Si is the 𝑖 𝑡ℎ soluble component,𝐒 = (S1 , … , Sn1 );
Xi is the ith particulate component,𝐗 = (Xn1 +1 , … , Xn2 );
Xprec is the ith precipitate, 𝐗 𝐩𝐫𝐞𝐜 = (Xprec,1 , … , Xprec,m3 );
Sgas,i is the ith soluble gas component, 𝐒𝐠𝐚𝐬 = (Sgas,1 , … , Sgas,n3 );
ρA,i (t, 𝐒)is the process rate for acid base reactions for ith soluble component;
ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 )is the gas transfer process rate for ith soluble component;
ρbio,j (t, 𝐒, 𝐗)is the rate for jth biochemical process;
ρcomplx,j (t, 𝐒, 𝐗)is the rate forjth complexation process;
ρprec,j (t, 𝐒, 𝐗)is the rate for jth precipitation process;
ρdissol,j (t, 𝐒, 𝐗)is the rate forjth dissolution process;
Vliq is the volume of liquid in the reactor.
S2. The additional main features/processes of the proposed model

 Liquid-gas processes: The model considers liquid-gas stripping processes for gaseous components formed during the biological degradation of
the organic complex matter. The model considers H2, CH4, CO2 and H2S gas components as significant intermediates. The concentrations of
biogas components have been estimated by solving gas transfer equations parallel to other biochemical and physicochemical processes.
 Liquid-Liquid processes: Protonation/deprotonation of VFAs, phosphate, carbonate, ammonium and sulfide species in AD processes have been
considered and implemented as differential equations. The corresponding pka values have been considered for each protonation/deprotonation
reaction. The charge balance as well as the corresponding protonation/deprotonation reactions have been modified to include the effect of EDTA
species.
 Liquid-Solid processes: The model considers precipitation/dissolution of TEs: precipitates of Fe, Ni, Co, Ca and Mg with carbonate, phosphate
and sulfide have been considered. A concentration dependent second order kinetic model has been adopted for the precipitation/dissolution
reactions Maharaj et al., 2018).
 Dose-response function: The effect of TEs on biological processes has been modeled based on a dose-response function. The multiplicative
inhibition factor𝐼𝑀𝑒 2+ , has been introduced to explicitly consider the multiple effects of TEs on microbial metabolism. This factor has been
modified in order to account for the effect of TE complexes.
 Phosphorus and sulfur: Complex organic matter has been also considered as a source for phosphorous, sulfur and TEs which arise in the system
from the disintegration step. Phosphorous has been supposed to be released as HPO 42-, which represents the most abundant form in the pH range
of 6 to 14. Similarly, the released sulfur has been considered in the form of HS -which has been depicted as the abundant form in the pH range of
6 to 12. Furthermore, the uptake of free and complexed TEs has been linked to the acetate and hydrogen uptake in the methanogenesis step. The
stoichiometry of the related processes has been updated.
S3: Stoichiometric matrix for EDTA protonation/deprotonation reactions.

Component 79 80 81 82 83 84 85 Process Rates
Process Sh6edta Sh5edta Sh4edta Sh3edta Sh2edta Shedta Sedta
A31 H6EDTA acid-base 1 −1 k A⁄ (Sh5edta− SH+ − K a,h6edta Sh6edta )
B,h6edta
−1 k A⁄ (Sh4edta− SH+ − K a,h5edta Sh5edta )
A32 H5EDTA acid-base 1
B,h5edta
1 k A⁄ (Sh3edta− SH+ − K a,h4edta Sh4edta )
A33 H4EDTA acid-base −1
B,h4edta
A34 H3EDTA acid-base 1 −1 k A⁄ (Sh2edta− SH+ − K a,h3edta Sh3edta )

B,h3edta
A35 1 −1 k A⁄ (Shedta− SH+ − K a,h2edta Sh2edta )
H2EDTA acid-base B,h2edta
A36 1 −1 k A⁄ (Sedta− SH+ − K a,hedta Shedta )
HEDTA acid-base B,hedta
S4: Stoichiometric matrix for VFA complexation reactions.
Components 4 5 6 7 38 40 42 44 46 65 66 67 68 69 70 71 72 73 74 75 76 77 78 Process Rate

Process Sva Sbu Spro Sac Sca2+ SMg2+SNi2+ SCo2+SFe2+ C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19
C6 [Ca-Va]+(l) Complexation −1 −1 1 Cx (k 6 Sva SCa2+ − k −6 SvaCa )
C7 [Ca-Bu]+(l) Complexation −1 −1 1 Cx (k 7 Sbu SCa2+ − k −7 SbuCa )
C8 [Ca-Pro]+(l) Complexation −1 −1 1 Cx (k 8 Spro SCa2+ − k −8 SproCa )
C9 [Ca-Ac]+(l) Complexation −1 −1 1 Cx (k 9 Sac SCa2+ − k −9 SacCa )
C10 [Mg-Bu]+(l) Complexation −1 −1 1 Cx (k10 Sbu SMg2+ − k −10 SbuMg )
C11 [Mg-Pro]+(l) Complexation −1 −1 1 Cx (k11 Spro SMg2+ − k −11 SproMg )
C12 [Mg-Ac]+(l) Complexation −1 −1 1 Cx (k12 Sac SMg2+ − k −12 SacMg )
C13 [Ni-Bu]+(l) Complexation −1 −1 1 Cx (k13 Sbu SNi2+ − k −13 SbuNi )
C14 [Ni-Pro]+(l) Complexation −1 −1 1 Cx (k14 Spro SNi2+ − k −14 SproNi )
C15 [Ni-Ac]+(l) Complexation −1 −1 1 Cx (k15 Sac SNi2+ − k −15 SacNi )
C16 [Co-Bu]+(l) Complexation −1 −1 1 Cx (k16 Sbu SCo2+ − k −16 SbuCo )
C17 [Co-Pro]+(l) Complexation −1 −1 1 Cx (k17 Spro SCo2+ − k −17 SproCo )
C18 [Co-Ac]+(l) Complexation −1 −1 1 Cx (k18 Sac SCo2+ − k −18 SacCo )
C19 [Fe-Ac]+(l) Complexation −1 −1 1 Cx (k19 Sac SFe2+ − k −19 SacFe )
S5: Stoichiometric matrix for EDTA complexation reactions.
Component 38 40 42 44 46 85 86 87 88 89 90 Process Rates

Process Sca2+ SMg2+ SNi2+ SCo2+ SFe2+ Sedta C1 C2 C3 C4 C5
C1 2-
[Ca-EDTA] (l) Complexation −1 1 C𝑥 (𝑘1 Sedta SCa2+ − k −1 SedtaCa )
C2 [Mg-EDTA]2-(l) Complexation −1 1 C𝑥 (𝑘2 Sedta SMg2+ − k −2 SedtaMg )
2-
C3 [Ni-EDTA] (l) Complexation −1 1 C𝑥 (𝑘3 Sedta SNi2+ − k −3 SedtaNi )
C4 [Co-EDTA]2-(l) Complexation −1 1 C𝑥 (𝑘4 Sedta SCo2+ − k −4 SedtaCo )
C5 [Fe-EDTA]2-(l) Complexation −1 1 C𝑥 (𝑘5 Sedta SFe2+ − k −5 SedtaFe )
S6: Model verification
In order to verify the model accuracy and prediction ability, the complexation submodule predictions were compared with the ones obtained through
Visual MINTEQ 3.1 (Gustafsson, J.P., 2018, https://vminteq.lwr.kth.se/download/). TE-complexation reactions with suitable initial concentrations of the
reacting species were considered in the Visual MINTEQ equilibrium solver. The ionic strength of the system was set to zero for all the predictions. In
order to compare model results with Visual MINTEQ, the simulation time was increased to get a steady state for all the variables considered in the
model. The model outputs were compared with the Visual MINTEQ predictions, confirming an accurate prediction of the TE complex formation. This
comparison holds good for all the complexation reactions taken into consideration.
Species [Ca-EDTA]2- [Co-EDTA]2- [Fe-EDTA]2- [Mg-EDTA]2- [Ni-EDTA]2- pH
Visual Minteq 2.0097E-12 9.5407E-07 7.4611E-09 2.7125E-14 9.0385E-06 9.774
Proposed Model 1.4589E-12 1.2732E-06 4.8761E-09 1.9874E-14 8.7219E-06 9.778
Species Concentration
Ca+2 0.00001
CaEDTA-2 2.0097E-12
Co+2 9.0459E-06
CoEDTA-2 9.5407E-07
EDTA-4 7.2967E-20
Fe+2 9.9925E-06
FeEDTA-2 7.4611E-09
H+1 1.6845E-10
H2EDTA-2 3.4439E-22
H3EDTA- 7.5946E-29
H4EDTA (aq) 2.0939E-36
H5EDTA+ 1.1154E-44
H6EDTA+2 1.4517E-54
HEDTA(ii)-3 1.0904E-18
Mg+2 9.7778E-06
MgEDTA-2 2.7125E-14
MgOH+ 2.2222E-07
Ni+2 9.6154E-07 Similarly,
NiEDTA-2 9.0385E-06
OH- 0.000059778
complexation
of TEs with VFAs has been verified taking into consideration each individual complexation reaction. Complexation reactions between Fe2+, Ni2+, Co2+
with acetate, propionate, valerate, butyrate as mentioned in Table 1 has been carried out. Complexation of acetate with Fe2+ has been presented here as
an example. The procedure of arriving at the verification remains the same, as mentioned earlier in this section.
Input Visual Minteq Proposed Model
Fe2+ (M) Acetate-1 (M) [Fe-Acetate]+ (M) pH [Fe-Acetate]+ (M) pH
1×10-3 1×10-3 2.393×10-5 10.997 2.393×10-5 11.00
1×10-4 1×10-3 6.275×10-7 4.284 6.269×10-7 4.28
1×10-3 1×10-4 2.444×10-6 11.276 2.444×10-6 11.27
1×10-4 1×10-4 2.499×10-7 9.997 2.499×10-7 10.00
S7: TE fractionation obtained from simulation studies.

CM1: EDTA:0.00 M; Ca and Mg: 0.001 M
Speciation (when total concentration of each TE is assumed to be 1 unit)
TE TES TECO3 (TE)3(PO4)2 TE2+ TE-Bu+ TE-Pro+ TE-Ac+ [TE-EDTA]2-
Fe 8.662×10-1 3.000×10-3 7.562×10-5 0.1301 N.A. N.A. 6.090×10-4 0.00
Co 8.618×10-1 7.400×10-4 0 0.13687 2.654×10-6 6.674×10-6 5.331×10-4 0.00
Ni 2.090×10-2 2.700×10-3 0 0.9712 2.371×10-5 5.4384e-05 4.987×10-3 0.00

Speciation (fraction)
Fe 9.596×10-1 2.550×10-2 8.052×10-5 0.0147 N.A. N.A. 1.620×10-4 0.00
Co 9.697×10-1 1.018×10-2 0 0.019881 7.615×10-7 1.822×10-6 1.830×10-4 0.00
Ni 4.190×10-2 2.740×10-2 5.102×10-8 0.91940 4.433×10-5 9.676×10-5 1.117×10-2 0.00

Fe 8.862×10-1 1.114×10-1 6.755×10-5 0.00227 N.A. N.A. 4.012×10-5 0.00
Co 9.371×10-1 5.670×10-2 0 0.006022 3.336×10-7 7.923×10-7 8.816×10-5 0.00
Ni 5.550×10-2 1.199×10-1 1.029×10-6 0.8087 5.640×10-5 1.221×10-4 1.560×10-2 0.00

Fe 6.993×10-1 2.988×10-1 5.040×10-5 0.001423 N.A. N.A. 2.38×10-4 0.00
Co 8.184×10-1 1.730×10-1 0 0.007402 3.863×10-6 9.088×10-6 1.02×10-3 0.00

Ni 5.380×10-2 3.212×10-1 1.815×10-6 0.5270 3.460×10-5 7.420×10-4 9.667×10-2 0.00

Fe 7.409×10-1 2.109×10-3 6.210×10-5 1.175×10-1 3.129×10-4 1.389×10-1
0 7.249×10-4
Co 9.403×10-5 4.485×10-4 3.2751×10-3 3.378×10-8 8.830×10-8 9.558×10-1
3.198×10-9
Ni 2.260×10-4 1.914×10-3 0 1.033×10-4 1.341×10-9 3.014×10-7 9.97×10-1
CM6: EDTA: 0.000025 M; Ca and Mg: 0.002 M

Fe 8.409×10-1 1.903×10-2 6.669×10-5 2.267×10-2 1.220×10-4 1.171×10-1
8.955×10-1
Co 9.701×10-2 6.539×10-3 0 7.045×10-4 1.105×10-8 2.847×10-8 3.154×10-6
0 2.308×10-5 4.558×10-10 1.071×10-9 1.362×10-7 9.818×10-1
Ni 9.136×10-4 1.716×10-2

Fe 8.091×10-1 8.453×10-2 5.674×10-5 1.181×10-1 9.403×10-7 9.438×10-2
Co 1.819×10-1 3.541×10-2 0 4.027×10-4 8.206×10-9 2.107×10-8 2.667×10-6 7.822×10-1
7.611×10-2 1.481×10 -5
3.800×10 -10
9.217×10-1
Ni 2.111×10-3 1.769×10-8 8.902×10-10 1.293×10-7

Fe 6.466×10-1 2.480×10-1 4.191×10 -5
9.684×10 -3
8.261×10-5 9.544×10-2
Co 1.929×10-1 1.233×10-1 0 2.900×10-4 6.551×10-9 1.718×10-8 2.062×10-6 6.833×10-1
2.284×10-1 1.100×10-7 7.691×10-1
Ni 2.325×10-3 1.346×10-7 1.176×10-5 3.338×10-10 7.986×10-10

Fe 5.335×10-1 1.394×10-3 4.111×10-5 9.123×10 -2
1.900×10-4 3.735×10-1
2.414×10-4 9.738×10-4 7.809×10-9 1.688×10-6
Co 1.025×10-2 0 2.063×10-8 9.885×10-1
7.086×10-10 9.986×10-1
Ni 4.658×10-5 1.282×10-3 0 2.913×10-5 2.941×10-10 6.658×10-8

Fe 6.349×10-1 1.303×10-2 4.579×10-5 2.729×10-2 1.011×10-4 3.245×10-1
1.008×10-6 9.627×10-1
Co 3.323×10-2 3.682×10-3 0 3.271×10-4 3.482×10-9 9.134×10-9
9.983×10-6 1.337×10-10 3.200×10-10 9.884×10-1
Ni 2.900×10-4 1.124×10-2 0 4.055×10-8

Fe 6.442×10-1 5.925×10-2 4.039×10 -5
1.761×10 -2
9.272×10-5 2.787×10-1
2.344×10-4 8.114×10-9 1.026×10-6
Co 6.582×10-2 1.934×10-2 3.105×10-9 9.145×10-1
0
7.492×10-6
Ni 6.449×10-4 4.614×10-2 0 1.249×10-10 2.977×10-10 4.324×10-8 9.532×10-1

Fe 5.212×10-1 1.952×10-1 3.000×10 -5
1.459×10 -2
8.240×10-5 2.687×10-1
8.602×10-2 0 7.047×10-9 8.786×10-7
Co 7.308×10-2 1.87×10-4 2.678×10-9 8.407×10-1
6.365×10-6 1.147×10-10 2.753×10-10 8.417×10-1
Ni 7.350×10-2 1.574×10-2 7.965×10-8 3.940×10-8
CM13: EDTA:0.000075M; Ca and Mg: 0.001 M
2+
TE TES TECO3 (TE)3(PO4)2 TE TE-Bu+ TE-Pro+ TE-Ac+ [TE-EDTA]2-
Fe 3.311×10-1 8.588×10-4 2.234×10 -5
6.483×10 -2
1.170×10-4 6.030×10-1
1.185×10-4 0 3.016×10-9 6.509×10-7 9.958×10-1
Co 3.588×10-3 4.310×10-4 8.004×10-9
1.275×10-5 1.123×10-10 2.538×10-8
Ni 1.316×10-5 7.700×10-4 0 2.718×10-10 9.992×10-1

2+
Fe 4.310×10-1 7.858×10-4 2.688×10 -5
2.860×10 -2
8.638×10-5 5.324×10-1
0 4.887×10-9 5.3664×10-7
Co 1.112×10-2 1.713×10-3 2.130×10-4 1.851×10-9 9.869×10-1
6.276×10-3 6.949×10-11 1.673×10-10
Ni 7.793×10-5 0 6.368×10-6 2.109×10-10 9.936×10-1

Fe 4.726×10-1 3.613×10-2 2.521×10-5 2.072×10-2 8.672×10-6 4.703×10-1
8.316×10-2 0 1.807×10-9 4.757×10-9 9.707×10-1
Co 2.079×10-2 1.725×10-4 6.006×10-7
0 5.230×10-6 1.655×10-10 2.399×10-8 9.771×10-1
Ni 1.510×10-4 2.265×10-2 6.895×10-11

Fe 3.886×10-1 1.426×10-1 1.908×10 -5
1.740×10 -2
7.799×10-5 4.511×10-1
6.100×10-2 0 4.225×10-9 5.319×10-7
Co 2.918×10-2 1.426×10-4 1.597×10-9 9.096×10-1
1.548×10-10 8.917×10-1
Ni 2.720×10-4 1.079×10-1 5.596×10-8 4.555×10-6 6.420×10-11 2.238×10-8
2+
Fe 1.430×10-1 4.580×10-4 7.311×10 -6
3.996×10 -2
6.771×10-5 8.16483×10-1
Co 6.575×10-2 6.433×10-5 0 1.960×10-4 1.291×10-9 3.433×10-9 2.772×10-7 9.988×10-1
0 5.792×10-6 4.788×10-11 1.160×10-10
Ni 1.581×10-6 4.030×10-4 1.076×10-8 9.995×10-1

2+
Fe 2.385×10-1 3.818×10-3 1.147×10 -6
2.686×10 -2
7.340×10-5 7.306×10-1
1.474×10-4 3.065×10-9 9.964×10-1
Co 2.629×10-3 7.458×10-3 0 1.157×10-9 3.351×10-7
4.355×10-6 4.305×10-11 1.039×10-10 1.304×10-8 9.972×10-1
Ni 8.672×10-6 2.776×10-3 0

Fe 3.038×10-1 1.678×10-1 1.255×10-5 2.167×10-2 8.056×10-7 6.575×10-1
2.960×10-3 0 1.310×10-4 1.222×10-9 3.228×10-9 4.070×10-7 9.917×10-1
Co 5.172×10-3
Ni 9.97×10-1 8.477×10-3 0 3.913×10-6 4.572×10-11 1.101×10-10 1.594×10-8 9.914×10-1

Fe 2.581×10-1 9.827×10-2 1.040×10-5 1.801×10-2 7.165×10-5 6.255×10-1
1.095×10-4 2.867×10-9 3.624×10-7 9.381×10-1

Co 1.367×10-2 4.802×10-2 0 1.080×10-9
3.410×10-6 1.025×10-10 1.487×10-8 9.187×10-1
Ni 1.3×10-4 8.106×10-2 3.850×10-8 4.236×10-11
S.8 ADM1 based TE complexation model parameters.
Parameter Name Value Unit

Vliq Liquid volume of the reactor 1.0 L
kdis Kinetic parameter for disintegration 0.5 d-1
khydch Kinetic parameter for carbohydrate hydrolysis 10 d-1
khydpr Kinetic parameter for protein hydrolysis 10 d-1
khydli Kinetic parameter for lipid hydrolysis 10 d-1
kmsu Monod specific maximum uptake rate for sugars 30 d-1
Kssu Half saturation value for sugars 0.5 g COD L-1
kmaa Monod specific maximum uptake rate for amino acids 50 d-1
Ksaa Half saturation value for amino acids 0.3 g COD L-1
kmfa Monod specific maximum uptake rate for fatty acids 6 d-1
Ks_fa Half saturation value for fatty acids 0.4 g COD L-1
kmc4 Monod specific maximum uptake rate for c4 20 d-1
Ksc4 Half saturation value for c4 0.2 g COD L-1
kmpro Monod specific maximum uptake rate for propionate 13 d-1
Kspro Half saturation value for propionate 0.1 g COD L-1
kmac Monod specific maximum uptake rate for acetate 8 d-1
Ksac Half saturation value for acetate 0.15 g COD L-1
kmh2 Monod specific maximum uptake rate for hydrogen 35 d-1
Ksh2 Half saturation value for hydrogen 7×10-6 g COD L-1
kdec_Xsu First order decay rate for sugar degraders 0.02 d-1
kdec_Xaa First order decay rate for amino acid degraders 0.02 d-1
kdec_Xfa First order decay rate for fatty acids degraders 0.02 d-1
kdec_Xc4 First order decay rate for butyrate and valerate degraders 0.02 d-1
kdec_Xpro First order decay rate for propionate degraders 0.02 d-1
kdec_Xac First order decay rate for acetate degraders 0.02 d-1
kdec_Xh2 First order decay rate for hydrogen degraders 0.02 d-1
Y_su Yield of sugar degraders on sugars 0.100 -
Y_aa Yield of amino acids degraders on amino acids 0.080 -
Y_fa Yield of fatty acids degraders on fatty acids 0.060 -
Y_c4 Yield of valerate and butyrate degraders on valerate and butyrate 0.060 -
Y_pro Yield of propionate degraders on propionate 0.040 -
Y_ac Yield of acetate degraders on acetate 0.050 -

Y_h2 Yield of hydrogen degraders on hydrogen 0.060 -
f_sI.xc Yield of soluble inert on complex matter 0.100 -
f_xI.xc Yield of particulate inert on complex matter 0.25 -
f_ch.xc Yield of carbohydrate on complex matter 0.20 -
f_pr.xc Yield of protein on complex matter 0.20 -
f_li.xc Yield of lipid on complex matter 0.25 -
f_hs.xc Yield of sulfide on complex matter 0.0006 -
f_hpo4.xc Yield of phosphate on complex matter 0.00006 -
f_fa.li Yield of fatty acid on lipid 0.950 -
f_bu.su Yield of butyrate on sugars 0.130 -
f_pro.su Yield of propionate on sugars 0.270 -
f_ac.su Yield of acetate on sugars 0.410 -
f_h2.su Yield of hydrogen on sugars 0.190 -
f_va.aa Yield of valerate on amino acids 0.230 -
f_bu.aa Yield of butyrate on amino acids 0.260 -
f_pro.aa Yield of propionate on amino acids 0.050 -
f_ac.aa Yield of acetate on amino acids 0.400 -
f_h2.aa Yield of hydrogen on amino acids 0.060 -
C_su Carbon content of sugars 0.03130 mole C (g COD)-1
C_aa Carbon content of amino acids 0.030 mole C (g COD)-1
C_bu Carbon content of butyrate 0.0250 mole C (g COD)-1
C_pro Carbon content of propionate 0.02680 mole C (g COD)-1
C_ac Carbon content of acetate 0.03130 mole C (g COD)-1
C_va Carbon content of valerate 0.0240 mole C (g COD)-1
C_bac Carbon content of biomass 0.03130 mole C (g COD)-1
C_ch4 Carbon content of methane 0.01560 mole C (g COD)-1
N_bac Nitrogen content of biomass 0.08/14 mole N (g COD)-1
N_aa Nitrogen content of amino acids 0.0070 mole N (g COD)-1
kla_h2 Gas-Liquid transfer coefficient for hydrogen 3.84 d-1
kla_met Gas-Liquid transfer coefficient for methane 2.16 d-1
kla_co2 Gas-Liquid transfer coefficient for carbon dioxide 60 d-1
KH_h2 Henry’s law coefficient for hydrogen 0.00078 kmole bar-1
KH_ch4 Henry’s law coefficient for methane 0.00116 kmole bar-1
KH_co2 Henry’s law coefficient for carbon dioxide 0.035 kmole bar-1

T Temperature 308.16 K
R Gas law constant 0.08314 bar kmole-1 K-1
Vgas Volume of gas 0.25 L
Pgaslim Limit of gas 0
KA Acid-base kinetic parameter for valerate 1010 M-1d-1
Ka Kinetic equilibrium coefficient for valerate 10-4.84 M
KB Acid-base kinetic parameter for butyrate 1010 M-1d-1
Kb Kinetic equilibrium coefficient for butyrate 10-4.82 M
KPRO Acid-base kinetic parameter for propionate 1010 M-1d-1
Kpro Kinetic equilibrium coefficient for propionate 10-4.88 M
KAC Acid-base kinetic parameter for acetate 1010 M-1d-1
Kac Kinetic equilibrium coefficient for acetate 1.8×10-5 M
KCO Acid-base kinetic parameter for bicarbonate 1010 M-1d-1
Kco Kinetic equilibrium coefficient for bicarbonate 10-6.35 M
KNI Acid-base kinetic parameter for ammonia 1010 M-1d-1
kni Kinetic equilibrium coefficient for ammonia 5.6×10-10 M
KCO3 Acid-base kinetic parameter for carbonate 1010 M-1d-1
Kco3 Kinetic equilibrium coefficient for carbonate 4.8×10-11 M
kh3po4 Kinetic equilibrium coefficient for H3PO4- 6.9×10-3 M
KH3PO4 Acid-base kinetic parameter for H3PO4- 1010 M-1d-1
khpo4 Kinetic equilibrium coefficient for HPO42- 6.2×10-8 M
KHPO4 Acid-base kinetic parameter for HPO42- 1010 M-1d-1
kpo4 Kinetic equilibrium coefficient for PO43- 4.8×10-13 M
KPO4 Acid-base kinetic parameter for PO43- 1010 M-1d-1
Pr_CaCO3 Precipitation kinetic parameter for CaCO3 102 M-1d-1
Ksp_CaCO3 Solubility product for CaCO3 8.48 -
Dis_CaCO3 Dissolution kinetic parameter for CaCO3 102 M-1d-1
Pr_MgCO3 Precipitation kinetic parameter for MgCO3 102 M-1d-1
Ksp_MgCO3 Solubility product for MgCO3 7.46 -
Dis_MgCO3 Dissolution kinetic parameter for MgCO3 102 M-1d-1
Ksp_NiCO3 Precipitation kinetic parameter for NiCO3 11.2 -
Pr_NiCO3 Solubility product for NiCO3 102 M-1d-1
Dis_NiCO3 Dissolution kinetic parameter for NiCO3 102 M-1d-1
Ksp_CoCO3 Solubility product for CoCO3 11.2 -

Pr_CoCO3 Precipitation kinetic parameter for CoCO3 102 M-1d-1
Dis_CoCO3 Dissolution kinetic parameter for CoCO3 102 M-1d-1
Ksp_FeCO3 Solubility product for FeCO3 10.59 -
Pr_FeCO3 Precipitation kinetic parameter for FeCO3 102 M-1d-1
Dis_FeCO3 Dissolution kinetic parameter for FeCO3 102 M-1d-1
Ksp_Ni3PO4_2 Solubility product for Ni3(PO4)2 31 -
Pr_Ni3PO4_2 Precipitation kinetic parameter for Ni3(PO4)2 102 M-1d-1
Dis_Ni3PO4_2 Dissolution kinetic parameter for Ni3(PO4)2 102 M-1d-1
Ksp_Fe3PO4_2 Solubility product for Fe3(PO4)2 37.76 -
Pr_Fe3PO4_2 Precipitation kinetic parameter for Fe3(PO4)2 102 M-1d-1
Dis_Fe3PO4_2 Dissolution kinetic parameter for Fe3(PO4)2 102 M-1d-1
Ksp_Co3PO4_2 Solubility product for Co 3(PO4)2 34 -
Pr_Co3PO4_2 Precipitation kinetic parameter for Co 3(PO4)2 102 M-1d-1
Dis_Co3PO4_2 Dissolution kinetic parameter for Co 3(PO4)2 102 M-1d-1
Ksp_Ca3PO4_2 Solubility product for Ca3(PO4)2 28.4 -
Pr_Ca3PO4_2 Precipitation kinetic parameter for Ca3(PO4)2 102 M-1d-1
Dis_Ca3PO4_2 Dissolution kinetic parameter for Ca3(PO4)2 102 M-1d-1
Ksp_NiS Solubility product for NiS 20.7 -
Pr_NiS Precipitation kinetic parameter for NiS 102 M-1d-1
Dis_NiS Dissolution kinetic parameter for NiS 102 M-1d-1
Ksp_CoS Solubility product for CoS 20.3 -
Pr_CoS Precipitation kinetic parameter for CoS 102 M-1d-1
Dis_CoS Dissolution kinetic parameter for CoS 102 M-1d-1
Ksp_FeS Solubility product for FeS 18 -
Pr_FeS Precipitation kinetic parameter for FeS 102 M-1d-1
Dis_FeS Dissolution kinetic parameter for FeS 102 M-1d-1
Ksp_MgNH4PO4 Solubility product for MgNH4PO4 13.26 -
Pr_MgNH4PO4 Precipitation kinetic parameter for MgNH4PO4 102 M-1d-1
Dis_MgNH4PO4 Dissolution kinetic parameter for MgNH4PO4 102 M-1d-1
pH_LL1 pH limitation 4.0 -
pH_UL1 pH limitation 5.5 -

KH6EDTA Acid-base kinetic parameter for H6EDTA2+ 1010 M-1d-1
kh6edta Kinetic equilibrium coefficient for H6EDTA2+ 100.0 M
KH5EDTA Acid-base kinetic parameter for H5EDTA+ 1010 M-1d-1
kh4edta Kinetic equilibrium coefficient for H5EDTA+ 10-1.5 M
KH4EDTA Acid-base kinetic parameter for H4EDTA 1010 M-1d-1
kh4edta Kinetic equilibrium coefficient for H4EDTA 10-2.0 M
KH2EDTA Acid-base kinetic parameter for H2EDTA2- 1010 M-1d-1
kh2edta Kinetic equilibrium coefficient for H2EDTA2- 10-2.6 M
KHEDTA Acid-base kinetic parameter for HEDTA3- 1010 M-1d-1
khedta Kinetic equilibrium coefficient for HEDTA3- 10-6.13 M
KEDTA Acid-base kinetic parameter for EDTA4- 1010 M-1d-1
kedta Kinetic equilibrium coefficient for EDTA4- 10-10.37 M
C_[Ca-EDTA]2- Complexation kinetic parameter for [Ca-EDTA]2- 1010 M-1d-1
C_[Mg-EDTA]2- Complexation kinetic parameter for [Mg-EDTA]2- 1010 M-1d-1
C_[Fe-EDTA]2- Complexation kinetic parameter for [Fe-EDTA]2- 1010 M-1d-1
C_[Ni-EDTA]2- Complexation kinetic parameter for [Ni-EDTA]2- 1010 M-1d-1
C_[Co-EDTA]2- Complexation kinetic parameter for [Co-EDTA]2- 1010 M-1d-1
C_[Ca-Va]+ Complexation kinetic parameter for [Ca-Va]+ 1010 M-1d-1
C_[Ca-Bu]+ Complexation kinetic parameter for [Ca-Bu]+ 1010 M-1d-1
C_[Ca-Pro]+ Complexation kinetic parameter for [Ca-Pro]+ 1010 M-1d-1
C_[Ca-Ac]+ Complexation kinetic parameter for [Ca-Ac]+ 1010 M-1d-1
C_[Mg-Bu]+ Complexation kinetic parameter for [Mg-Bu]+ 1010 M-1d-1
C_[Mg-Pro]+ Complexation kinetic parameter for [Mg-Pro]+ 1010 M-1d-1
C_[Mg-Ac]+ Complexation kinetic parameter for [Mg-Ac]+ 1010 M-1d-1
C_[Ni-Bu]+ Complexation kinetic parameter for [Ni-Bu]+ 1010 M-1d-1
C_[Ni-Pro]+ Complexation kinetic parameter for [Ni-Pro]+ 1010 M-1d-1
C_[Ni-Ac]+ Complexation kinetic parameter for [Ni-Ac]+ 1010 M-1d-1
C_[Co-Bu]+ Complexation kinetic parameter for [Co-Bu]+ 1010 M-1d-1
C_[Co-Pro]+ Complexation kinetic parameter for [Co-Pro]+ 1010 M-1d-1
C_[Co-Ac]+ Complexation kinetic parameter for [Co-Ac]+ 1010 M-1d-1
C_[Fe-Ac]+ Complexation kinetic parameter for [Fe-Ac]+ 1010 M-1d-1

K_[Ca-EDTA]2- Complexation stability constant for [Ca-EDTA]2- 10.7 -
K_[Mg-EDTA]2- Complexation stability constant for [Mg-EDTA]2- 8.7 -
K_[Ni-EDTA]2- Complexation stability constant for [Fe-EDTA]2- 20 -
K_[Co-EDTA]2- Complexation stability constant for [Ni-EDTA]2- 18 -
K_[Fe-EDTA]2- Complexation stability constant for [Co-EDTA]2- 16 -
K_[Ca-Va]+ Complexation stability constant for [Ca-Va]+ 0.3 -
K_[Ca-Bu]+ Complexation stability constant for [Ca-Bu]+ 0.94 -
K_[Ca-Pro]+ Complexation stability constant for [Ca-Pro]+ 0.93 -
K_[Ca-Ac]+ Complexation stability constant for [Ca-Ac]+ 1.18 -
K_[Mg-Bu]+ Complexation stability constant for [Mg-Bu]+ 0.96 -
K_[Mg-Pro]+ Complexation stability constant for [Mg-Pro]+ 0.97 -
K_[Mg-Ac]+ Complexation stability constant for [Mg-Ac]+ 1.26 -
K_[Ni-Bu]+ Complexation stability constant for [Ni-Bu]+ 0.691 -
K_[Ni-Pro]+ Complexation stability constant for [Ni-Pro]+ 1.19 -
K_[Ni-Ac]+ Complexation stability constant for [Ni-Ac]+ 1.44 -
K_[Co-Bu]+ Complexation stability constant for [Co-Bu]+ 0.591 -
K_[Co-Pro]+ Complexation stability constant for [Co-Pro]+ 1.13 -
K_[Co-Ac]+ Complexation stability constant for [Co-Ac]+ 1.38 -
K_[Fe-Ac]+ Complexation stability constant for [Fe-Ac]+ 1.40 -
S9: Initial concentration of the various dynamic state variables considered in the model.
EDTA with optimal EDTA with Variable Ca/Mg

State No Variable TEs sub-optimal Unit
TEs
1 Ssu 0 0 0 g COD L-1

2 Saa 0 0 0 g COD L-1
3 Sfa 0 0 0 g COD L-1
4 Sva 0 0 0 g COD L-1
5 Sbu 0 0 0 g COD L-1
6 Spro 0 0 0 g COD L-1
7 Sac 0 0 0 g COD L-1
8 Sh2 0 0 0 g COD L-1
9 Sch4 0 0 0 g COD L-1
10 Sco2 0 0 0 g COD L-1
11 Snh3 0 0 0 g COD L-1
12 SI 0 0 0 g COD L-1
13 Xc 2 2 2 g COD L-1
14 Xch 0 0 0 g COD L-1
15 Xpr 0 0 0 g COD L-1
16 Xli 0 0 0 g COD L-1
17 Xsu 0.12 0.12 0.12 g COD L-1
18 Xaa 0.12 0.12 0.12 g COD L-1
19 Xfa 0.12 0.12 0.12 g COD L-1
20 Xc4 0.12 0.12 0.12 g COD L-1
21 Xpro 0.12 0.12 0.12 g COD L-1
22 Xac 0.12 0.12 0.12 g COD L-1
23 Xh2 0.12 0.12 0.12 g COD L-1
24 XI 0 0 0 g COD L-1
31 Shva 0 0 0 g COD L-1
32 Shbu 0 0 0 g COD L-1
33 Shpro 0 0 0 g COD L-1
34 Shac 0 0 0 g COD L-1
35 Shco3 0 0 0 M
36 Snh4 0.006 0.006 0.006 M
37 Sco3 0.004 0.004 0.004 M
38 SCa 0.002 0.002 Variable* M
40 SMg 0.0025 0.0025 Variable* M
42 SNi 0.00001 0.000001 0.00001 M
44 SCo 0.0000001 0.00000001 0.0000001 M
46 SFe 0.0001 0.00001 0.0001 M
51 Spo4 0.006 0.006 0.006 M
58 Ss2 0.001 0.001 0.001 M
71 Sedta4- Variable* Variable* Variable* M
ADM1-based mathematical model of trace elements speciation in
anaerobic digestion systems
Bikash Chandra Maharaja,b,d*, Maria Rosaria Matteib, Luigi Frunzob, Eric D. van Hullebuschc,d,e
and Giovanni Esposito a
a
University of Cassino and the Southern Lazio, Department of Civil and Mechanical
Engineering, Cassino, Italy

b
University of Naples Federico II, Department of Mathematics and Applications "Renato
Caccioppoli", Naples, Italy

c
Université Paris-Est, Laboratoire Géomatériaux et Environnement (LGE), EA 4508, UPEM,
77454 Marne-la-Vallée, France

d
IHE Delft Institute for Water Education, Department of Environmental Engineering and Water
Technology, Delft, The Netherlands

e
Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Université Paris Diderot, UMR
7154, CNRS, F-75005 Paris, France
*Corresponding Author:
Department of Civil and Mechanical Engineering,
University of Cassino and Southern Lazio,
Via di Basio 43, 03043 Cassino (FR), Italy
E-mail address: [email protected]

Abstract:
Due to the multiplicity of physicochemical processes taking place in AD systems and limitations
of the available analytical techniques, assessing the bioavailability of TEs is a challenging task.
Determination of TE speciation can be partly accomplished by developing a mathematical model
able to consider the variety of physicochemical processes affecting TEs dynamics. To this aim, a
new model based on the anaerobic digestion model no.1 (ADM1) approach has been proposed to
simulate TEs adsorption-desorption reactions with biomass, inert and mineral precipitates, as
well as TE precipitation/dissolution, complexation reactions and biodegradation processes. The
developed model was integrated numerically, and numerical simulations have been run to
investigate the model behavior. In particular, the simulation scenarios predict the effect of
organic matter concentration, initial TE concentration, initial Ca-Mg concentration, initial EDTA
concentration and change in TE binding site density for biomass, inert and precipitate, on
cumulative methane production and TE speciation.
Keywords: Anaerobic digestion, Trace elements; Mathematical modeling; ADM1;
Physicochemical processes; Adsorption.

1. Introduction
In recent years, anaerobic digestion (AD) of organic wastes has gained attention as an alternative
source of sustainable energy (Appels et al., 2011). Efforts have been devoted to improve the
biogas production by recruiting process optimization and scale up techniques (Appels et al.,
2008; Fermoso et al., 2009; Li et al., 2011; Nguyen et al., 2015). Dosing of trace elements (TEs)
in AD is one such avenues which has attracted attention of researchers (Choong et al., 2016;
Thanh et al., 2015; Zandvoort et al., 2006). This is primarily due to the sensitivity of the
microbial community to TEs, in particular acetogens and methanogens. Indeed, the synchrony
between acetogens and methanogens is largely attributed to optimum availability and supply of
TEs in the surrounding. Any lack or excess of TEs in the reactor leads to acidification and
toxicity respectively. Moreover, the complex biogeochemistry of TEs and the increasing
environmental concern over heavy metal misuse generating from digestate, demands a thorough
analysis regarding the fate of TEs in AD systems (Fermoso et al., 2015).
TEs are structural components of important enzymes involved in AD biochemical processes
(Mudhoo and Kumar, 2013). Fe is a structural component of the majority of enzymes in AD
pathways. Also, it serves as a redox carrier. Ni is a part of coenzyme F430 ring structure of
methyl reductase which facilitates reductions of methyl co-enzyme M to methane during
methanogenesis. Co is essential to both acetoclastic methanogens and hydrogenotrophic
methanogens. It is an integral part of coronoid structure of vitamin B12 which binds to coenzyme
methylase (Thanh et al., 2015). These TEs influence the chemical equilibrium of the AD
bioreactors by forming precipitates, complexes and by adsorbing on the particulate components
constituting the anaerobic system (Arican et al., 2002; Fermoso et al., 2015; Van Hullebusch et
al., 2006; Worms et al., 2006). Due to the significant role played by TEs in AD systems, the
difficulty in measuring the bioavailable TE fractions, mainly due to low resolution of
spectroscopy methods, as well as the intricate biogeochemistry, developing a mathematical
model able to consider the complex dynamics of TEs in anaerobic environments is necessary
(van Hullebusch et al., 2016).
The bioavailable fraction of TEs in an AD system is a function of the biouptake, precipitation,
complexation and adsorption processes (Worms et al., 2006; Yekta et al., 2016). Proper
availability of TEs (i.e. Fe, Ni, Co, Se, Mn, Zn and Mo) in an AD system is necessary for an
efficient methane production (van Hullebusch et al., 2016). Microbial growth depends on the
uptake and the transport of these essential TEs from their local milieu. As a result, there is a
necessity to maintain a hydrophilic interface. The latter is achieved by having a large outer
surface area and projecting anionic organic ligands on the cell surface. These anionic organic
ligands become point of adsorption-desorption for a facilitated diffusion of TEs for the process
of TEs internalization. It has been reported that the adsorption-desorption processes of TEs, on
the cell surface and the available particulate matter (i.e. mineral precipitates and inert), strongly
affect the availability of TEs in an AD system (van Hullebusch et al., 2016).
The main objective of this work is to develop an ADM1 based mathematical model to: 1)
mechanistically describe all the main biochemical and physicochemical processes involved in an
AD system, including adsorption of TEs (on inert, biomass and precipitates), EDTA/VFA
complexation, precipitation/dissolution and uptake of TE; 2) investigate into the speciation of
TEs in AD, in particular the fractionation of TEs among adsorbed species, precipitates and
dissolved complexes in relation to the change in initial concentration of complex organic matter,
TEs, calcium and magnesium and EDTA. To this aim, new TE-adsorption reactions have been
defined and added to the ADM1-based model developed in (Maharaj et al., 2019). Hydrolysis,
acidogenesis, acetogenesis, methanogenesis and acid-base equilibrium have been combined to
TE-adsorption, complexation and precipitation/dissolution and the effects on total methane
production have been evaluated.
2. Methods: Model Development and Simulation
2.1 The Mathematical model
The proposed model simulates the speciation of TEs in AD processes. Adsorption, precipitation
and complexation, as three primary TE fluxes, are analyzed in conjugation with methane
production during the anaerobic biodegradation of complex organic matter. Overall, the
proposed model tracks the dynamics of 175 state variables which constitute the components of
the proposed AD model. For each component considered in the liquid, gas and solid phase the
general mass balance can be written as:
m5
dVliq Si
= qin Sin,i − qout Si +Vliq (∑ δi,j ρA,j (t, 𝐒) − ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ))
dt
𝑗=1
m1 m2
+ Vliq (∑ αi,j ρbio,j (t, 𝐒, 𝐗) + ∑ βi,j ρcmplx,j (t, 𝐒)) (1)

j=1 j=1
m3 m3 m4
̅ 𝐚𝐝𝐬 )),
+Vliq (∑ γi,j ρprec,j (t, 𝐒, 𝐗 𝐏 ) + ∑ γi,j ρdissol,j (t, 𝐒, 𝐗 𝐏 ) + ∑ εi,j ρs,j (t, 𝐒, 𝐗 𝐚𝐝𝐬 , 𝐗
j=1 j=1 j=1
i = 1, … , n1 , t > 0,
m1
dVliq Xi
dt (2)
j=1
i = n1 + 1, … , n2 , t > 0,
dSgas,i Sgas,i qgas

dt Vgas (3)
i = 1, … , n1 , t > 0,
dVliq Xp,i
= qin Xpin − qout Xp,i + Vliq (𝛼̅𝑖 ρprec,i (t, 𝐒, 𝐗 𝐏 ) + β̅𝑖 ρdissol,i (t, 𝐒, 𝐗 𝐩 )),
dt (4)
i = 1, … , m3 , t > 0,
where:
n1 is the number of soluble components;
n2 − n1 is the number of particulate components, including complex organic matter and inert;
m1 is the number of biochemical processes considered;
m2 is the number of complexation processes considered;
m3 is the number of precipitation/dissolution processes considered;
m4 is the number of adsorption/desorption processes considered;
m5 is the number of acid-base reactions considered;
αi,j is the stoichiometric coefficient of the ith species on biochemical process j;
βi,j is the stoichiometric coefficient of the ith species on complexation process j;
γi,j is the stoichiometric coefficient of the ith species on precipitation/dissolution process j;
δi,j is the stoichiometric coefficient of the ith species on acid-base reaction j;
εi,j is the stoichiometric coefficient of the ith species on adsorption/desoprtion reaction j;
̅𝑖 and β̅𝑖 are the stoichiometric coefficients for the precipitate formation and dissolution;
α
Si is the ith soluble component, 𝐒 = (S1 , … , Sn1 );
Xi is the ith particulate component, 𝐗 = (Xn+1 , … , Xn2 );
Xp,i is the ith precipitate, 𝐗 𝐩 = (X p,1 , … , Xp,m3 );
Xads,i is the concentration of free binding sites for a specific adsorbent component, 𝐗 𝐚𝐝𝐬 =
(Xads,1 , … , Xads,n3 );
̅ ads,i is the concentration of occupied binding sites for a specific adsorbent component, 𝐗
X ̅ 𝐚𝐝𝐬 =
̅ ads,1 , … , X
(X ̅ ads,n );
3
n3 is the number of adsorbing components, including microbial species, inert and precipitates
(n2 − n1 + m3 );
Sgas,i is the ith soluble gas component, 𝐒𝐠𝐚𝐬 = (Sgas,1 , … , Sgas,n3 );
ρA,j (t, 𝐒) is the process rate for acid base reaction j;
ρT,i (t, 𝐒, 𝐒𝐠𝐚𝐬 ) is the gas transfer process rate for the ith soluble component;
ρbio,j (t, 𝐒, 𝐗) is the rate for the jth biochemical process;
ρcomplx,j (t, 𝐒, 𝐗) is the rate of the jth complexation process;
ρprec,j (t, 𝐒, 𝐗 𝒑 ) is the rate for the jth precipitation process;
ρdissol,j (t, 𝐒, 𝐗 𝒑 )is the rate for the jth dissolution process;
̅ 𝐚𝐝𝐬 ) is the rate for the jth sorption/desorption process;
ρs,j (t, 𝐒, 𝐗 𝐚𝐝𝐬 , 𝐗
Vliq and Vgas are the liquid and gas volume of the reactor.
Before explaining the adsorption module of the proposed model, it is imperative to introduce the
basic model features which entail the ADM1 framework. The model description here follows the
same structure of the original ADM1 report (Batstone et al., 2002). The TEsorption module will
be detailed in an appropriate section. The model considers a three phase AD system (solid-
liquid-gas). Reactions occurring in the liquid to solid phase are central to the TE adsorption
module and they can be classified in three groups: adsorption on biomass, inert and precipitates.
Protonation/deprotonation and TE-complexation with VFAs and EDTA have been considered in
the complexation module (Maharaj et al., 2019). Precipitation and redissolution taking place in
the solid phase and have been modeled according to (Maharaj et al., 2018). H2, CH4, CO2 and
H2S are considered in the gas phase.
In general, organic complex matter undergoes a series of kinetically controlled biological
degradation reactions to produce methane. The methane thus produced in the liquid phase strips
into the gas phase. The intermediates of the degradation processes include carbohydrates,
proteins, lipids, monosaccharides, amino acids and LCFAs and VFAs. The main
features/processes of the proposed model are detailed in the following lines:

 Liquid-gas processes: The model considers liquid-gas stripping processes for gaseous
components formed during the biological degradation of organic complex matter. The
model considers H2, CH4, CO2 and H2S gas components as significant intermediates and
have strong effect on biological processes. The concentration of biogas components have
been estimated by solving gas transfer equations together with other biochemical and
physicochemical processes.
 Liquid-Liquid processes: In addition to the biochemical reactions occurring in the liquid
phase, the model considers complexation reactions occurring between synthetic/organic
ligands and TEs. Protonation/deprotonation of VFAs, phosphate, carbonate, ammonium
and sulfide species in AD have been considered and implemented as differential
equations (Maharaj et al., 2019). The corresponding pka values have been considered for
each protonation/deprotonation reaction. The charge balance has been modified to
include the effect of the new ion pairing species.
 Liquid-Solid processes: The model considers the sorption/desoprtion of TEs on biomass,
inert and precipitates. Additionally, precipitation/dissolution of TEs such as Fe, Ni, Co,
as well as Ca and Mg with carbonate, phosphate and sulfide have been considered. A
concentration dependent second order kinetic model has been adopted for the
precipitation/dissolution reactions (Maharaj et al., 2018).
 Dose-response function: The effect of TEs on biological processes has been modelled
based on a dose-response function. A multiplicative inhibition factor𝐼𝑀𝑒 2+ has been
introduced to explicitly take into account the multiple effects of TEs on microbial
metabolism. This factor has been modeled according (Maharaj et al., 2019).
 Phosphorus and sulfur: Complex organic matter has been also considered as a source for
phosphorous, sulfur and TEs which arise in the system from the disintegration step.
Phosphorous has been supposed to be released as HPO43-, which represents the most
abundant form in the pH range of 6 to 14. Similarly, the released sulfur has been
considered in the form of HS - which has been depicted as the most abundant form in the
pH range of 6 to 12. Furthermore, the uptake of TEs has been linked to the acetate and
hydrogen uptake. The stoichiometry of the related processes has been updated.
2.1.1 Sorption/desorption Module
TE sorption/desorption has been modeled using a kinetic approach. Specifically, a reversible
second order kinetic model has been considered for sorption (forward) and sorption (backward)
reactions. Three types of adsorption surfaces have been considered: 1) Biomass; 2) Inerts; and 3)
Precipitates. For what concerns the mineral precipitates, FeS has been recognized as the
component playing the most crucial role in sorption processes (Van Hullebusch et al., 2006;
Watson et al., 1995). Seven biomass species, as originally formulated in the ADM1, are
construed to participate in the sorption/desorption reactions with all the five metals considered
(TEs: Fe, Ni and Co; Light metals: Ca and Mg). Inerts have been construed to take part in
sorption/sorption reaction with Fe, Ni, Ca, Mg and Ca, while, only Co and Ni have been
considered to participate with precipitates (FeS). Thus, arriving at a system which consists of 35
sorption/desorption reactions for biomass, 5 sorption/desorption reactions for inert and 2 for
sorption/sorption reactions on FeS. sorption/desorption kinetics
The sorption/desorption processes are kinetically controlled and can be represented as follows:
𝑘1
𝑀𝑒 2+ + 𝑀𝑒𝑆 → (𝑀𝑒 ≡ 𝑀𝑒𝑆) (5)
𝑘−1
(𝑀𝑒 − 𝑀𝑒𝑆) → 𝑀𝑒 2+ + 𝑀𝑒𝑆 (6)
Where, k1 is the 𝑀𝑒 − 𝑀𝑒𝑆 formation rate constant; k-1 is the dissociation constant. Two pairs
of reactions in (2) have equilibrium reactions and constants defined by:
𝐾𝑀𝑒𝑀𝑒𝑆
𝑀𝑒 2+ + 𝑀𝑒𝑆 ↔ (𝑀𝑒 ≡ 𝑀𝑒𝑆) (7)
Where, 𝐾𝑀𝑒𝑀𝑒𝑆 is the equilibrium constant for 𝑀𝑒 − 𝑀𝑒𝑆. The rate equation for the mechanism
in equation (7) and introduced in equation (1) 𝜌𝑠 = 𝜌𝑎𝑑𝑠 − 𝜌𝑑𝑒𝑠 can be written as:
𝜌𝑎𝑑𝑠𝑜𝑟𝑝 = 𝐾𝑥 ∙ 𝑘1 𝑋ads ∙ 𝑆𝑀𝑒 2+ (8𝑎)
𝜌𝑑𝑒𝑠𝑜𝑟𝑝 = 𝐾𝑥 ∙ 𝑘−1 ∙ 𝑋̅ads (8𝑏)
where, SMe2+, Xads and Xads are the dynamic state variable for free TE concentration, free binding
site concentration, adsorbed complex and Kx is the sorption/desorption kinetic rate constant. Note
that 𝑋𝑎𝑑𝑠 and 𝑋𝑑𝑒𝑠 constitute additional state variables of the system, whose dynamics is
explicitly tracked through the sorption/desorption rate. The current model formulation considers
the biomass, inert and precipitates as having similar binding characteristics. It should also be
noted that the model does not distinguish between binding sites based on chemical
characteristics. The idea here is to adapt a universal behaviour for inert, biomass and precipitates
based on binding site density which can be latter improved to incorporate specific binding
characteristics such as carboxylate, phosphate and sulphate surfaces.

Binding sites
The adsorption module has been conceptualized on: 1) formation of free binding sites; 2)
reaction of free binding sites with TEs and light metals to form occupied binding sites-sorption
reaction; 3) desorption of occupied binding sites; 4) release of TEs due to decay of biomass and
5) competition among TEs for free binding site. A schematic representation is reported in Figure
1. Each sorbent component is characterized by a “binding site density”. The binding site density
may be defined as the moles of binding sites per unit gram of the respective surface expressed as
moles/gram. Unless otherwise mentioned, a binding site density of 2 × 10−3 has been used as a
default value for biomass and inert surfaces whereas a value of 2 × 10−5 has been set for
precipitate surfaces. The decay of occupied binding sites results in formation of free binding sites
and TEs. It should be noted that the decay of binding sites is different from desorption processes.
The competition of TEs and light metals for free binding sites has been included in the module in
form of competing mass fluxes for particular free binding sites.
2.1.2Effect of TEs on biochemical processes
Since TEs have been reported as micronutrients in AD systems and are constituents of co-factors
in enzyme systems, their effect on biochemical processes has been explicitly taken into account
by introducing an additional term 𝐼𝑀𝑒 2+ in the inhibition expressions used in ADM1 framework
(Batstone et al., 2002). The function takes into account both the stimulation and inhibiting action
of TEs on biochemical rate depending on the concentration range. This includes an additional
non-competitive biostatic inhibition function, 𝐼𝑀𝑒 2+ ,which keeps in track the effect of TEs on the
biochemical rates and acts as a growth limiting factor. 𝐼𝑀𝑒 2+ is expressed as:
2+
𝑎1 ∙ (𝑀𝑒 2+ + 𝐶 𝑀𝑒 ) + 𝑎2
𝐼𝑀𝑒 2+ = 2+ 2+ , (10)
(𝑀𝑒 2+ + 𝐶𝑀𝑒 )2 + 𝑏1 ∙ (𝑀𝑒 2+ + 𝐶𝑀𝑒 ) + 𝑏2
where, 𝑎1 , 𝑎2 , 𝑏1 , 𝑏2 are assumed constants which can been adjusted to obtain a desirable
optimum dose-response function at a particular TE concentration, 𝑀𝑒 2+ denotes the

2+
concentration of bioavailable TEs within the bulk liquid and 𝐶 𝑀𝑒 is the fraction of TE which
undergoes complexation reaction to form [Me-EDTA]2- (Hu et al., 2008; Vintiloiu et al., 2013).
Note that the amount of TEs adsorbed on the various surfaces present in AD is not involved in
the inhibition function 𝐼𝑀𝑒 2+ as the adsorbed TEs are supposed to be unavailable for the
biochemical reactions.
2.2 Model simulation plan
The capability of the model to cover a wide range of input variables and parameters is a crucial
factor.. In this regard, five different simulation scenarios have been considered to check the
model prediction capability. Scenario 1 analyzed the impact of an increase in adsorbent
concentration (complex organic matter) when the amount of TEs in the system is constant. The
Scenario 2 evaluated the change in speciation of TEs when the initial concentration of a single
TE is changed. Here model simulations were performed for Fe, Ni and Co. The Scenario 3
examined any possible competition among carbonate and sulfide systems to demobilize TEs.
Scenario 4 was used to assess the response of the sorption reactions to the presence of a synthetic
chelating agent.. Lastly, Scenario 5 showcased the variation in the model prediction due to
change in binding site density for biomass, inert and precipitate.
2.3 Model parameters and initial conditions

An in-silico anaerobic batch digester of working volume 0.75 l (head space volume 0.25 l) was
selected for numerical investigations. The temperature of the reactor was set at 35 °C. The initial
conditions for the particular scenario were selected on the basis of experience and literature
which have been reported in Table 1 and 2. The TEs concentration range was selected from
literature (Thanh et al., 2015). The initial values for VFAs were set to zero. The initial values for
ammonia and bicarbonate were adjusted at certain value to run the reactor around neutral pH.
The initial amount of calcium and magnesium were taken from literature (Musvoto et al., 2000).
The values for the saturation constants for sorption processes were collected from various
sources.
2.4 Model Implementation
The implementation of the developed mathematical model has involved: i) the update of the
biochemical framework to include the effect of TEs on microbial kinetics as well as the release
of sulfur, phosphorous and TEs during the disintegration step; ii) the introduction of new acid-
base reactions; iii) inclusion of new liquid-liquid processes (TE-EDTA and TE-VFA
complexation); iv)the inclusion of solid-liquid processes (precipitation); v) incorporation of
dynamic state variables and processes for sorption/desorption processes; vi) the collection of the
parameters values governing the sorption processes. In particular, these values have been
collected from suitable literature and scientific reports.
The system of ordinary differential equations constituting the extended-ADM1 model has been
implemented in an original code and has been solved using the algorithm ODE 15s, a multistep,
variable-order solver on MATLAB® platform. Numerical simulations of specific scenarios have
been performed to test model reliability.

3. Results
3.1 Scenario 1: Effect of adsorbent dosage on TE sorption/desorption in anaerobic digestion
The purpose of this simulation scenario (Figure 2) is to assess the change in TE sorption with
increase in adsorbent amount in the reactor. Adsorbent here refers to composite organic matter
𝑋𝑐 . For all the simulation RUNs the binding site density was fixed at 2 × 10 -3 moles/g of dry
weight. The initial TEs concentration were set at optimal values (Fe: Ni: Co:: 1 × 10 -4 M: 1 × 10-5
M: 1 × 10-7 M). Table 1 and 2 provide the various initial concentrations values for all the process
variables. As shown in Figure 2, with increase in adsorbent amount in the anaerobic system, the
amount of sulfide precipitate decreased with increase in initial complex organic matter.
Moreover, sulfide fraction for Fe was higher than Ni and Co for corresponding initial complex
organic matter concentration. The majority of Ni and Co was found with EDTA complexes.
What stands out in this scenario is the slow increase in TE in inert fraction with increase in initial
complex organic matter concentration (adsorbent). This is clearly evident from the inert fraction
of Fe where the inert fraction increases from around 45% of Fe to about 70% of Fe. For Co there
is increase in the inert fraction, but it is lower than Fe inert fraction. Consequently, the fraction
for FE-EDTA decreased with increase in adsorbent amount. For Ni and Co the TE-EDTA
fraction remained more or less constant across all adsorbent amounts (1-5 gCOD of composite
organic matter) considered in this simulation set.
The cumulative methane production profiles were estimated for 100 days. As shown in the
Figure 2, the cumulative methane production increased proportionally with increase in the
adsorbent amount 𝑋𝑐 . It is interesting to note here that although the amount of TEs are fixed at
one value (Table 1 and 2) for all the model runs, the cumulative methane production increased
proportionally.
3.2 Scenario 2: Effect of initial Fe, Ni and Co concentration on TE adsorption in anaerobic
digestion
In order to assess the changes in speciation pattern of TEs due to change in initial concentrations,
8 model simulations were performed separately for Fe, Ni and Co (Figure 3a, 3b and 3c). The
initial concentration for the TE under study was varied (Table 1 and 2) while the other two TEs
concentrations were set at an optimal value.
Change in initial Fe concentration
8 simulations were carried out with 8 different initial concentrations of Fe (Figure 3a, Table 1
and 2). For all simulations Ni and Co were set at optimal values. The different initial
concentration was chosen keeping in mind the optimal concentration value for Fe, which is 1 ×
10-4 M of Fe. In this scenario, sulfide fraction for Fe and Co was higher than Ni. Ni was found
more with complex fraction. The amount of Fe adsorbed on inert matter is found to be higher at
both lower and higher initial concentration of Fe, in comparison at optimal value of Fe. In RUN
1 through RUN 8 which represent lower to higher initial Fe concentration, we have higher
amount of TE (Fe, Ni and Co) in the inert fraction. Additionally, the analysis of the fractionation
pattern (Figure 3a) shows that with increase in initial Fe concentration the carbonate fraction
increases. The carbonate fraction for RUN 8 (with highest initial Fe concentration, 3 × 10 -3 M of
Fe) is the highest amount for all the model runs among Fe, Ni and Co. It is also interesting to
note that complexation fraction is only observed in case of Ni and Co. The complexation fraction
for both Ni and Co decreases as the initial Fe concentration approaches higher values. The free
TE fraction does not change significantly for Fe. In case of Ni and Co, the free TE fraction
increases with increase in Fe initial concentration.

The cumulative methane production was also examined with change in initial Fe concentration.
There is a slight decrease in methane production as Fe initial concentration approaches higher
value. Subsequently, at the highest initial methane concentration (RUN 8), even if the cumulative
methane production is similar to rest of the RUNs, there was a markedly slower production of
methane.
Change in initial Ni concentration
Figure 3b depicts 8 simulations carried out with 8 different initial concentrations (Table 1 and
2) of Ni. Fe and Co was set at optimal values in this simulation set. The different initial
concentration of Ni was chosen around the optimal concentration value for Ni (1 × 10-5 M of Ni).
As observed in the previous in-silico experiment, sulfide fraction was predominantly found for
Fe and Co. Ni was found more as EDTA complex with lower Ni initial concentration.
Conversely, with higher initial concentration inert fraction of Ni was dominant. The same case is
with the Co fractionation. In case of Fe speciation, the fraction for sulfide and inert remains
constant for all the values of Ni initial concentration. A thorough analysis of the fractionation
pattern (Figure 3b) shows that with increase in initial Ni concentration the carbonate fraction
does not increase significantly for Fe and Ni. It is also interesting to note here that complexation
fraction decreases with increase in Ni initial concentration. For Fe, the complexation fraction
decreases with higher Ni initial concentration. The free TE fraction changes with increase in Ni
initial concentration for Co and for Ni. In case of Fe, the free TE remained more or less constant
throughout all the simulations.
The cumulative methane production was also examined (Figure 3b) with change in initial Ni
concentration. RUN 5 showed the highest cumulative methane production. Subsequently, the
methane production decreased with further increase in initial Ni concentration. The cumulative
methane production for RUN 1, with 5 × 10-7 M of Ni is around 2.3 × 10-2 M. For RUN 8
methane production is the lowest among all simulation runs.
Change in initial Co concentration
Figure 3c presents 8 simulations carried out with 8 different initial concentrations (Table 1 and
2) of Co. Fe and Ni was set at optimal values in this simulation set. Different initial
concentration of Co was chosen keeping in mind the optimal concentration value for Co (1 × 10 -7
M of Co). The trend of fractionation among Fe, Ni and Co is similar to the previous simulation
sets for change in initial Fe and Ni concentrations (Figure 3a and Figure 3b). Higher amount of
sulfide was observed in Fe and Co fractions, and EDTA complexation fraction was higher for
Ni. Higher amount of Fe was found adsorbed on inert matter at all initial concentration of Co. A
very minute fraction of complexation is observed in all model runs for Fe. The fractionation of
Fe shows little or no change in initial Co concentration. The carbonate fraction does not increase
significantly for all the TEs. The free TE fraction does not change significantly for any of TEs
other than the RUN 8 which did not have any EDTA. The cumulative methane production was
also examined (Figure 3c) with change in initial Co concentration. It is evident from the figure
that change in initial concentration of Co did not change drastically the cumulative methane
production. The RUNs (1-8) resulted in cumulative methane production close to around 2.25 ×
10-2 M.
3.3 Scenario 3: Effect of initial Ca-Mg concentration on TE adsorption in anaerobic digestion
The scenario investigated the effect of change in initial Ca and Mg concentrations in terms of
speciation of TEs and cumulative methane production. Figure 4 depicts the effect of Ca and Mg
on the TE speciation. Eight different initial conditions have been considered in this scenario
(Table 1 and 2). The simulations showcased that sulfide fraction is predominantly found with Fe.
EDTA complexation fraction was found across all the simulations other than for Fe. Fe was
found associated more with inert fraction as well as biomass fraction with lower Ca-Mg
concentration. With increase in Ca-Mg initial concentration less inert and biomass fractions were
observed for Fe, Ni and Co. For example, the amount of Fe bound to inert matter at lower initial
concentration of Ca-Mg is higher (RUN 1). However, at higher initial Ca-Mg concentration the
amount of Fe bound to inert matter decreased (RUN 8). Conversely, amount of carbonate
precipitates for all the TEs increased with increase in Ca-Mg. Free TE fraction was only noticed
with Fe. For Ni and Co the free TE fraction was close to zero. Cumulative methane production
was also examined for corresponding Ca-Mg initial concentrations. From Figure 4 it is apparent
that with increase in Ca-Mg dosage the rate of cumulative methane production decreased, while
the cumulative amount of methane produced remained more or less constant for RUN 1 to 7.
However, cumulative methane production in RUN8 is less as compared to the rest of the
simulation RUNs.
3.4 Scenario 4: Effect of initial EDTA concentration on TE sorption/desorption in anaerobic
digestion
Model simulations were carried out to study the effect of initial EDTA concentrations on
sorption of TEs and cumulative methane production. 10 simulations were considered for this
scenario. The initial concentrations of Fe, Ni and Co are set at optimum values for all simulations
(Table 1 and 2). The initial concentration of EDTA was varied from 1 × 10 -5 M to 1 × 10-4 M
(Table 1 and 2). The model simulations presented in Figure 5 demonstrated that out of all the
TE fractions Fe was found predominantly in inert fraction with lower initial concentration of
EDTA. As the initial concentration of EDTA increased, the Fe was observed more in the
complexed fraction. However, Ni and Co were predominantly found with complex fraction, even
with lower initial EDTA concentration. It is interesting to note that, Fe is also found with sulfide
fraction, whereas, Ni and Co is found with sulfide and inert fraction only at lower initial
concentration of EDTA. The fraction of Fe bound to inert is higher as compared to the sulfide
and complexed fraction in lower initial EDTA concentration. Also it is evident that the amount
of Fe-inert fraction decreases slowly with increase in initial EDTA concentration. For example,
the Fe-inert fraction is higher at RUN 1 as compared to RUN 10. The free form of TEs decreased
with increase in EDTA initial concentration. For Fe, the free form decreased with increase in
EDTA initial concentration and a small fraction (~5% Fe) was found in the free form in RUN
10. However, such a fraction of free Ni and Co was not observed in the model simulations other
than at RUN 1. The cumulative methane production was also estimated in relation to the change
in initial EDTA concentration. With increase in EDTA initial concentration the amount of
cumulative methane production does not change drastically. The amount of methane produced at
RUN 1 is close to 2.45 × 10-2 M and the amount of methane produced during RUN 10 is
approximately 2.5 × 10-2 M
3.5 Scenario 5: Effect of binding site density on TE sorption/desorption in anaerobic digestion
Change in biomass binding site density
9 simulations were carried out with different binding site density for biomass (Figure 6a, Table
1 and 2). For all simulations Fe, Ni and Co were set at optimal values. The different binding site
density was chosen keeping in mind the binding site density reported in literature, which is 2 ×
10-3 moles/gram of biomass. In this scenario, change in binding site density for biomass changes
the speciation of TEs in a particular run. For Fe, with decrease in binding site density the fraction
of Fe bound to biomass decrease (RUN 1 to 9). A similar trend has been observed for Co. This is
also true for FeS and CoS. Change in binding site density could affect the speciation of Ni but
the distribution of TEs across different fractions remained more or less constant after RUN 5.
Maximum changes in free TE fraction were observed for RUN 1 and 5. RUN 1 having highest
amount of TEs had lowest cumulative methane production. However, the rest of the simulations
predicted more or less similar amount methane production
Change in inert binding site density
Figure 6b depicts 9 simulations carried out with 9 different initial concentrations (Table 1 and
2) of Ni. Fe, Ni and Co was set at optimal values in this simulation set. So, the different inert
binding site density was chosen similar to the binding site density for biomass. Change in
binding site density changed the inert fraction for simulations from across all the model RUNs.
The inert fraction for Fe decreased as the binding site density decreased from RUN 1 to RUN 9.
Whereas, the increase in binding site density for inert resulted in an increase in free TE fraction,
biomass adsorbed TE fraction, and sulfide fraction. For Co and Ni similar speciation was
observed but with two prominent characteristics. One, the free TE fraction decreased with
decrease in inert binding site density for Ni and Co where as in case of Fe, the free TE fraction
increased upto RUN 5 and then decreased in RUN 9. Two, the TE complexation fractions for Ni
and Co were comparatively higher as compared to EDTA complexation fraction of Fe. The
cumulative methane production was also examined (Figure 6b) with change in inert binding site
density. RUN 9, which were for lowest inert binding site density, showed the highest cumulative
methane production.
Change in precipitate binding site density
Figure 6c presents 9 simulations carried out with 9 different binding site densities for
precipitates. In this model only FeS has been considered to harbour TE binding sites. The
binding site density of FeS for Ni and Co has been assumed in this study. The changes in binding
site density of mineral precipitate did not result in change of TE speciation. The speciation for
Fe, Ni and Co remains intact all throughout the RUNs from 1 to 9. Same is the case with
cumulative methane production.
4. Discussion
4.1 Does the model consider the major TE fluxes operating in the AD system?
In an AD system, depending upon the reactor conditions (i.e. pH, OLR, redox state and residence
time), TEs are found in different forms or species. At a given reactor condition, speciation of
TEs depends primarily on various biogeochemical processes occurring simultaneously at
different rates. Precipitation, dissolution, organic complexation, adsorption, biodegradation of
TE containing intermediates and bio uptake are such determining processes. The resultant
amount of free TE which eventually remains in the liquid phase depends on the amount of TEs
consumed by the mentioned biogeochemical fluxes. Precipitation/dissolution is a major flux
which results in formation of sulfide, carbonate and phosphate minerals in the AD system, and
consequently scavenges the majority of TEs from the system. Similarly, organic complexation
utilizes a portion of free TEs. TEs end up complexed with natural or synthetic organic ligands
present in AD systems. A fraction of TEs is uptaken by biological cells and hence initiates other
cascading processes (such as gene regulation, storage, excretion). However, some amount of TEs
are contributed into the system by the biodegradation (in our case disintegration of complex
organic matter) of TE containing intermediates of AD. Adsorption/desorption of TE is an
important and major physicochemical process, particularly when a large amount of surface area
(biomass, inorganic precipitate and inert material) is available in the AD systems. The proposed
model under discussion, includes the significant processes (including adsorption) involved in
speciation of TEs in AD. Apart from the standard biodegradation reactions (as in ADM1) the
model includes 40 adsorption/desorption reactions, 13 precipitation/dissolution reactions and 15
complexation reactions. The adsorption of TEs have been considered on the biomass, precipitate
and inert matter. Overall, the proposed model contains a TE speciation module in an ADM1
framework.
4.2 What are the biogeochemical characteristics of different surfaces considered for
adsorption?
Anaerobic digestion system provides excellent support material (adsorbent) for adsorption of
TEs. The seven microbial communities (as considered in ADM1) provide bacterial surfaces for
adsorption of TEs. Precipitation products, such as sulfide precipitates, are also a point of
attachment for TEs. Inert material, produced during disintegration step of anaerobic digestion, is
a major source of adsorbent because it harbours many complex polymeric substances, such as
lignocelluloses, cellulose and various complex protein moieties. The proposed model considers
biomass, inert and precipitates as three major groups of adsorbent.
Biomass surface
The surface of microbes contains numerous attachment sites (carboxyl, hydroxyl and phosphoryl
functional groups) for divalent TEs. These biomass surfaces can vary widely among different
species and site. To rather complicate, inter species microbial diversity may be recalled. Presence
of wide range of binding site does not necessarily demand recruitment of different adsorption
behaviour. In this proposed model, universal biomass adsorption behaviour has been adopted. It
has been hypothesized that the TE bacterial surface interactions are common over the seven
groups of microbes considered in the original ADM1. The universal biomass adsorption has been
supported by many studies. For example, (Jiang, 2006) demonstrated that total reflectance FTIR
spectra of both gram positive and negative bacteria are similar. These similarities imply a
common binding environment. Similarly,(Johnson et al., 2007) presented a averaged set of
universal thermodynamic proton binding and site density parameters for modelling bacterial
adsorption reactions in geologic systems.
Inert materials
Inert material is the most abundant adsorbent available in AD systems. Disintegration of
composite particulate material (100%) results in carbohydrates, proteins, lipids and more
importantly inert (10%). The inert are composed of both organic and inorganic material.
However, in the proposed model there is no distinction made on this basis. Rather, inert have
been considered as a surface, similar to biomass, which can attach and bind TEs. Inert can adsorb
TEs as well as Ca and Mg.
Mineral precipitate
Mineral precipitates provide a large surface area for small ions (here TEs) to adsorb. Mineral
precipitates, particularly sulfide, adsorb TEs in AD. (Watson et al 1995) Co and Ni were found
to be adsorbed on FeS. Similarly, Co has a high affinity to iron sulfide precipitates in anaerobic
granular sludge (Van Hullebusch et al., 2006). It should be noted here that, adsorption of TEs on
mineral precipitates is different from co-precipitation. Co-precipitation is the inclusion of metals

into the solid crystal where as adsorption is a reaction occurring on the surface of the mineral
crystal. In the proposed model FeS has been considered as the only mineral precipitate on which
adsorption of other TEs takes place. This is primarily due to unavailability of supportive studies
to incorporate other minerals as adsorbent. Moreover, because higher amount of FeS (Watson et
al., 1995) formed in the AD it is logical and sufficient to consider FeS as the sole mineral
adsorbent.
4.3 Is binding site density necessary in modelling TE adsorption in AD system?
Several reports have mentioned the importance of binding site density in the context of TE
adsorption to microbial and mineral surfaces. In order to evaluate the importance of binding site
density in case of adsorption we employed a simple methodology. In this, a preliminary model
was developed which had all the adsorption components in terms of dynamic state variable and
thermodynamic constants. But lacked binding site density factor. The model prediction capacity
was tested with the already developed models. It was seen that without binding site density, the
model over predicts adsorption of TE. Almost all the TEs are in adsorbed fraction. This results in
a diminishing potential for methane production (Figure 2). A possible explanation for this is the
unavailability of TEs to stimulate methane production. Moreover, if methane production by
ADM1 is considered as control, then inclusion of precipitation processes in the anaerobic
digestion process should result in a decrease in methane production. And further inclusion of
complexation reactions, with EDTA, the methane production should increase. This was exactly
observed (Maharaj et al., 2019, 2018). However, when adsorption processes were incorporated,
the methane production decreased significantly. Considering first principles, the methane
production should not drastically change with inclusion of adsorption processes in the model.
Recalling that TE concentration is set at optimal range for the dose-response function, such a
decrease in methane production is not surprising. Thus, as a solution, binding site density was
incorporated in the proposed model.
The binding site density considered in this study is universal in nature. And the binding site
density can be modified for any given adsorbent. Keeping in mind previous model simulation
experience and literature, binding site density chosen for RUN 5 was selected for further
investigative simulations. Moreover, these simulations suggest that binding site density is an
important parameter in modeling adsorption of TEs in AD systems. Binding site density has been
used both in surface complexation models as well as waste-water process models (Hauduc et al.,
2015; Schwarz and Rittmann, 2007). The approach varies from case to case. For example, when
modelling zinc complexation by organic microbial products binding site density in terms of
moles/g bacteria (dry wt) was introduced in a surface complex model (Schwarz and Rittmann,
2007). Such an approach suits a steady state system. However, in a dynamic physicochemical
model for chemical phosphorous removal, adsorption of phosphates were carried out and the
adsorption model incorporated a term (SiteF/SiteT) n which decrease the kinetic rate of adsorption
(Hauduc et al., 2015). To summarize, binding site density is necessary in modelling TE
adsorption processes in AD and hence implemented in the proposed model.
4.4 How adsorption fraction compare with other fraction of TEs?
The proportional variations and dynamics of chemical fractions for adsorbed TEs under different
initial adsorbent concentration (Figure 2), different initial TE (each for Fe, Ni and Co)
concentration (Figure 3), different initial Ca-Mg concentration (Figure 4), different initial
EDTA concentration (Figure 5) and effect of change in binding site density (Figure 6) have
been studied. Throughout all the simulation scenarios, the adsorption fraction for inert is more
than the adsorption fraction for biomass and precipitates. A possible explanation for this
observation is that amount of inert material formed by the end of AD process is higher as
compared to the amount of biomass and mineral precipitate. The maximum amount of biomass
and mineral precipitate (FeS) is 1.2 × 10-1 gCOD and 1 × 10-4 M, whereas, the inert is around 3.5
× 10-1 gCOD. Secondly, the % fraction is calculated at the end of the 100 days when the amount
of biomass is already at lower limits and the inert materials reaching maximum values. Thirdly,
the binding site density may also influence the adsorption fraction. Now let’s discuss each
simulation scenario one by one. Due to scarcity of literature we have tried to offer possible
explanations based on theoretical understandings, first principles and our experience.
Initial adsorbent concentration
Change in initial adsorbent concentration (𝑋𝑐 ), as reported in Figure 2, affects the TE adsorption
fraction. The increase in TE adsorption fraction is more visible for inert. There is no significant
increase in adsorption to biomass and precipitate. With increase in adsorbent concentration in the
system, the amount of adsorbed TEs increases (particularly with inert matter). This is valid for
Fe, Ni and Co. This relationship can be partly explained as increase in adsorbent amount
provides greater number of surface binding sites and hence the increase in adsorption. The
decrease in free TEs can be attributed to the uptake of TEs by microbial communities and also
increase in adsorption flux.
Initial Fe, Ni and Co concentration
The main objective of the scenarios with change in initial Fe (Figure 3a), Ni (Figure 3b)) and
Co (Figure 3c) was to predict the adsorption behaviour when limited and over supplied (in
comparison to optimal concentration values) amount of TEs are present in the AD system. From
the simulation scenarios it can be claimed that Fe, Ni and Co have distinct pattern of effect on
adsorption. Adsorption pattern as seen in case of different initial Fe concentrations (Figure 4a)
implies a decrease in adsorption fraction in the optimal TE zone (for Fe and Co). This result may
be explained by the fact that around the optimal TE concentrations (RUN 5) the microbial
activity is maximum and the biological TE utilization is at minimum, and the remaining TEs end
up in sulfide fraction due to a suitable pH regime. However, in the sub-optimal initial
concentrations for Fe, a comparatively larger fraction of inert is due to change in reactor pH.
This is also evident from the fact that sub-optimal Fe concentration have inhibited methane
production, which may be translated into accumulation of VFAs and hence a change in pH. Also,
the difference in cumulative methane production for lower Fe concentration (RUN 1) and higher
Fe concentration (RUN 8) is small because addition of more Fe into the system does not affect
the system. However, this is not the case with change in initial Ni concentration (Figure 4b)
where a small change in initial concentration significantly affects the cumulative methane
production. This may be because there are already enough free Ni in the system. Addition of a
small amount of Ni into the system greatly affects the system both in terms of methane
production and speciation of TEs. The increase in sulfide fraction and decrease in inert fraction
implies a underlying mass transfer from inert fractions to the sulfide fraction. It can also be
postulated that first the TEs get adsorbed on the surfaces and slowly gets precipitated with
increase in Ni dosing. Unlike Fe and Ni, change in initial Co concentration did not translate into
any significant change in the TE speciation pattern (Figure 4c). This may be attributed to the
low concentration of Co in the reactor system which can alter the reactor equilibrium
significantly. However, surprisingly, the model is able to capture small change in the cumulative
methane production due to change in Co concentrations. This is a interesting aspect of the model,
particularly when sensitivity of methane production to small concentration of TEs is considered.
Initial Ca-Mg concentration
Ca and Mg have been reported to exhibit competitive effect with TEs in terms of occupying
binding sites. This is because of the similar charge of Ca and Mg as with divalent TE ions. The
decrease in methane production (Figure 4) due to increase in Ca-Mg addition (Table 1 and 2),
can be attributed to the pH of the system and the availability of TEs. More over the decrease in
adsorption fraction with increase in Ca- Mg is due to increase in sulfide precipitate. This is
similar to the change in speciation due to change in initial Fe concentration for Co and Fe.
However, interestingly, the reappearance of inert fraction at higher Ca-Mg initial concentration is
not observed here (Figure 4, Fe-RUN 7). At higher concentration of Ca-Mg TEs prefers to go to
carbonates rather than inert. This can be explained by the increase in saturation level of calcium-
carbonate-magnesium systems due to increase in Ca-Mg concentration. This exercise confirms
that with increase in TE dose the cumulative methane production first increases then decreases
when the dose amount exceeds the optimal concentration limit. However, with Ca-Mg dosage
increase there is only decrease in cumulative methane production. This can partly be explained
by the usage of inhibition function only for the TEs. Secondly, availability of free TEs reduces at
the higher concentration of Ca-Mg dosage.
Initial EDTA concentration
Change in initial EDTA concentration, as reported in Figure 5, affects the TE adsorption
fraction. The decrease in TE adsorption fraction with addition of EDTA confirms its chelating
ability. With decrease in sulfide precipitates, the decrease in inert fraction as evident in case of
Fe speciation is interesting to note. The decrease in sulfide precipitate is in accordance to our
experience (Maharaj et al., 2019).
4.5 What are the limitations and perspectives of the proposed model?
The modelling effort required to strengthen the concept that methane production potential is a
function of TE adsorption, complexation, precipitation and biouptake keeping in mind the usage
of model to recruit a control strategy, may require additional model complexity in terms of
significant processes and components. Considering EPS as a storage house for adsorption of
TEs, which in turn affect the binding site density and hence adsorption behaviour, it is necessary
to incorporate EPS and SMP in order to arrive at a more accurate mathematical model.
Nevertheless, the goal of this study is to build a model framework based on ADM1 which can
simulate the effect of adsorption, complexation and precipitation process on AD. This model can
be used to formulate a TE dosing scheme or a reduced version of this can be used as a model
predictive control of TE dosing in AD. At this stage, with unavailability of proper experimental
data to calibrate and validate such a model, it is necessary to individuate and define other
biochemical and physicochemical processes (EPS and redox state) which can affect the
speciation of TEs in anaerobic digesters. The use of such dynamic model based on ADM1 to
predict the speciation of TEs in anaerobic digestion system is still under development and some
limitations and gap areas in this study can be summarized below:
1) Amino acids and proteins can be considered as a source for TEs.
2) Redox state of the system can be predicted which can improve the TEs speciation
prediction.
3) The effect of diffusion phenomenon on TE across biofilms within the framework of
ADM1 may be studied to assess the role of speciation in anaerobic systems.
4) Production of soluble microbial product as a part of biochemical processes can be
considered.
5) Incorporating a rigorous modeling approach to define and predict binding site density by
taking help of surface complexation models.
5. Conclusions
A general approach to model TEs adsorption/desorption and its effect on methane production has
been implemented to extend an ADM1 based model. An original tailor code to solve the aqueous
chemistry of TE adsorption reactions has been used. In addition to estimating precipitate,
complex formation, the extended ADM1 is able to quantify the amount and chemical form
(adsorbed on biomass, inert and precipitate) of TEs present at a particular point of time in the AD
reactor. TE-inert was found to be a major adsorption species in anaerobic digestion system.
Further, variation of complex organic matter, TEs, EDTA and Ca-Mg initial concentration affect
the adsorption of TEs in anaerobic batch reactor. Binding site density controls the amount of TE
to be adsorbed on the available free surfaces.
6. References
Appels, L., Baeyens, J., Degrève, J., Dewil, R., 2008. Principles and potential of the anaerobic
digestion of waste-activated sludge. Prog. Energy Combust. Sci. 34, 755–781.
https://doi.org/10.1016/j.pecs.2008.06.002
Appels, L., Lauwers, J., Degrve, J., Helsen, L., Lievens, B., Willems, K., Van Impe, J., Dewil,
R., 2011. Anaerobic digestion in global bio-energy production: Potential and research
challenges. Renew. Sustain. Energy Rev. 15, 4295–4301.
https://doi.org/10.1016/j.rser.2011.07.121
Arican, B., Gokcay, C.F., Yetis, U., 2002. Mechanistics of nickel sorption by activated sludge.
Process Biochem. 37, 1307–1315. https://doi.org/10.1016/S0032-9592(02)00015-8
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S. V., Pavlostathis, S.G., Rozzi, A.,
Sanders, W.T., Siegrist, H., Vavilin, V.A., 2002. The IWA Anaerobic Digestion Model No
1 (ADM1). Water Sci. Technol. 45, 65–73. https://doi.org/10.2166/wst.2008.678
Choong, Y.Y., Ismail, N., Abdullah, A.Z., Yhaya, M.F., 2016. Impacts of trace element
supplementation on the performance of anaerobic digestion process: A critical review.
Bioresour. Technol. 209, 369–379. https://doi.org/10.1016/j.biortech.2016.03.028
Fermoso, F.G., Bartacek, J., Jansen, S., Lens, P.N.L., 2009. Metal supplementation to UASB
bioreactors: from cell-metal interactions to full-scale application. Sci. Total Environ. 407,
3652–3667. https://doi.org/10.1016/j.scitotenv.2008.10.043
Fermoso, F.G., Hullebusch, E.D. Van, Guibaud, G., Collins, G., Svensson, B.H., Vink, J.P.M.,
Esposito, G., Frunzo, L., 2015. Fate of Trace Metals in Anaerobic Digestion, in: Gubitz, G.,
Bauer, A., Bochmann, G., Gronauer, A., Weiss, S. (Eds.), Biogas Science and
Technology.Advances in Biochemical Engineering/Biotechnology. pp. 171–195.
https://doi.org/10.1007/978-3-319-21993-6
Hauduc, H., Smith, S., Szabo, A., Murthy, S., Tak, I., 2015. A dynamic physicochemical model
for chemical phosphorus removal 3.
Hu, Q.H., Li, X.F., Liu, H., Du, G.C., Chen, J., 2008. Enhancement of methane fermentation in
the presence of Ni2+ chelators. Biochem. Eng. J. 38, 98–104.
https://doi.org/10.1016/j.bej.2007.07.002
Hullebusch, E.D. Van, Guibaud, G., Simon, S., Lenz, M., Yekta, S.S., Fermoso, F.G., Jain, R.,
Duester, L., Roussel, J., Guillon, E., Skyllberg, U., Almeida, C.M.R., Garuti, M., Frunzo,
L., Esposito, G., Carliell-, C., Ortner, M., Collins, G., 2016. Methodological approaches for
fractionation and speciation to estimate trace element bioavailability in engineered
anaerobic digestion ecosystems : An overview. Crit. Rev. Environ. Sci. Technol. 46, 1324–
1366. https://doi.org/10.1080/10643389.2016.1235943
Jiang, B., 2006. The effect of trace elements on the metabolism of methanogenic consortia.
Johnson, K.J., Szymanowski, J.E.S., Borrok, D., Huynh, T.Q., Fein, J.B., 2007. Proton and metal
adsorption onto bacterial consortia : Stability constants for metal – bacterial surface
complexes 239, 13–26. https://doi.org/10.1016/j.chemgeo.2006.12.002
Li, Y., Park, S.Y., Zhu, J., 2011. Solid-state anaerobic digestion for methane production from
organic waste. Renew. Sustain. Energy Rev. 15, 821–826.
https://doi.org/10.1016/j.rser.2010.07.042
Maharaj, B.C., Mattei, M.R., Frunzo, L., Hullebusch, E.D. va., Esposito, G., 2019. ADM1 based
mathematical model of trace element complexation in anaerobic digestion processes.
Bioresour. Technol. 253–259. https://doi.org/10.1016/j.biortech.2018.12.064
Maharaj, B.C., Mattei, M.R., Frunzo, L., van Hullebusch, E.D., Esposito, G., 2018. ADM1 based
mathematical model of trace element precipitation/dissolution in anaerobic digestion
processes. Bioresour. Technol. 267, 666–676.
Mudhoo, A., Kumar, S., 2013. Effects of heavy metals as stress factors on anaerobic digestion
processes and biogas production from biomass. Int. J. Environ. Sci. Technol. 10, 1383–
1398. https://doi.org/10.1007/s13762-012-0167-y
Musvoto, Wentzel, M.C.M., Ekama, G.A.M., 2000. Integrated chemical-physical processes
modelling-II. Simulating aeration treatment of anaerobic digester supernatants. Water Res.
34, 1868–1880. https://doi.org/10.1016/S0043-1354(99)00335-8.
Nguyen, D., Gadhamshetty, V., Nitayavardhana, S., Khanal, S.K., 2015. Automatic process
control in anaerobic digestion technology: A critical review. Bioresour. Technol. 193, 513–
522. https://doi.org/10.1016/j.biortech.2015.06.080
Schwarz, A.O., Rittmann, B.E., 2007. A biogeochemical framework for metal detoxification in
sulfidic systems. Biodegradation 18, 675–692. https://doi.org/10.1007/s10532-007-9101-2
Thanh, P., Ketheesan, B., Zhou, Y., Stuckey, D.C., 2015. Trace metal speciation and
bioavailability in anaerobic digestion : A review. Biotechnol. Adv. 34, 122–136.
https://doi.org/10.1016/j.biotechadv.2015.12.006
Van Hullebusch, E.D., Gieteling, J., Zhang, M., Zandvoort, M.H., Daele, W. Van, Defrancq, J.,
Lens, P.N.L., 2006. Cobalt sorption onto anaerobic granular sludge: Isotherm and spatial
localization analysis. J. Biotechnol. 121, 227–240.
https://doi.org/10.1016/j.jbiotec.2005.07.011
Vintiloiu, A., Boxriker, M., Lemmer, A., Oechsner, H., Jungbluth, T., Mathies, E., Ramhold, D.,
2013. Effect of ethylenediaminetetraacetic acid (EDTA) on the bioavailability of trace
elements during anaerobic digestion. Chem. Eng. J. 223, 436–441.
https://doi.org/10.1016/j.cej.2013.02.104
Watson, J.H.P., Ellwood, D.C., Deng, Q., Mikhalovsky, S., Haytert, C.E., Evanst, J., 1995.
HEAVY METAL ADSORPTION ON BACTERIALLY PRODUCED FeS 8, 1097–1108.
Worms, I., Simon, D.F., Hassler, C.S., Wilkinson, K.J., 2006. Bioavailability of trace metals
to aquatic microorganisms: importance of chemical, biological and physical processes
on biouptake. Biochimie 88, 1721–1731. https://doi.org/10.1016/j.biochi.2006.09.008
Yekta, S.S., Skyllberg, U., Danielsson, Å., Björn, A., Svensson, B.H., 2016. Chemical speciation
of sulfur and metals in biogas reactors—Implications for cobalt and nickel bio-uptake
processes. J. Hazard. Mater. 324, 110–116. https://doi.org/10.1016/j.jhazmat.2015.12.058
Zandvoort, B.M.H., Hullebusch, E.D. Van, Fermoso, F.G., Lens, P.N.L., 2006. Trace Metals in
Anaerobic Granular Sludge Reactors : Bioavailability and Dosing Strategies. Eng. Life Sci.
6, 293–301. https://doi.org/10.1002/elsc.200620129.
Table 1: Initial concentration of various dynamic state variables considered in the model.
Variable Unit
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5
COM Variable TEs Ca-Mg EDTA Variable Binding density
Fe Biomass Inert Precipitate
1 Ssu 0 0 0 0 0 0 0 0 0 g COD L-1
2 Saa 0 0 0 0 0 0 0 0 0 g COD L-1
3 Sfa 0 0 0 0 0 0 0 0 0 g COD L-1
4 Sva 0 0 0 0 0 0 0 0 0 g COD L-1
5 Sbu 0 0 0 0 0 0 0 0 0 g COD L-1
6 Spro 0 0 0 0 0 0 0 0 0 g COD L-1
7 Sac 0 0 0 0 0 0 0 0 0 g COD L-1
8 Sh2 0 0 0 0 0 0 0 0 0 g COD L-1
9 Sch4 0 0 0 0 0 0 0 0 0 g COD L-1
10 Sco2 0 0 0 0 0 0 0 0 0 g COD L-1
11 Snh3 0 0 0 0 0 0 0 0 0 g COD L-1
12 SI 0 0 0 0 0 0 0 0 0 g COD L-1
13 Xc Variable* 2 2 2 2 2 2 2 2 g COD L-1
14 Xch 0 0 0 0 0 0 0 0 0 g COD L-1
15 Xpr 0 0 0 0 0 0 0 0 0 g COD L-1
16 Xli 0 0 0 0 0 0 0 0 0 g COD L-1
17 Xsu 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
18 Xaa 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
19 Xfa 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
20 Xc4 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
21 Xpro 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
22 Xac 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
23 Xh2 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 0.12 g COD L-1
24 XI 0 0 0 0 0 0 0 0 0 g COD L-1
31 Shva 0 0 0 0 0 0 0 0 0 g COD L-1
32 Shbu 0 0 0 0 0 0 0 0 0 g COD L-1
33 Shpro 0 0 0 0 0 0 0 0 0 g COD L-1
34 Shac 0 0 0 0 0 0 0 0 0 g COD L-1
35 Shco3 0 0 0 0 0 0 0 0 0 M
36 Snh4 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 M
37 Sco3 0.004 0.004 0.004 0.004 0.004 0.004 0.004 0.004 0.004 M
38 SCa 0.0025 0.0025 Variable* 0.0025 Variable* 0.0025 0.0025 0.0025 0.0025 M
40 SMg 0.0025 0.0025 Variable* 0.0025 Variable* 0.0025 0.0025 0.0025 0.0025 M
42 SNi 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 M
44 SCo 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001 M
46 SFe 0.0001 Variable* 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 M
51 Spo4 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 M
58 Shs 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 M
71 Sedta4- 0.0001 0.00001 0.0001 Variable* 0.0001 Variable* 0.0001 0.0001 0.0001 M
* Refer Table 2
Table 2. Variable initial concentration (M) of complex organic matter, TEs, EDTA, Ca and Mg used in the model simulations.
Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5

RUN
Variable Variable TEs Variable Ca- Variable Variable Binding density

COM Mg EDTA
Fe Ni Co Biomass Inert Precipitate
-6 -7 -9 -3 -3 1 1
1 1 5.0×10 5.0×10 5.0×10 1.0×10 1.0×10 2.0×10 2.0×10 2.0×10-1
2 2 9.0×10-6 9.0×10-7 9.0×10-9 1.5×10-3 1.5×10-3 2.0 2.0 2.0×10-2
3 3 3.0×10-5 3.0×10-6 3.0×10-8 2.0×10-3 2.0×10-3 2.0×10-1 2.0×10-1 2.0×10-3
4 4 7.0×10-5 7.0×10-6 7.0×10-8 2.5×10-3 2.5×10-3 2.0×10-2 2.0×10-2 2.0×10-4
5 5 1.0×10-4 1.0×10-5 1.0×10-7 3.0×10-3 3.0×10-3 2.0×10-3 2.0×10-3 2.0×10-5
6 5.0×10-4 5.0×10-5 5.0×10-7 3.5×10-3 3.5×10-3 2.0×10-4 2.0×10-4 2.0×10-6
7 9.0×10-4 9.0×10-5 9.0×10-7 4.0×10-3 4.0×10-3 2.0×10-5 2.0×10-5 2.0×10-7
8 3.0×10-3 3.0×10-4 3.0×10-6 2.0×10-6 2.0×10-6 2.0×10-8
9 2.0×10-7 2.0×10-7 2.0×10-9
Figure 1: A simple illustration of the sorption /desorption model.
CH
4 Fe
0.06
100
0.05
80
FeS(s)
0.04 FeCO3(s)
% Fraction
Moles
60 Fe3(PO4)2(s)
0.03 =I-Fe(s)
1 gCOD =B-Fe-(s)
40
0.02 2 gCOD [FeEDTA]2- (l)
3 gCOD
Fe2+ (l)
4 gCOD 20
0.01
5 gCOD
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5
Days
Ni Co
100 100
NiS(s) CoS(s)
80 80
NiCO3(s) CoCO3(s)
% Fraction
% Fraction
Ni3(PO4) 2(s) Co3(PO4)2(s)
60 60
=I-Ni(s) =I-Co(s)
=B-Ni-(s) =B-Co-(s)
40 =P-Ni(s) 40 =P-Co(s)
2-
[NiEDTA] (l) [CoEDTA]2- (l)
Ni2+ (l) Co2+ (l)
20 20
0 0
1 2 3 4 5 1 2 3 4 5
Figure 2: TE speciation and methane production in an anaerobic batch reactor, under different initial complex organic matter concentrations.
CH
4 Fe
0.025
100
FeS(s)
0.02 R1-5*10-6Fe
80 FeCO3(s)
R2-9*10-6Fe
Fe3(PO4) 2(s)
% Fraction
R3-3*10-5Fe
0.015
Moles
60 =I-Fe(s)
R4-7*10-5Fe
=B-Fe-(s)
R5-1*10-4Fe
0.01 [FeEDTA]2- (l)
R6-5*10-4Fe 40
-4 Fe2+(l)
R7-9*10 Fe
0.005 R8-3*10-3Fe 20
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8
Days
Ni Co
100 100
CoS(s)
80 NiS(s) 80 CoCO3(s)
NiCO3(s) Co3(PO4) 2(s)
% Fraction
% Fraction
60 Ni3(PO4)2(s) 60 =I-Co(s)
=I-Ni(s) =B-Co-(s)
=B-Ni-(s) =P-Co(s)
40 40
=P-Ni(s) [CoEDTA]2- (l)
[NiEDTA]2- (l) Co2+(l)
20 Ni2+ (l) 20
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 3.a. TE speciation and methane production in an anaerobic batch reactor, under different initial Fe concentrations.
CH
4 Fe
0.025
100
0.02 FeS(s)
80 FeCO3(s)
R1-5*10-7Ni
Fe3(PO4)2(s)
% Fraction
0.015 R2-9*10-7Ni
Moles
60 =I-Fe(s)
R3-3*10-6Ni
R4-7*10-6Ni =B-Fe-(s)
0.01
R5-1*10-5Ni 40 [FeEDTA]2- (l)
Fe2+ (l)
R6-5*10-5Ni
0.005 R7-9*10-5Ni 20
R8-3*10-4Ni
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8
Days
Ni Co
100 100
CoS(s)
NiS(s)
80 80 CoCO3(s)
NiCO3(s)
Co3(PO4) 2(s)
% Fraction
% Fraction
Ni3(PO4)2(s)
60 60 =I-Co(s)
=I-Ni(s)
=B-Co-(s)
=B-Ni-(s)
=P-Co(s)
40 =P-Ni(s) 40
[CoEDTA]2-(l)
[NiEDTA]2-(l)
2+
Co2+(l)
20 Ni (l) 20
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 3.b. TE speciation and methane production in an anaerobic batch reactor, under different initial Ni concentrations.
CH
4 Fe
0.025
100
0.02 FeS(s)
R2-5*10-9Co 80
FeCO3(s)
R2-9*10-9Co
% Fraction
R3-3*10-8Co Fe3(PO4)2(s)
0.015
Moles
60
R4-7*10-8Co =I-Fe(s)
R5-1*10-7Co =B-Fe-(s)
0.01
R6-5*10-7Co
40 [FeEDTA]2- (l)
R7-9*10-7Co Fe2+ (l)

0.005 -6
R8-3*10 Co 20
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8
Days
Ni Co
100 100
NiS(s) CoS(s)
80 80 CoCO3(s)
NiCO3(s)
Co3(PO4) 2(s)
% Fraction
% Fraction
Ni3(PO4) 2(s)
60 =I-Ni(s) 60 =I-Co(s)
=B-Ni-(s) =B-Co-(s)
40 =P-Ni(s) 40 =P-Co(s)
[NiEDTA]2- (l) [CoEDTA]2-(l)
Ni2+(l) Co2+(l)
20 20
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 3.c. TE speciation and methane production in an anaerobic batch reactor, under different initial Co concentrations.
CH
4 Fe
0.025
100
0.02
80
R1-1*10-3Ca-Mg FeS(s)
% Fraction
0.015 R2-1.5*10-3Ca-Mg FeCO3(s)
Moles
60
-3
R3-2*10 Ca-Mg Fe3(PO4)2(s)
R4-2.5*10-3Ca-Mg
0.01 40 =I-Fe(s)
R5-3*10-3Ca-Mg =B-Fe-(s)
-3
R6-3.5*10 Ca-Mg [FeEDTA]2- (l)
0.005 -3
R7-4*10 Ca-Mg 20
Fe2+(l)
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8
Days
Ni Co
100 100
80 80
NiS(s) CoS(s)
% Fraction
% Fraction
NiCO3(s) CoCO3(s)
60 60
Ni3(PO4)2(s) Co3(PO4) 2(s)
=I-Ni(s) =I-Co(s)
40 40 =B-Co-(s)
=B-Ni-(s)
=P-Ni(s) =P-Co(s)
20 [NiEDTA]2- (l) 20 [CoEDTA]2-(l)
Ni2+ (l) Co2+ (l)
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 4. TE speciation and methane production in an anaerobic batch reactor, under different initial Ca-Mg concentrations.
CH
4 Fe
0.025
100
R1-1*10-5EDTA
0.02
R2-2*10-5EDTA 80
FeS(s)
R3-3*10-5EDTA
% Fraction
FeCO3(s)
0.015 R4-4*10-5EDTA
Moles
60
Fe3(PO4) 2(s)
R5-5*10-5EDTA
-5 =I-Fe(s)
0.01 R6-6*10 EDTA
40 =B-Fe-(s)
R7-7*10-5EDTA
[FeEDTA]2- (l)
R8-8*10-5EDTA
0.005 20 Fe2+ (l)
R9-9*10-5EDTA
-5
R10-10*10 EDTA
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10
Days
Ni Co
100 100
80 80
NiS(s) CoS(s)
NiCO3(s)
% Fraction
% Fraction
CoCO3(s)
60 Ni3(PO4) 2(s) 60 Co3(PO4)2(s)
=I-Ni(s) =I-Co(s)
40 =B-Ni-(s) 40 =B-Co-(s)
=P-Ni(s) =P-Co(s)
2-
20 2+ 20
Ni (l) Co2+ (l)
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Figure 5. TE speciation and methane production in an anaerobic batch reactor, under different initial EDTA concentrations.
CH
4 Fe
0.025
100
0.02
R1-2.0*101 80
FeS(s)
R2-2.0
FeCO3(s)
% Fraction
0.015 R3-2.0*10-1
Moles
60 Fe3(PO4) 2(s)
R4-2.0*10-2
R5-2.0*10-3 =I-Fe(s)
0.01 40 =B-Fe-(s)
R6-2.0*10-4
[FeEDTA]2- (l)
R7-2.0*10-5
0.005 -6 Fe2+(l)
R8-2.0*10 20
-7
R9-2.0*10
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9
Days
Ni Co
100 100
80 80
NiS(s) CoS(s)
NiCO3(s) CoCO3(s)
% Fraction
% Fraction
60 Ni3(PO4) 2(s) 60 Co3(PO4)2(s)
=I-Ni(s) =I-Co(s)
40 =B-Ni-(s) 40 =B-Co-(s)
=P-Ni(s) =P-Co(s)
2-
[NiEDTA] (l) [CoEDTA]2-(l)
20 2+ 20
Ni (l) Co2+ (l)
0 0
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9
Figure 6.a. TE speciation and methane production in an anaerobic batch reactor under different initial biomass binding site density.
CH
4 Fe
0.025
100
0.02
R1-2.0*101 80
R2-2.0 FeS(s)
% Fraction
R3-2.0*10-1 FeCO3(s)
0.015
Moles
60
R4-2.0*10-2 Fe3(PO4)2(s)
R5-2.0*10-3 =I-Fe(s)
0.01 40
R6-2.0*10-4 =B-Fe-(s)
R7-2.0*10-5 [FeEDTA]2-(l)
0.005 R8-2.0*10-6 20 Fe2+ (l)
R9-2.0*10-7
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9
Days
Ni Co
100 100
80 NiS(s) 80 CoS(s)
NiCO3(s) CoCO3(s)
% Fraction
% Fraction
Ni3(PO4)2(s) Co3(PO4) 2(s)
60 60
=I-Ni(s) =I-Co(s)
=B-Ni-(s) =B-Co-(s)
40 =P-Ni(s) 40
=P-Co(s)
2-
2+
20 Ni (l) 20 Co2+(l)
0 0
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9
Figure 6.b. TE speciation and methane production in an anaerobic batch reactor under different initial inert binding site density.
CH
4 Fe
0.02
100
0.015 R1-2.0*10-1 80
FeS(s)
R2-2.0**10-2
FeCO3(s)
% Fraction
-3
R3-2.0*10
Moles
60 Fe3(PO4) 2(s)
0.01 R4-2.0*10-4
=I-Fe(s)
R5-2.0*10-5
40 =B-Fe-(s)
R6-2.0*10-6
[FeEDTA]2- (l)
0.005 R7-2.0*10-7
-8
Fe2+ (l)
R8-2.0*10 20
0 0
0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8
Days
Ni Co
100 100
80 NiS(s) 80 CoS(s)
NiCO3(s) CoCO3(s)
% Fraction
% Fraction
Ni3(PO4) 2(s) Co3(PO4)2(s)
60 60
=I-Ni(s) =I-Co(s)
=B-Ni-(s) =B-Co-(s)
40 =P-Ni(s) 40 =P-Co(s)
[NiEDTA]2- (l) [CoEDTA]2-(l)
20 Ni2+ (l) 20 Co2+ (l)
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 6.c. TE speciation and methane production in an anaerobic batch reactor, under different initial precipitate binding site density.
Supplementary Material
S1: Stoichiometric matrix for Fe adsorption reactions on biomass.
Components 46 96 100 102 104 106 108 110 97 101 103 105 107 109 111 Process Rate
Process SFe2+
Ad1 [Fe-Xsu] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑠𝑢
Ad2 [Fe-Xaa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑎𝑎
Ad3 [Fe-Xfa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑓𝑎
Ad4 [Fe-Xc4] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑐4
Ad5 [Fe-Xpro] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑝𝑟𝑜
Ad6 [Fe-Xac] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋𝑎𝑐
Ad7 [Fe-Xh2] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝐹𝑒_𝐹𝐵𝑋ℎ2
Des1 [Fe-Xsu] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑠𝑢
Des2 [Fe-Xaa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑎𝑎
Des3 [Fe-Xfa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑓𝑎
Des4 [Fe-Xc4] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑐4
Des5 [Fe-Xpro] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑝𝑟𝑜
Des6 [Fe-Xac] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋𝑎𝑐
Des7 [Fe-Xh2] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝐹𝑒_𝑂𝐵𝑋ℎ2
Dec1 [Fe-FBXsu] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑠𝑢𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑠𝑢 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐹𝑒 … )
Dec3 [Fe-FBXaa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑎𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑎𝑎 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐹𝑒 … )
Dec5 [Fe-FBXfa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑓𝑎𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐹𝑒 … )
𝐹𝑒
Dec7 [Fe-FBXc4] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑐4𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑐4𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑐4𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑐4𝐹𝑒 … )
Dec9 [Fe-FBX pro] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑝𝑟𝑜 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐹𝑒 … )
Dec11 [Fe-FBXac] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑐𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥𝑎𝑐 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐹𝑒 … )
Dec13 [Fe-FBX h2] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥ℎ2𝐹𝑒 ∙ (𝑋𝐹𝐵𝑥ℎ2𝐹𝑒 ⁄𝑋𝐹𝐵𝑥ℎ2𝐹𝑒 + 𝑋𝑂𝐵𝑥ℎ2𝐹𝑒 … )
Dec2 [Fe-OBXsu] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑠𝑢𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑠𝑢 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐹𝑒 … )
Dec4 [Fe-OBXaa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑎𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑎𝑎 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐹𝑒 … )
Dec6 [Fe-OBXfa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑓𝑎𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐹𝑒 … )
𝐹𝑒
Dec8 [Fe-OBXc4] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑐4𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑐4𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑐4𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑐4𝐹𝑒 … )
Dec10 [Fe-OBXpro] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑝𝑟𝑜 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐹𝑒 … )
Dec12 [Fe-OBXac] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑐𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥𝑎𝑐 𝐹𝑒 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐹𝑒 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐹𝑒 … )
Dec14 [Fe-OBXh2] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥ℎ2𝐹𝑒 ∙ (𝑋𝑂𝐵𝑥ℎ2𝐹𝑒 ⁄𝑋𝐹𝐵𝑥ℎ2𝐹𝑒 + 𝑋𝑂𝐵𝑥ℎ2𝐹𝑒 … )
S2: Stoichiometric matrix for Co adsorption reactions on biomass.
Process SCo2+
Ad8 [Co-Xsu] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝑠𝑢𝐶𝑜
Ad9 [Co-Xaa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑎𝐶𝑜
Ad10 [Co-Xfa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝑓𝑎𝐶𝑜
Ad11 [Co-Xc4] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝑐4𝐶𝑜
Ad12 [Co-Xpro] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝑝𝑟𝑜𝐶𝑜
Ad13 [Co-Xac] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑂2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑐𝐶𝑜
Ad14 [Co-Xh2] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋ℎ2𝐶𝑜
Des8 [Co-Xsu] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑠𝑢𝐶𝑜
Des9 [Co-Xaa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑎𝐶𝑜
Des10 [Co-Xfa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑓𝑎𝐶𝑜
Des11 [Co-Xc4] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑐4𝐶𝑜
Des12 [Co-Xpro] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑝𝑟𝑜𝐶𝑜
Des13 [Co-Xac] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑐𝐶𝑜
Des14 [Co-Xh2] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋ℎ2𝐶𝑜
Dec15 [Co-FBXsu] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑠𝑢 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑜 … )
Dec17 [Co-FBXaa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑎𝑎 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑜 … )
Dec19 [Co-FBXfa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑜 … )
𝐶𝑜
Dec21 [Co-FBXc4] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑐4𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑐4𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑐4𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑐4𝐶𝑜 … )
Dec23 [Co-FBXpro] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑝𝑟𝑜 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑜 … )
Dec25 [Co-FBXac] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥𝑎𝑐 𝐶𝑜⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑜 … )
Dec27 [Co-FBXh2] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥ℎ2𝐶𝑜 ∙ (𝑋𝐹𝐵𝑥ℎ2𝐶𝑜 ⁄𝑋𝐹𝐵𝑥ℎ2𝐶𝑜 + 𝑋𝑂𝐵𝑥ℎ2𝐶𝑜 … )
Dec16 [Co-OBXsu] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑠𝑢 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑜 … )
Dec18 [Co-OBXaa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑎𝑎 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑜 … )
Dec20 [Co-OBXfa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑜 … )
𝐶𝑜
Dec22 [Co-OBXc4] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑐4𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑐4𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑐4𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑐4𝐶𝑜 … )
Dec24 [Co-OBXpro] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑝𝑟𝑜 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑜 … )
Dec26 [Co-OBXac] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥𝑎𝑐 𝐶𝑜 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑜 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑜 … )
Dec28 [Co-OBXh2] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥ℎ2𝐶𝑜 ∙ (𝑋𝑂𝐵𝑥ℎ2𝐶𝑜 ⁄𝑋𝐹𝐵𝑥ℎ2𝐶𝑜 + 𝑋𝑂𝐵𝑥ℎ2𝐶𝑜 … )
S3: Stoichiometric matrix for Ni adsorption reactions on biomass.
Process SNi2+
Ad15 [Ni-Xsu] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑠𝑢𝑁𝑖
Ad16 [Ni-Xaa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑎𝑁𝑖
Ad17 [Ni-Xfa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑓𝑎𝑁𝑖
Ad18 [Ni-Xc4] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑐4𝑁𝑖
Ad19 [Ni-Xpro] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑝𝑟𝑜𝑁𝑖
Ad20 [Ni-Xac] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑐𝑁𝑖
Ad21 [Ni-Xh2] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋ℎ2𝑁𝑖
Des15 [Ni-Xsu] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑠𝑢𝑁𝑖
Des16 [Ni-Xaa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑎𝑁𝑖
Des17 [Ni-Xfa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑓𝑎𝑁𝑖
Des18 [Ni-Xc4] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑐4𝑁𝑖
Des19 [Ni-Xpro] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑝𝑟𝑜𝑁𝑖
Des20 [Ni-Xac] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑐𝑁𝑖
Des21 [Ni-Xh2] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋ℎ2𝑁𝑖
Dec29 [Ni-FBXsu] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑠𝑢𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑠𝑢 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑠𝑢𝑁𝑖 … )
Dec31 [Ni-FBXaa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑎𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑎𝑎 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑎𝑎𝑁𝑖 … )
Dec33 [Ni-FBXfa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑓𝑎𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑓𝑎𝑁𝑖 … )
𝑁𝑖
Dec35 [Ni-FBXc4] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑐4𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑐4𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑐4𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑐4𝑁𝑖 … )
Dec37 [Ni-FBX pro] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑝𝑟𝑜 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑁𝑖 … )
Dec39 [Ni-FBXac] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑐𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥𝑎𝑐 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑎𝑐𝑁𝑖 … )
Dec41 [Ni-FBX h2] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥ℎ2𝑁𝑖 ∙ (𝑋𝐹𝐵𝑥ℎ2𝑁𝑖 ⁄𝑋𝐹𝐵𝑥ℎ2𝑁𝑖 + 𝑋𝑂𝐵𝑥ℎ2𝑁𝑖 … )
Dec30 [Ni-OBXsu] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑠𝑢𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑠𝑢 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑠𝑢𝑁𝑖 … )
Dec32 [Ni-OBXaa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑎𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑎𝑎 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑎𝑎𝑁𝑖 … )
Dec34 [Ni-OBXfa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑓𝑎𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑓𝑎𝑁𝑖 … )
𝑁𝑖
Dec36 [Ni-OBXc4] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑐4𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑐4𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑐4𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑐4𝑁𝑖 … )
Dec38 [Ni-OBXpro] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑝𝑟𝑜 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑁𝑖 … )
Dec40 [Ni-OBXac] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑐𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥𝑎𝑐 𝑁𝑖 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝑁𝑖 + 𝑋𝑂𝐵𝑥𝑎𝑐𝑁𝑖 … )
Dec42 [Ni-OBXh2] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥ℎ2𝑁𝑖 ∙ (𝑋𝑂𝐵𝑥ℎ2𝑁𝑖 ⁄𝑋𝐹𝐵𝑥ℎ2𝑁𝑖 + 𝑋𝑂𝐵𝑥ℎ2𝑁𝑖 … )
S4: Stoichiometric matrix for Ca adsorption reactions on biomass.
Process SCa2+
Ad22 [Ca-Xsu] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑠𝑢𝐶𝑎
Ad23 [Ca-Xaa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑎𝐶𝑎
Ad24 [Ca-Xfa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑓𝑎𝐶𝑎
Ad25 [Ca-Xc4] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑐4𝐶𝑎
Ad26 [Ca-X pro] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑝𝑟𝑜𝐶𝑎
Ad27 [Ca-Xac] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑐𝐶𝑎
Ad28 [Ca-X h2] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋ℎ2𝐶𝑎
Des22 [Ca-Xsu] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑠𝑢𝑁𝑖
Des23 [Ca-Xaa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑎𝑁𝑖
Des24 [Ca-Xfa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑓𝑎𝑁𝑖
Des25 [Ca-Xc4] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑐4𝑁𝑖
Des26 [Ca-X pro] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑝𝑟𝑜𝑁𝑖
Des27 [Ca-Xac] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑐𝑁𝑖
Des28 [Ca-X h2] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋ℎ2𝑁𝑖
Dec43 [Ca-FBXsu] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑠𝑢 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑎 … )
Dec45 [Ca-FBXaa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑎𝑎 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑎 … )
Dec47 [Ca-FBXfa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑎 … )
𝐶𝑎
Dec49 [Ca-FBXc4] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑐4𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑐4𝐶𝑎⁄𝑋𝐹𝐵𝑥𝑐4𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑐4𝐶𝑎 … )
Dec51 [Ca-FBXpro] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑝𝑟𝑜 𝐶𝑎⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑎 … )
Dec53 [Ca-FBXac] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥𝑎𝑐 𝐶𝑎⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑎 … )
Dec55 [Ca-FBXh2] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥ℎ2𝐶𝑎 ∙ (𝑋𝐹𝐵𝑥ℎ2 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥ℎ2𝐶𝑎 + 𝑋𝑂𝐵𝑥ℎ2𝐶𝑎 … )
Dec44 [Ca-OBXsu] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑠𝑢 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑠𝑢𝐶𝑎 … )
Dec46 [Ca-OBXaa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑎𝑎 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑎𝑎𝐶𝑎 … )
Dec48 [Ca-OBXfa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑓𝑎𝐶𝑎 … )
𝐶𝑎
Dec50 [Ca-OBXc4] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑐4𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑐4𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑐4𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑐4𝐶𝑎 … )
Dec52 [Ca-OBX pro] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑝𝑟𝑜 𝐶𝑎 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝐶𝑎 … )
Dec54 [Ca-OBXac] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥𝑎𝑐 𝐶𝑎⁄𝑋𝐹𝐵𝑥𝑎𝑐𝐶𝑎 + 𝑋𝑂𝐵𝑥𝑎𝑐𝐶𝑎 … )
Dec56 [Ca-OBX h2] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥ℎ2𝐶𝑎 ∙ (𝑋𝑂𝐵𝑥ℎ2𝐶𝑎 ⁄𝑋𝐹𝐵𝑥ℎ2𝐶𝑎 + 𝑋𝑂𝐵𝑥ℎ2𝐶𝑎 … )
S5: Stoichiometric matrix for Mg adsorption reactions on biomass.
Process SCa2+
Ad29 [Mg-Xsu] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑠𝑢𝑀𝑔
Ad30 [Mg-Xaa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑎𝑀𝑔
Ad31 [Ni-Xfa] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑓𝑎𝑀𝑔
Ad32 [Ni-Xc4] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑐4𝑀𝑔
Ad33 [Ni-Xpro] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑝𝑟𝑜𝑀𝑔
Ad34 [Ni-Xac] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝑎𝑐𝑀𝑔
Ad35 [Ni-Xh2] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋ℎ2𝑀𝑔
Des29 [Ni-Xsu] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑠𝑢𝑀𝑔
Des30 [Ni-Xaa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑎𝑀𝑔
Des31 [Ni-Xfa] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑓𝑎𝑀𝑔
Des32 [Ni-Xc4] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑐4𝑀𝑔
Des33 [Ni-Xpro] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑝𝑟𝑜𝑀𝑔
Des34 [Ni-Xac] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝑎𝑐𝑀𝑔
Des35 [Ni-Xh2] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋ℎ2𝑀𝑔
Dec57 [Ni-FBXsu] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑠𝑢𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑠𝑢 𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑠𝑢𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑠𝑢𝑀𝑔 … )
Dec59 [Ni-FBXaa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑎𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑎𝑎 𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑎𝑎𝑀𝑔 … )
Dec61 [Ni-FBXfa] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑓𝑎𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑓𝑎𝑀𝑔 … )
𝑀𝑔
Dec63 [Ni-FBXc4] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑐4𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑐4𝑀𝑔⁄𝑋𝐹𝐵𝑥𝑐4𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑐4𝑀𝑔 … )
Dec65 [Ni-FBX pro] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑝𝑟𝑜 𝑀𝑔⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑀𝑔 … )
Dec67 [Ni-FBXac] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥𝑎𝑐𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥𝑎𝑐 𝑀𝑔⁄𝑋𝐹𝐵𝑥𝑎𝑐𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑎𝑐𝑀𝑔 … )
Dec69 [Ni-FBX h2] (s) Decay −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝐹𝐵𝑥ℎ2𝑀𝑔 ∙ (𝑋𝐹𝐵𝑥ℎ2𝑀𝑔 ⁄𝑋𝐹𝐵𝑥ℎ2𝑀𝑔 + 𝑋𝑂𝐵𝑥ℎ2𝑀𝑔 … )
Dec58 [Ni-OBXsu] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑠𝑢𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑠𝑢 𝑀𝑔⁄𝑋𝐹𝐵𝑥𝑠𝑢𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑠𝑢𝑀𝑔 … )
Dec60 [Ni-OBXaa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑎𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑎𝑎 𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑎𝑎𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑎𝑎𝑀𝑔 … )
Dec62 [Ni-OBXfa] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑓𝑎𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑓𝑎 ⁄𝑋𝐹𝐵𝑥𝑓𝑎𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑓𝑎𝑀𝑔 … )
𝑀𝑔
Dec64 [Ni-OBXc4] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑐4𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑐4𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑐4𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑐4𝑀𝑔 … )
Dec66 [Ni-OBXpro] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑝𝑟𝑜 𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑝𝑟𝑜𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑝𝑟𝑜𝑀𝑔 … )
Dec68 [Ni-OBXac] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥𝑎𝑐𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥𝑎𝑐 𝑀𝑔 ⁄𝑋𝐹𝐵𝑥𝑎𝑐𝑀𝑔 + 𝑋𝑂𝐵𝑥𝑎𝑐𝑀𝑔 … )
Dec70 [Ni-OBXh2] (s) Decay 𝑌𝑇𝐸 −1 𝐷𝑥 ∙ 𝑘𝑑 ∙ 𝑋𝑂𝐵𝑥ℎ2𝑀𝑔 ∙ (𝑋𝑂𝐵𝑥ℎ2𝑀𝑔 ⁄𝑋𝐹𝐵𝑥ℎ2𝑀𝑔 + 𝑋𝑂𝐵𝑥ℎ2𝑀𝑔 … )
S5: Stoichiometric matrix for Fe, Co, Ni, Ca, Mg adsorption reactions on inert.
Process SFe2+ SCo2+ SNi2+ SCa2+ SMg2+
Ad2 [Fe-X I] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐹𝑒2+ ∙ 𝑋𝐹𝐵𝑋𝐼𝐹𝑒
Des2 [Fe-X I] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝐼𝐹𝑒
Ad23 [Co-XI] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝑋𝐼𝐶𝑜
Des23 [Co-XI] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝐼𝐶𝑜
Ad24 [Ni-X I] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝑋𝐼𝑁𝑖
Des24 [Ni-X I] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝐼𝑁𝑖
Ad25 [Ca-X I] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑎2+ ∙ 𝑋𝐹𝐵𝑋𝐼𝐶𝑎
Des25 [Ca-X I] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝐼𝐶𝑎
Ad26 [Mg-XI] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑀𝑔2+ ∙ 𝑋𝐹𝐵𝑋𝐼𝑀𝑔
Des26 [Mg-XI] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝑋𝐼𝑀𝑔
S5: Stoichiometric matrix for Co and Ni adsorption reactions on FeS.
Components 46 144 145 44 146 147 Process Rate

Process SCo2+ SNi2+
Ad2 [Co-XI] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝐶𝑜2+ ∙ 𝑋𝐹𝐵𝐹𝑒𝑆𝐶𝑜
Des2 [Co-XI] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝐹𝑒𝑆𝐶𝑜
Ad23 [Ni-X I] (s) Adsorption −1 −1 1 𝑘𝑎/𝑑 ∙ 𝐾𝑎𝑑𝑠 ∙ 𝑆𝑁𝑖2+ ∙ 𝑋𝐹𝐵𝐹𝑒𝑆𝑁𝑖
Des23 [Ni-X I] (s) Desorption 1 1 −1 𝑘𝑎/𝑑 ∙ 𝐾𝑑𝑒𝑠 ∙ 𝑋𝑂𝐵𝐹𝑒𝑆𝑁𝑖

TH2019PESC2039

Uploaded by

Copyright:

Available Formats

TH2019PESC2039

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

TH2019PESC2039

Uploaded by

Copyright:

Available Formats

Joint PhD degree in Environmental Technology

Docteur de l’Université Paris-Est

Dottore di Ricerca in Tecnologie Ambientali

Degree of Doctor in Environmental Technology

Thesis for the degree of Doctor of Philosophy in Environmental Technology

PhD thesis –Väitöskirja – Proefschrift – Tesi di Dottorato – Thèse

Bikash Chandra Maharaj

To be defended on 10/12/2019, Paris

Prof. Patrice Chatellier Chairman, Reviewer

Prof. Lorenzo Bertin

Prof. Blaž Stres

Prof. Piet N. L. Lens

Prof. Jukka Rintala

Asst. Prof. Luigi Frunzo

Dr. Maria Rosaria Mattei

The thesis is comprised of the following four publications:

Publication I Bikash C Maharaj performed a descriptive literature survey, wrote the

Publication II Bikash C Maharaj carried out the mathematical model formulation,

Publication IV Bikash C Maharaj carried out the mathematical model formulation,

134 1.1 Thesis Overview

147 Publication I- Mathematical modelling of trace element dynamics in anaerobic digestion

220 AD is a collection of processes by which microorganisms, associated to each other by metabolic

244  it is a simple and sustainable method;

375 Figure 2. Deficiency, stimulation, optimal, inhibition and toxicity of TE concentration.

TE Mineral pKsp Reference

Log Stability Constants

Ethelyenediaminetetracetic acid (EDTA) EDTA-Metal Complex

Three organic ligands constitute:

(Propionate) The extent of Complex formation is strongly pH

(Sulphur containing amino acid)

639 Figure 5: Biochemical conversions in ADM1.

Batstone et al., 2003 First extention to consider precipitation of mineral CaCO3.

667 2.3 Modeling TE-associated physiochemical reactions in AD

668 2.3.1 Modeling precipitation processes

821 2.3.3 Modeling TE Adsorption processes in AD

864 2.3.4 Modeling TE-Biouptake in AD

890 Microbial growth rate was defined as:

961 Table 5. A general matrix representation of model

999 3.2.1 The mathematical model

+ Vliq (∑ αi,j ρbio,j 𝐼𝑖 (t, 𝐒, 𝐗) )

+Vliq (∑ γi,j ρprec,j (t, 𝐒, 𝐗 𝐩 ) + ∑ γi,j ρdissol,j (t, 𝐒, 𝐗 𝐏 )),

1004 n1 is the number of soluble components;

1007 m1 is the number of biochemical processes considered;

1008 m2 is the number of precipitation/dissolution processes considered;

1009 αi,j is the stoichiometric coefficient of species i on biochemical process j;

1010 γi,j is the stoichiometric coefficient of species i on precipitation/dissolution process j;

1012 Si is the 𝑖 𝑡ℎ soluble component, 𝐒 = (S1 , … , Sn1 );

1013 Xi is the ith particulate component, 𝐗 = (Xn1+1 , … , Xn2 );

1014 Xp,i is the ith precipitate, 𝐗 𝐩 = (Xp,1 , … , Xp,m2 );

1018 ρbio,j (t, 𝐒, 𝐗) is the rate for jth biochemical process;

1019 ρprec,j (t, 𝐒, 𝐗 𝒑 ) is the rate for jth precipitation process;

1020 ρdissol,j (t, 𝐒, 𝐗 𝒑 ) is the rate for jth dissolution process;

1021 Vliq is the volume of liquid in the reactor.

1022 Vgas is the volume of gas in the reactor

𝐼1 = 𝐼𝑝𝐻 ∙ 𝐼𝐼𝑁_𝑙𝑖𝑚 ∙ 𝐼𝑀𝑒 2+ , (3.5)

𝜌𝑇,𝑖 = 𝑘𝑙 𝑎(𝑆𝑙𝑖𝑞,𝑖 − 𝐾𝐻,𝑖 𝑃𝑔𝑎𝑠,𝑖 ) (3.9)

𝜌𝐴⁄ = 𝑘𝐴⁄ (𝑆𝑙𝑖𝑞.𝐴− ∙ 𝑆𝑙𝑖𝑞.𝐻 + − 𝐾𝑎 ∙ 𝑆𝑙𝑖𝑞.𝐻𝐴 ) (3.10)

𝑆𝐴𝐶 − 𝑆𝑃𝑅𝑂 − 𝑆𝐵𝑈 − 𝑆𝑉𝐴−

1064 3.2.2 Model implementation and simulations

1102 3.3.1 The mathematical model