Algal Research 44 (2019) 101667

Contents lists available at ScienceDirect

Algal Research
journal homepage: www.elsevier.com/locate/algal

Optimization of algae production on urine T

a a a b a
Kanjana Tuantet , Hardy Temmink , Grietje Zeeman , René H. Wijffels , Cees J.N. Buisman ,
Marcel Janssenb,*
a
Department of Environmental Technology, Wageningen University, the Netherlands
b
Bioprocess Engineering, AlgaePARC, Wageningen University, the Netherlands

A R T I C LE I N FO A B S T R A C T

Keywords: Urine is a potential source of nutrients to grow microalgal biomass to be re-used as fertilizer and soil conditioner.
Microalgae production In this study the impact of photobioreactor dilution rate on microalgae productivity and photosynthetic effi-
Nutrient removal ciency was assessed and used to determine operating conditions to reach both full nitrogen removal from urine
Urine treatment and high biomass productivity. In addition, the possibility to work under day/night cycling was tested. To this
Photobioreactor control
end, the microalga Chlorella sorokiniana was grown on artificial urine and real human urine in bench-scale panel
photobioreactors with short optical paths.
At a light intensity of 1530 μmol⋅ m−2⋅s-1 photobioreactor productivity and photosynthetic efficiency was
demonstrated to be maximal at reactor dilution rates between 0.10 and 0.15 h-1. A biomass yield of 1 g dry
matter per mol of PAR photons was achieved. Biomass concentration, and accordingly nutrient removal effi-
ciency, decreased at increasing reactor dilution rate. The experimental results could be reproduced by model
simulations. These simulations allowed to demonstrate that the system must be operated at a dilution rate of less
than 0.01 h-1 in order to reach complete nitrogen removal. In that scenario more than half of the potential
biomass productivity is lost due to severe self-shading within the algal culture. Experiments with real human
urine illustrated the problem of incomplete nitrogen removal and ammonium inhibition of growth at too high
dilution rates. It is therefore suggested to apply an optimized pre-dilution of pure urine prior to treatment in a
photobioreactor.
Experiments under day-night cycles demonstrated that microalgal cultures quickly acclimate to such variable
light conditions. Additional model simulations illustrated that a phototrophic system is most effective when
diluted urine is fed to the photobioreactor during day time only. In that situation the lowest nitrogen con-
centration in the effluent can be reached at a maximal areal removal rate and photosynthetic efficiency.

1. Introduction urine accounts only for 1% of the volume of household wastewater but
contains 69% of the nitrogen load and 40% of the phosphorus load
Nutrients need to be removed from wastewater in order to prevent [11].
eutrophication of surface waters and consequent blooms of algae or A previous study by Tuantet et al. [12] demonstrated the feasibility
cyanobacteria. The same microbes therefore could offer a way for nu- of minimally diluted human urine for microalgae cultivation. It was
trient removal in contained treatment systems. As such, the nutrients possible to cultivate the microalga Chlorella sorokiniana on urine that
are captured within the algal biomass and can be re-used, for example was only diluted by a factor of two. It was demonstrated that a short
as fertilizer and soil conditioner [1,2]. For this reason the combination light path was a prerequisite to grow algae on concentrated urine. Urine
of microalgal biomass production with wastewater treatment has hydrolysis appeared to be a complicating factor as it is accompanied by
gained a lot of attention during the last decade [3–8]. In our study we a pH increase inducing precipitation of struvite (magnesium ammo-
focussed on household wastewater and we started from the point of nium phosphate). Precipitation of this struvite led to a depletion of
view that source separation of different types of wastewaters opens new magnesium in the liquid phase of hydrolysed urine and microalgae did
avenues for efficient treatment steps [9,10]. Source-separated human not grow or grew slowly [12]. Supplementation of magnesium therefore
urine is very rich in nutrients and therefore could serve as growth is mandatory to support optimal microalgal growth. Fresh urine, on the
medium for microalgae cultivation. More specifically, source-separated contrary, was suitable for the cultivation of microalgae.

⁎
Corresponding author at: Bioprocess Engineering, Wageningen University and Research, P.O. Box 16, 6700AA, Wageningen, the Netherlands.
E-mail address: [email protected] (M. Janssen).

https://doi.org/10.1016/j.algal.2019.101667
Received 22 February 2019; Received in revised form 17 September 2019; Accepted 17 September 2019
2211-9264/ © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
K. Tuantet, et al. Algal Research 44 (2019) 101667

Nomenclature ms Maintenance requirement for sugar, (mols⋅molx−1⋅s−1)

Mx Molecular mass of biomass, (gx⋅molx−1)
Abbreviations qsm Maximal specific rate of photosynthetic sugar production,
(mols⋅molx−1⋅s−1)
HRT Hydraulic retention time rNarea Areal nitrogen removal rate, (mgN⋅ m−2⋅ h-1 or gN⋅ m−2⋅d-
1
PAR Photosynthetic active radiation, 400–700 nm )
PBR Photobioreactor rxarea Areal biomass production rate, (gx⋅ m−2⋅ h-1)
rxvol Volumetric biomass production rate, (gx⋅L−1⋅ h−1)
Symbols SVR Surface to volume ratio PBR, (m−1)
TN Total nitrogen concentration wastewater, (mgN⋅L−1)
μ Specific growth rate, (h−1) VR Volume PBR, (m3)
acf Photoacclimation factor, (-) xN Molar nitrogen fraction biomass, (molN⋅molx−1)
AR Illuminated area PBR, (m2) xP Molar phosphorus fraction biomass, (molP⋅molx−1)
ax Specific light absorption coefficient, (m2⋅molx−1) Ysphm Maximal photosynthetic yield of sugar on PAR photons,
CN Dissolved nitrogen concentration, (mgN⋅L−1 or molN⋅ m-3) (mols⋅mol ph−1)
Cx Biomass concentration, (gx⋅L−1 or molx⋅ m-3) Yxph Observable biomass yield on PAR photons,
COD Chemical oxygen demand wastewater, (mgO2⋅L−1) (molx⋅mol ph−1)
D Dilution rate, (h−1) Yxs Biomass yield on photosynthetically produced sugar,
Enλ Normalized spectral distribution light source, ( nm−1) (molx⋅mols−1)
F Liquid flow rate, (L⋅ h−1)
Iph(z) PAR photon flux density (i.e. light intensity) at location z Subscripts
in PBR, ((μ)molph⋅ m−2⋅s-1)
Iphin PAR photon flux density at surface PBR (i.e. incident light x Biomass
intensity), ((μ)molph⋅ m−2⋅s-1) ph PAR photons
Iphin24 Daily averaged PAR photon flux density, (molph⋅ m−2⋅d-1) s Sugar (triose)
KsN Saturation constant for nitrogen uptake, (mgN⋅L−1 or N Nitrogen
molN⋅ m-3) P Phosphorus
LPenh Light path enhancement factor, (-) λ Wavelength dependent parameter or variable

Also other researchers have demonstrated the potential of urine as a concentrations are required to guarantee maximal nutrient removal
nutrient source for microalgae cultivation [13–15]. In a follow up study efficiency (≈ 100%) since the nutrients removed from the water are
by Tuantet et al. [16] the algae production process on urine was op- immobilized within the biomass formed. High biomass concentrations,
erated under non-sterile conditions for 270 days, demonstrating the however, will increase the dark volume of a photobioreactor leading to
robustness of the process. In this study a flat panel photobioreactor a reduction in biomass production and photosynthetic efficiency. This
reactor was applied with an optical path of effectively 5 mm illumi- will result in a larger area required for a photosynthetic process.
nated at solar light levels of 1500 μmol⋅ m−2⋅s-1 and a hydraulic re- In this study the relation between photobioreactor dilution rate,
tention time (HRT) of 1 day. Under these conditions full phosphorus biomass density, reactor productivity and photosynthetic efficiency was
removal was obtained with only 3 times diluted urine. Nitrogen re- determined for the microalga Chlorella sorokiniana growing on urine.
moval efficiency was 80% and the organic matter present in urine Chlorella sorokiniana is a mesophilic alga characterized by a relatively
(expressed as chemical oxygen demand, COD) was degraded for 60%, high optimal temperature (30–40 °C) and maximal specific growth rate
most likely by a symbiotic bacterial population. Areal biomass pro- (≈ 0.27 h−1) [23]. It grows very well on urine [12,16], and a mathe-
ductivity was 6.2 g⋅ m−2⋅ h-1 corresponding to a photosynthetic effi- matical model is available to predict this growth under light limitation
ciency of 0.55 g⋅ mol-1. This photosynthetic efficiency is relatively low [24]. Experimental results were compared to model simulations of
for Chlorella sorokiniana as efficiencies of 1 g⋅ mol-1 have been achieved photobioreactor productivity as a function of dilution rate. The com-
in comparable systems [17]. It was suspected that a HRT of 1 day is too bined results allow for balancing the conflicting interests of high ni-
long to warrant maximal biomass productivity. trogen removal efficiency and high biomass productivity in relation to
The productivity of photobioreactors depends on the light intensity, urine pre-dilution. Most experiments were performed with synthetic
temperature, and the reactor dilution rate. Considering outdoor solar urine, but a limited number of conditions were also tested employing
production conditions the light intensity falling on the reactor is a fact real human urine. In order to assess the applicability of microalgae
of life and cannot be controlled. The dilution rate, however, can be based urine treatment under solar conditions the process was also va-
directly controlled determining the biomass concentration and, as such, lidated under a day-night cycle regime. Additional model simulations
the light regime within the culture. In wastewater related processes the were then applied to identify suitable photobioreactor operating stra-
reactor dilution rate (D) is usually converted into the hydraulic re- tegies under solar conditions.
sidence time (HRT = 1/D) In a dense microalgal culture (obtained at a
long HRT) a light gradient will develop with maximal light intensity at
2. Materials and methods
the reactor surface to almost complete darkness on the opposing side
[18]. The relative magnitude of the dark zone determines the amount of
2.1. Microorganism and culture media
biomass which is lost by cellular maintenance or endogenous respira-
tion [19]. This is also the reason for any photobioreactor to reach a
maximal biomass concentration when operated in batch under light Chorella sorokiniana CCAP211/8K was obtained from the Culture
limited conditions [20,21]. The relation between reactor dilution, Collection of Algae and Protozoa, Oban, UK. The microalgae were pre-
biomass density, and reactor productivity has been studied in the past cultivated in 250-ml enclosed shake flasks with 100 ml of M8a medium
[25]. The cultures were maintain in a shaking incubator at 25 °C, a light
[22], but not in relation to nutrient removal from wastewater. In this
application a conflict of interest is expected. High biomass intensity of 20–40 μmolph⋅ m−2⋅s-1, and a 16/8 h day/night cycle. A few
days before inoculation, the cultures were placed in an incubator with

2
K. Tuantet, et al. Algal Research 44 (2019) 101667

continuous illumination of 165 μmolph⋅ m−2⋅s-1 and 2%-CO2 enriched pressure sodium lamps (MIG 400 W, LEG Illumination, Italy) placed on
air in order to reach higher cell density. both sides of the reactor surface. Before and after the experiments, the
Both synthetic urine and real human urine were used. Synthetic light intensity was measured over 26 spots distributed homogeneously
urine composition was as follows (amounts per L): 15 g CO(NH2)2 inside the empty growth chamber of the reactor using a 2π PAR
(urea), 2.75 g K2HPO4, 0.75 g Na2HPO4.2H2O, 0.53 g CaCl2.2H2O, quantum sensor (LI-190SA, LiCOR, USA) connected to a Li-COR hand-
1.23 g MgSO4.7H2O, 1.4 g K2SO4, and 8 g NaCl. To support dense mi- held meter (LI-250A, Li-COR, USA). The reported light intensity was
croalgae cultures, 30 ml Fe-EDTA stock solution (13 ml for day-night averaged over the reactor surface. The reactor inner surface was
experiments), and 15 ml micronutrient stock solution were supple- cleaned daily to remove biofilms using a teflon-coated magnetic stirring
mented (6.5 ml for day night experiment). Fe-EDTA stock solution bar placed within the reactor.
consisted of (amounts per L stock solution) 58 g EDTA ferric sodium salt During the experiments, temperature was controlled at 38 °C and
and 18.6 g Na2EDTA.2H2O. Micronutrient stock solution consisted of pH was maintained at 7.0 ± 0.2. Air enriched with CO2 (at different
(amounts per L of stock solution) 13 g MnCl2.4H2O, 3.2 g ZnSO4.7H2O, concentrations) was fed into the reactor by means of mass flow con-
1.83 g CuSO4.5H2O, and 0.062 g H3BO3. The composition of this syn- trollers to supply enough inorganic carbon for microalgae growth and
thetic urine was comparable to averaged urine concentration reported to provide mixing. For each experiment the CO2 concentration was
by Kujawa-Roeleveld and Zeeman [11]. The pH of synthetic urine was adjusted to minimise the amount of acid needed for pH control. The
adjusted to 7 with 3 M HCl prior to application and 0.032% (v/v) of actual CO2 concentrations used at different dilution rates are given in
antifoam B silicone emulsion (J.T.Baker, Germany) was added to pre- Table 1, and ranged between 8 and 20% v/v. The averaged light in-
vent foaming. tensity over both reactor surfaces was 1530 μmolph⋅ m−2⋅s-1. This light
Synthetic urine was diluted 1.8–10 times to avoid inhibition by intensity represents photons in the so-called PAR (Photosynthetic Ac-
ammonia, which is formed when urea is hydrolysed. Ammonium was tive Radiation) ranging from 400 to 700 nm, the wavelength range
observed to inhibit microalgal growth at concentrations of 1440 which can be used by microalgae.
mgN⋅L−1 or higher12. At the same time, dilution was also minimized to For the experiment performed under day-night conditions the
a level which was expected not to result in any nutrient limitation. The photobioreactor was illuminated from one side only with a different
dilution factors applied to synthetic urine are shown in Table 1. The light source. This light source (type Reallight 24, Technical
dilution factor is defined as the ratio of the volume of diluted urine to Development Studio, Wageningen UR, NL) was equipped with 24
the original aliquot volume of non-diluted urine, and hence a dilution warm-white LEDs (Bridgelux BXRA W1200). Light intensity was con-
factor of 1 is equivalent to non-diluted urine. trolled with pulse-width-modulation (PWM) at a frequency of 12 KHz.
Human urine was collected from separation toilets (Gustavsberg, The change in light intensity during the day was automatically con-
Berger Biotechnik GmbH, Germany) receiving urine from male and trolled based on a half sine curve resulting in a day length of 12 h and a
female employees of the Department of Environmental Technology, peak light intensity of 1340 μmolph⋅ m−2⋅s-1.
Wageningen University, the Netherlands. Since an amount of flush
water is used in those toilets, the urine was less concentrated than the 2.3. Experimental procedures
synthetic urine, as will be shown later in the results. The urine was
maintained at 4 °C and darkness during storage and experiments. A For the experiments under constant illumination with synthetic
volume of 1 l urine was supplemented with 30 ml of Fe-EDTA stock urine two experimental series with different microalgal inocula were
solution and 15 ml of micronutrient stock solution, which was also conducted. Both series were started in batch mode by inoculating mi-
applied for the synthetic urine. The urine was also supplemented with croalgae in diluted synthetic urine at low light intensity utilising one
0.032% (v/v) of antifoam B silicone emulsion (J.T.Baker, Germany) to lamp only and shading it with light-diffusing glass plates. As soon as the
prevent foaming. In addition, magnesium was supplemented to real biomass density started to increase, the light intensity was gradually
urine. A stock solution of 8 g⋅L−1 of MgSO4.7H2O was pumped directly increased by removing the glass plates and eventually the second lamp
into the reactor at 1.8 ml⋅ h−1 using a diaphragm metering pump
(STEPDOS® 03 RC, Germany). Table 1
Overview continuous chemostat experiments with synthetic and real urine.
2.2. Photobioreactor and general operating conditions Culture Dilution Aeration rate CO2 Urine Steady
medium rate (h−1) (L⋅L−1⋅ min−1) content dilution state
A panel photobioreactor with a light-path of 10 mm was used. The (%) factora period
illuminated surface area of each reactor side was 0.0903 m2 and the (days)

working volume of the reactor was 0.9 L. The cultivation system (Fig. 1, Continuous light at 1530 μmolph⋅ m−2⋅s-1from both sides; T = 38 °C, pH = 7.0 ± 0.2
and schematic in Appendix 1) was composed of 6 parts: the photo- Synthetic 0.04 0.68 20 2.5 7
bioreactor, influent and effluent pumps, a temperature control system, urine 0.05 0.69 16 1.8 9
a pH control system, an aeration system, and a light source. 0.07 0.68 20 3.0 10
0.10 0.67 11 3.3 4
The photobioreactor consisted of a cultivation chamber and a water
0.15 0.70 12 4.8 5
jacket for temperature control. In order to prevent evaporation a con- 0.20 0.69 11 7.4 3
denser was installed at the gas outlet on top of the cultivation chamber, 0.24 0.69 8 8.5 3
which was cooled with water of 4 °C. Peristaltic pumps (Watson Marlow Human 0.05 0.67 19 Non- 9
b
120U, Watson-Marlow pumps, UK) were used to feed and withdraw the urine diluted
0.11 0.67 19 Non- 6
influent and effluent. The influent and effluent bottles were placed at diluted
4 °C in darkness to avoid further urine hydrolysis. The culture tem- Experiment under 12 h/12 h day-night cycle at 1340 μmolph⋅ m−2⋅s-1from one side;
perature was controlled via a temperature sensor connected to a water T = 38 °C, pH = 6.9 ± 0.3
bath (Julabo-F25-HE, Julabo Labortechnik GMBH, Germany) circu- Synthetic 0.016 0.57 17.2 3 3
urine
lating water through the water jacket. The culture pH was controlled
with concentrated hydrochloric acid and sodium hydroxide solutions. A a
Dilution factor is defined as the ratio between the volume of diluted urine
gas distribution tube was placed above the bottom of the reactor over volume and the original volume of non-diluted urine (i.e. a dilution factor of 3
two thirds of the reactor width. This ‘partial’ gassing created an air-lift means 1 aliquot urine is diluted with 2 aliquots demineralised water).
mixing resulting in a well-mixed liquid without any stagnant zones. b
Human urine was supplemented with trace elements and magnesium (see
The photobioreactor was continuously illuminated by two high- text).

3
K. Tuantet, et al. Algal Research 44 (2019) 101667

Reported values are averages of the samples taken over these three
days.

2.4. Analytical procedures

The samples taken were partly used for the analyses of optical
density and biomass dry weight concentration (Cx). The remaining
sample was centrifuged at 10,000 rpm for 10 min to separate algal
biomass from the water. The cell-free supernatant was stored at 4 °C
and later used for analyses. The optical density (OD) was measured at
680 and 750 nm using a Xion 500 spectrophotometer (Hach Lange,
Germany). Dry weight was determined by collecting suspended biomass
on Whatman GF/F glass microfiber filters (Ø 55 mm pore size 0.7 μm).
Pre-washed and pre-dried filters were first weighed. Aliquots of 5 ml of
culture broth withdrawn from the reactor were diluted 15 times with
pre-filtered, demineralised water. The diluted samples were subse-
quently filtered followed by washing with 50 ml of pre-filtered, demi-
neralised water to remove adhering inorganic salts. Filters were dried
overnight at 105 °C and cooled in a desiccator for a minimum of 2 h.
The dry weight was expressed as the difference in weight of the dry
filters and the dry filters containing microalgae.
The supernatant was analysed for total nitrogen (TN) and ortho-
phosphate (PO43−P) for the experiments carried out with synthetic
urine. Supernatant from real urine samples was analysed for total ni-
trogen (TN), ammonium nitrogen (NH4+-N), total phosphorus (TP) and
the chemical oxygen demand (COD). All of the analyses were done
photo-metrically according to Standard Methods (APHA, 1998) using
Dr Lange® test kits (Hach Lange GMBH, Germany).

Fig. 1. Photograph of laboratory scale panel photobioreactor, illumination and 2.5. Calculations
auxiliary equipment.
The volumetric biomass productivity (rxvol) is the product of the
was employed illuminating the opposite side of the panel reactor. This biomass dry weight concentration (CX, gx⋅L−1), and reactor dilution
was followed by starting continuous chemostat operation at low dilu- rate (D, h−1), provided that the bioreactor is at steady state and no
tion rate. In total 7 different reactor dilution rates were tested during accumulation of biomass takes place in the reactor, and was calculated
both experimental series (see Table 1). according to Eq. 1.
After finishing the experiments with synthetic urine, a new experi- rxvol = C x ∙D (1)
ment with human urine was started with a new inoculum. The same
−2
starting-up procedure was performed as with synthetic urine. When the The areal productivity (Pxarea, gx⋅ m ⋅ h ) was calculated based on
-1

second lamp was turned on, human urine was pumped to the reactor at the volumetric productivity and the reactor dimensions (i.e. surface
a low dilution rate together with synthetic urine so that microalgae area to volume ratio, SVR, m-1) according to Eq. 2.
could gradually adapt to real urine. During 10 days the ratio between rxvol ∙1000
human urine and synthetic urine was gradually increased until solely rxarea =
SVR (2)
human urine was supplied. The applied reactor dilution rates of real
urine experiment were 0.11 and 0.05 h−1 (Table 1). The observed biomass yield on light energy (Yxph, gx⋅mol ph−1) is
The average dilution rate was checked daily by weighing the in- used to express the photosynthetic efficiency. It was calculated based on
fluent vessel, and the bottles containing acid, base, and magnesium the biomass concentration (CX, gx⋅L−1), liquid flow (F, L d−1), daily
solutions. Samples of the microalgal culture were taken during steady integral light intensity (Iphin24, mol⋅ m-2⋅d−1), and reactor surface area
state, a period having a constant biomass density for a duration of at (AR, m2) according to Eq. 3.
least 3 times the applied hydraulic retention time (HRT, i.e. the inverse C x ∙F C x ∙F
Yxph = =
of the dilution rate). At low dilution rates (0.04-0.10 h−1), samples Iphin24 ∙AR Iphin ∙10−6∙24∙3600∙AR (3)
were taken every 24 h. At high dilution rates (0.15-0.24 h−1), two
samples were taken per day. At least three replicate samples were taken The light intensity (Iphin) averaged over the illuminated surface area
for every dilution rate. was 1530 μmolph⋅ m−2⋅s-1 for the experiments under constant illumi-
For the experiment under the day-night cycle the photobioreactor nation. The photobioreactor width was 10 mm and illuminated from
was inoculated with a new algal inoculum. The experiment was started both reactor sides, the photobioreactor SVR became 200 m-1. The total
in batch mode under low and constant light. After three days when the reactor surface area (both sides) was 0.181 m2. For the experiment
biomass concentration was 3.4 gx⋅L−1 continuous dilution and the day- under day-night light regime the reactor was illuminated form one side
night light regime were initiated. The culture was continuously diluted only (SVR =100 m-1 and AR = 0.091 m2) according to a half sine cure
day and night at a fixed rate of 0.016 h−1 (HRT = 2.6 days). The in- with period 12 h and peak intensity of 1340 μmolph⋅ m−2⋅s-1. the daily
fluent vessel, effluent vessel, acid and base bottles were weighed every integral light intensity (Iphin24) then became 36.85 molph⋅ m−2⋅d-1.
day to calculate and check the dilution rate. A stable performance of the
microalgal culture was reached after running the photobioreactor for 2.6. Model simulations algal growth in photobioreactors
seven day-night cycles. Following this stabilization period culture
samples were taken every two-hours for a period of three days. Microalgal growth, photobioreactor productivity, and nutrient re-
moval during the experimental conditions tested was described with

4
K. Tuantet, et al. Algal Research 44 (2019) 101667

model simulations. This model allowed for an analysis of the relation 0.1 which is equivalent to a minimal quantum requirement of 10 for the
between light-limited algae production in photobioreactors and reactor fixation of CO2 and the production of O2. The parameter Yxs represents
dilution rate at optimal temperature. It must be noted that inclusion of the yield of algal biomass on sugar, where again biomass is normalized
temperature dependency is required in order to use this model under to 1 mol of carbon. Parameter ms represents the maintenance require-
outdoor conditions with variable temperature. The model employed ment for sugar. The values of the model parameters are presented in
was identical to the one developed for Chlorella sorokiniana by Blanken Table 2. More details of this model can be found elsewhere [18,24].
et al. (2016) [24] and Janssen (2016) [18]. This model can only be The model of Blanken was adapted to describe the experimental
applied to flat panel reactors as the local light intensity Iphλ(z) is cal- results better and to include photoacclimation which has been de-
culated assuming light transfer over a single dimension z based on scribed in detail by de Mooij et al. (2017) 27. It was reported that at low
Lambert Beer. light intensity, and corresponding low specific growth rate, the specific
light absorption coefficient ax is 2.2 to 2.5 times higher than at a high
Iphλ (z) = Iphinλ ∙e−axλ∙Cx ∙ LPenh ∙z (4)
light intensity and specific growth rate. Considering that in the current
The incident light intensity at the reactor surface is designated as study specific growth rate μ was controlled by fixing the reactor dilu-
Iphinλ and the biomass specific light absorption coefficient is represented tion rate D the process of photoacclimation could be directly related to
by axλ. Parameter LPenh has a value between 1 and 2 which represents μ by means of an inverted sigmoid function. Parameters were chosen
the lengthening of the light path because of non-perpendicular light such (Table 2) that the acclimation from a low light (LL) to a high light
entrance and forward light scattering by the microalgae cells. This phenotype gradually occurs when the specific growth rate increases
parameter needs to be fitted to experimental data. Out-scattering of from 0.05 h−1 to 0.1 h−1.
light at the front side of the panel is thus neglected. The spectral de- − c∙μ
pendence of light intensity and light absorption is taken into account ax = axLL ∙ acf and acf= 2.4 − 1.4∙e− b∙ e (9)
with a spectral resolution of 1 nm. In order to assess nitrogen and phosphorus removal rates it was
700 assumed that the nitrogen and phosphorus quota in the biomass were
Iph (z) = ∑ Iphinλ ∙e−axλ∙Cx ∙LPenh ∙z∙Δλ constants (Table 2) and that no other processes occurred other than
400 (5) assimilation of nitrogen and phosphorus in the algal biomass.
Iphinλ = Enλ ∙Iphin (6) Algal growth was simulated for a flat panel photobioreactor iden-
tical to the lab-scale system. Simulations were done for a range of di-
Variable Iphin represent the incident light intensity integrated over lution rates under constant light and a limited number of day-night
the PAR range. The parameter Enλ (unit nm−1) represents the PAR cycles. Identical light regimes were chosen as those applied at lab-scale.
normalized emission spectrum of the light source used. The spectra The simulation was founded in the numerical solution of the biomass
used for a high pressure sodium lamp and a warm-white Bridgelux LED balance over the reactor by means of PTC Mathcad version 15 em-
were taken from Blanken et al. (2016) [24]. ploying the ODEsolve routine with a fixed step size. The mass balances
Photosynthesis is defined as solely sugar (triose) production in the are based on an ideally mixed liquid phase.
chloroplast (qs) and is described by the model of Jassby and Platt [26].
Photosynthesis is followed by the formation of functional biomass (i.e. dC x (t)
= (μ(C x (t),t) − D(t)) ∙C x (t)
dt (10)
growth μ) described by the Pirt relation.
700 For the simulations under day night cycles also nitrogen limitation
⎛ Ysphm∙∑400 axλ ∙Iphλ (z) ∙Δλ ⎞ was included as a Monod relation in addition to light limitation. This
q s (z) = q sm ∙tanh ⎜ ⎟
q sm (7) resulted in an extended growth equation including a saturation constant
⎝ ⎠
for nitrogen KsN.
d
μ= ⎛ ∫0 q s (z) ∙dz − m s ⎞ ∙Yxs
(8) CN d
⎝ ⎠ μ= ∙⎛
CN + K sN ⎝
∫0 q s (z) ∙dz − m s ⎞ ∙Yxs
⎠ (11)
In order to calculate the specific growth rate μ of the algal culture
the local rates of photosynthesis are first integrated over the optical In addition, the nitrogen balance was included in the numerical
depth d. The parameter qsm represents the maximal specific rate of simulation of microalgal growth in the photobioreactor.
photosynthesis (i.e. production of triose sugar) inside the chloroplast.
dCN (t)
The parameter Ysphm represents the maximal yield of triose on photons = D(t) ∙ (CNin − CN (t)) − μ(CN (t), C x (t), t) ∙C x (t) ∙xN
dt (12)
where triose is normalized to 1 mol of carbon. We assumed a value of

Table 2
Model parameter used for simulating growth Chlorella sorokiniana in panel photobioreactors.
Parameter Value References / comments

−1
Mx 24 g⋅molx Blanken et al. (2016) [24]
qsm 1.3⋅10−4 mols⋅molx-1⋅s-1
Ysphm 0.1 mols⋅mol ph−1
Yxs 0.59 molx⋅mols−1
ms 2.5⋅10−6 mols⋅molx-1⋅s-1
ax (high light) 3.29 m2⋅molx−1 PAR average; ax(λ) is reported at 1 nm resolution for PAR range by de Mooij et al. (2017) [27]
ax (low light) 7.44 m2⋅molx−1
acf −c ∙ μ See text for more details. Factors 2.4 and -1.4 are based on maximal and minimal absorption coefficient measured by de Mooij et al.
acf = 2.4 − 1.4∙e−b ∙ e
(2017) [27]
b = 50
c = 70
LPenh 1.5 Fitting parameter describing lengthening of light path due to non-perpendicular light entrance, and scattering of light within the
culture
xN 0.14 molN⋅molx−1 Kliphuis et al. (2012) [28]
xP 0.011 molP⋅molx−1
KsN 0.1 molN⋅ m−3 Higher than published data, but smallest number still allowing for numerical integration.

5
K. Tuantet, et al. Algal Research 44 (2019) 101667

All simulations were started at a biomass concentration between 50 more light saturation and light dissipation at low reactor dilutions rates
and 500 molx⋅ m−3 (1.2–12 gx⋅L-1) and performed for 20 days, which [27]. At dilution rates above 0.2 h−1 the biomass concentration is not
was sufficient to reach a steady state with constant biomass con- high enough to absorb all incident light.
centration for chemostats at constant illumination. In the case of day- The observed relation between reactor dilution rate, biomass con-
night cycles a repetitive cycle of biomass evolution in time was reached centration, and photosynthetic efficiency could be reproduced by model
within 20 days. Based on the steady state model results the same per- simulations (Figs. 2 and 3). This mathematical model includes the most
formance indicators could be derived as those calculated for the ex- relevant biological and physical processes determining microalgal
periments: Pxvol, Pxarea, Yxph. In addition to those parameters also the growth in photobioreactors. The effects of cellular maintenance and
areal nitrogen removal rate (rNarea unit gN⋅ m-2⋅d-1) and nitrogen re- light saturation discussed above are also included. The model slightly
moval efficiency were calculated. underestimates the measured biomass concentration and photo-
synthetic efficiency and it could be further improved to describe the
3. Results and discussion results even better. Specifically the light model (Lambert Beer) and the
description of photoacclimation could be improved. However, this
3.1. Photobioreactor dilution rate, biomass productivity and photosynthetic would require additional measurements and mathematical techniques,
efficiency and is beyond the scope of this study. The current model is considered
to be a good tool for the analysis of microalgae based nutrient removal
The panel photobioreactor was operated as a chemostat, starting of urine, and wastewaters in general.
with a dilution rate of 0.04 h−1 and ending at 0.24 h−1 Table 4. Higher
dilution rates were not possible because the maximal growth rate of
3.2. Photobioreactor dilution rate and nutrient removal from urine
Chlorella sorokiniana is 0.27 h−1 and when approaching this value the
system becomes unstable. Synthetic urine with a nitrogen content of
A limited number of experiments with real urine were performed.
6993 mgN⋅L−1 (15 g⋅L−1 of urea) was used as a cultivation medium.
The human urine collected for this study was of a relatively dilute
Depending on the dilution rate the synthetic urine was pre-diluted to
nature with a total nitrogen (TN) concentration of 2626 mgN⋅L−1. The
prevent ammonium/ammonia inhibition. These pre-dilution factors
urine also contained phosphate at a concentration of 146 mgP⋅L−1 and
were already presented in Table 1. The microalgae were growing under
contained organic pollutants at a COD (chemical oxygen demand) level
light limited conditions as nutrient and carbon dioxide were supplied in
of 3270 mgO2⋅L−1. The urine was fed undiluted to the 10 mm panel
excess and the algae were grown at their optimal temperature (38 °C)
photobioreactor which was illuminated from both sides, identical to the
and pH range (6.8–7.2)
experiments with synthetic urine discussed above. Three different di-
Under light limited conditions the steady state biomass concentra-
lution rates were tested and the results are presented in Figs. 2 and 3.
tion was expected to decrease when increasing the dilution rate. This is
An overview of the removal characteristics can be found in Table 3.
indeed what was observed where the biomass concentration drops from
At the lowest dilution rate tested of 0.05 h−1 the system performed
16 g⋅L−1 at a dilution rate of 0.05 h−1 to 4 g⋅L−1 at a dilution rate of
very well. Biomass concentration and biomass yield on light were
0.25 h−1 (Fig. 2). When the dilution rate is increased the microalgae
higher in comparison to the application of synthetic urine, and also
are forced to grow at higher specific rates. To support such higher
higher than the model simulations (Figs. 2 and 3). Nitrogen and COD
specific growth rates light must be able to penetrate further in the
removal efficiency were 71 and 75% respectively. Phosphate removal
culture which corresponds to lower biomass concentrations.
efficiency was 100%. These results illustrate that the N:P ratio in urine
The steady state biomass concentration is important with respect to
is too high to allow for complete removal of nitrogen. This was already
microalgae based nutrient removal from wastewaters. The mechanism
studied previously16. It must be noted that in the synthetic urine ex-
of nutrient removal is the anabolic conversion of nitrogen and phos-
periments we always made sure that phosphate was never limiting by
phorus into cellular building blocks (e.g. amino acids) and hence the
minimizing the urine pre-dilution and checking phosphate levels during
immobilization of nitrogen and phosphorus within the algal biomass.
the different experiments (Table 1). The COD removal is most likely
Based on a typical nitrogen mass fraction of 8.2% w/w [28] a steady
state biomass concentration of 16 g⋅L−1 is accompanied by a nitrogen
removal capacity of 1312 mgN⋅L-1. This implies, however, that at a di-
lution rate as low as 0.04 h-1 full nitrogen removal of undiluted urine
with a TN concentration of 6993 mgN⋅L-1 cannot be accomplished. The
biomass concentration in this photobioreactor is already relatively high
for microalgal cultures, even for reactors with a short optical path of
5 mm [29]. To achieve even higher biomass concentrations and a better
nitrogen removal efficiency from urine even shorter optical paths are
needed or, alternatively, the urine must be pre-diluted.
In a light limited process the efficiency of light use directly de-
termines reactor productivity and nutrient removal rate. In other
words, photosynthetic efficiency, photobioreactor productivity, and
nutrient removal rates are linearly correlated. In Fig. 3 the photo-
synthetic efficiency is shown for the different experiments as the yield
of biomass dry weight gained on the absorption of one mole of PAR
Fig. 2. Biomass concentration Cx as a function of PBR dilution rate D at steady
photons. This metric allows for easy comparison with other studies. The state. Blue diamonds: experiments with pre-diluted synthetic urine; orange
biomass yield on light, and thus photobioreactor productivity, is max- triangles: experiments with real human urine TN = 2626 mgN⋅L−1). Other
imal between 0.1 and 0.15 h−1 dilution rate and reaches a value of 0.97 conditions: Chlorella sorokinina, flat panel PBR, 1 cm deep, double-sided illu-
gx⋅mol ph−1. This yield is comparable to, or higher than, those reached mination at 1530 μmolph⋅ m-2⋅s−1, T = 38 °C, pH = 7. Dashed blue line re-
in other studies performed with Chlorella sorokiniana [17,25,30]. At presents the model simulations under experimental conditions. Error bars re-
lower dilution rates the photosynthetic efficiency decreases because of present the standard deviation of the consecutive daily measurements during
an increased fraction of metabolic energy required to maintain a denser the steady state period as reported in Table 1. (For interpretation of the re-
algal culture. In addition, algal pigmentation is expected to increase at ferences to colour in this figure legend, the reader is referred to the web version
lower specific growth rates (i.e. photoacclimation) possibly resulting in of this article.)

6
K. Tuantet, et al. Algal Research 44 (2019) 101667

to negatively impact algal growth. In a previous study it was already

found that ammonium inhibits growth of Chlorella sorokiniana at con-
centrations above 1400 mgN⋅L−1 at neutral pH12. The algae performed
better at a dilution rate of 0.101 than 0.107 h−1. The 0.101 h−1 ex-
periment was performed with the same inoculum, but after the
0.107 h−1 experiment, giving the algal culture more time to acclimate
to the high ammonium levels. COD removal rates were comparable at
50% for both dilution rates, but also phosphate removal was higher at
0.101 h−1 than at 0.107 h−1.
The experiments with real urine demonstrate it is not trivial to apply
it undiluted as growth medium for microalgae. While algal biomass
productivity is highest between a dilution rate of 0.1 and 0.2 h−1, ni-
trogen removal efficiency is low and ammonium levels become in-
Fig. 3. Biomass yield on light Yxph as a function of PBR dilution rate D at steady hibiting for microalgal growth. Clearly urine must be pre-diluted to be
state. Bleu diamonds: experiments with pre-diluted synthetic urine; orange able to grow microalgae at high photosynthetic efficiency (high pho-
triangles: experiments with real human urine TN = 2626 mgN⋅L−1). Other tobioreactor productivity) and guarantee a good removal efficiency at
conditions: Chlorella sorokinina, flat panel PBR, 1 cm deep, double-sided illu- the same time. We used the model simulations to assess the urine pre-
mination at 1530 μmolph⋅ m-2⋅s−1, T = 38 °C, pH = 7. Dashed blue line re- dilution that is required to accomplish full nitrogen removal as a
presents the model simulations under experimental conditions. Error bars re- function of the (reactor) dilution rate applied. The results are shown in
present the standard deviation of the consecutive daily measurements during Fig. 4 where we assumed urine with a total nitrogen content of 6993
the steady state period as reported in Table 1. (For interpretation of the re-
mgN⋅L-1. In these calculations it was implicitly assumed that phosphorus
ferences to colour in this figure legend, the reader is referred to the web version
was never limiting.
of this article.)
The model simulations illustrate that even at a dilution rate of
0.015 h−1 the urine needs to be diluted with a factor 2.5 in order to
related to the growth of aerobic heterotrophs. When assuming a typical allow for full nitrogen removal. In Fig. 4 also the areal nitrogen removal
biomass yield of 0.46 g biomass per gram of COD removed we esti- rate is depicted with unit mgN⋅ m-2⋅ h-1. This areal nitrogen removal rate
mated that 7% of the biomass was heterotrophic biomass. Hetero- is proportional to the biomass yield on light (Fig. 3) and at a dilution
trophic COD removal thus could partly explain the higher biomass rate of 0.015 h-1 the nitrogen removal capacity is only half of the
concentration and biomass yield on light than with the application of maximal capacity observed between 0.1 and 0.15 h-1. At a dilution rate
synthetic urine. of 0.1 h-1 the urine needs to be diluted with a factor 9 to be able to
Besides aerobic heterotrophic bacteria also nitrifying bacteria could remove all nitrogen while using the full potential of the photo-
grow in urine based media. In our experiments, however, we did not bioreactor. In practice it is not very attractive to pre-dilute urine with
expect the presence of nitrifying bacteria as the lowest dilution rate clean fresh water. For this reason it might be more attractive to use the
applied of 0.05 h−1 was still above the maximal specific growth rate of water after biomass separation to dilute the urine. The feasibility of
nitrifying bacteria, which is reported to be 0.03 h−1 ( = 0.7 d−1) [31]. water recycling within microalgal production processes has been de-
Moreover, our algal based process was carried out at a pH of 7, which is monstrated in several studies [33–36]. Whether this is really feasible, or
0.5–1 pH unit lower than the optimal pH for a nitrifying culture [31]. In not, for urine needs to be studied as certain recalcitrant products (e.g.
case of operation at very low dilution rates (< 0.03 h−1) a nitrifying pharmaceuticals) could accumulate in the recycled water.
bacterial population potentially could establish but this would not limit The design of a large-scale system for algae based urine is challen-
algae growth as both processes can run in parallel. Co-cultivation of ging since it must cheap. The economic returns of the process are
microalgae and nitrifying bacteria already has been established in other coming from the algal biomass produced, which probably can only be
studies [32]. used as a fertilizer or a soil improver closing the nutrient cycle and
The experiments with real urine at a dilution rate of 0.101 and adding organic carbon to the soil [1,2]. In addition, the prevention of
0.107 h−1 resulted in a lower biomass concentration and lower biomass the need of traditional wastewater treatment systems presents an an-
yield on light than the experiments with synthetic urine and the cor- other source of income. Both, fertilizer production and wastewater
responding model simulations (Figs. 2 and 3). The nitrogen removal treatment are processes with low economic returns and therefore re-
efficiency was 37 and 29% and the dissolved nitrogen level was 1665 quire cheap and scalable systems. The scenario presented here is a
and 1862 mgN⋅L−1 respectively. The dissolved nitrogen most likely was scenario with a minimal requirement for pre-dilution as it concerns an
predominantly in the form of ammonium (Table 3). It is hypothesized optically thin reactor with a light path of 5 mm. Closed photo-
that because of the higher dilution rate and the lower biomass con- bioreactors with such a short light path are relatively expensive and
centration the ammonium nitrogen concentration became high enough difficult to scale-up. A cheaper and scalable system would be an open

Table 3
Removal characteristics of the treatment of real human urine in a panel photobioreactor. Conditions: Chlorella sorokinina, flat panel PBR, 1 cm deep, double-sided
illumination at 1530 μmolph⋅ m−2⋅s-1, T = 38 °C, pH = 7.
D h−1 TN mgN⋅L−1 NH4+-N mgN⋅L−1 PO43−P mgP⋅L−1 COD mgO2⋅L−1 Cx gx⋅L−1 rxarea gx⋅ m−2⋅ h-1 Removal efficiency / %

TN PO43−P COD

Urine 2626 2116 145.6 3270

0.050 767 625 0.7 825 15.7 3.94 71 100 75
0.101 1665 1357a 32.5 1582 7.35 3.71 37 78 52
0.107 1862 1517a 50.8 1673 5.86 3.13 29 65 49

Symbols and acronyms: D, reactor dilution rate; TN, total nitrogen; NH4+-N, ammonium nitrogen; PO43−-P, phosphate phosphorus; COD, chemical oxygen demand;
Cx, biomass concentration: rxarea, areal biomass productivity.
a
Calculated based on the ratio NH4+-N/TN derived from the experiment at 0.05 h−1.

7
K. Tuantet, et al. Algal Research 44 (2019) 101667

Table 4
Detailed analysis of simulations day night cycles with urine containing 6993 mgN⋅L−1 (total nitrogen, TN).
HRT Timing dilution Db h−1 Urine dilution CN influent mgN⋅L−1 CN effluent mgN⋅L−1 N-removal % rNarea gN⋅ m−2⋅ h-1 Yxph gx⋅moL ph−1

2.6 24/24 0.016 3 2331 1814 22.2 2.03 0.674

2.6 24/24 0.016 13 538 56 89.6 1.85 0.615
2.6 10/24a 0.038 13 538 36 93.4 1.96 0.650
1.0 10/24a 0.100 25 280 27 90.4 2.53 0.841

Symbols and acronyms: HRT, hydraulic retention time; D, reactor dilution rate; CN, total nitrogen concentration: rNarea, areal nitrogen removal rate; Yxph, biomass
yield on photons.
a
Dilution activated 1 h after sunrise and 1 h before sunset.
b
Rate during actual dilution.

Fig. 4. Model simulations of the areal nitrogen removal rate rNarea

(solid orange line) and required urine pre-dilution factor in order
to reach 100% nitrogen removal (blue dashed line). Conditions
simulated: urine with TN = 6993 mgN⋅L−1, Chlorella sorokinina,
flat panel PBR, 1 cm deep, double-sided illumination at 1530
μmolph⋅ m-2⋅s−1, T = 38 °C, pH = 7. (For interpretation of the
references to colour in this figure legend, the reader is referred to
the web version of this article.)

raceway pond. Open raceways, however, are optically deep requiring a translates into a daily light dose of 36.85 mol⋅ m-2⋅d−1, which is the
100 fold pre-dilution of urine [14,37–40] which is not very attractive average light level during the sunniest half year in the Netherlands. The
for treatment of concentrated wastewaters such as urine. Currently, the reactor was operated at a hydraulic retention time of 2.6 days which
most realistic option would be the application of open thin layer sys- translates into a dilution rate of 0.016 h−1, approximately the lowest
tems with an optical path in the order of a few millimeters to a few dilution rate simulated for constant light (Fig. 4). The reactor was di-
centimeters [41]. These systems have been applied at small scale under luted continuously at this specific rate, day and night. The resulting
outdoor conditions for wastewaters [3,42]. daily variation in biomass concentration is shown in Fig. 5.
The diurnal pattern of increasing and decreasing biomass con-
3.3. Microalgae growth and nutrient removal under day-night cycles centration was similar for consecutive days as expressed by the error
bars on the biomass concentration measurements presented in Fig. 5.
In a next step the performance of a microalgal culture composed of This repetitive pattern is therefore interpreted as a ‘steady state’ under a
Chlorella sorokiniana was tested under a day-night cycle with 3 times day-night cycle. During the day period the biomass concentration in-
diluted synthetic urine (TN undiluted = 6993 mgN⋅L−1). The same creases since the specific growth rate of the microalgae is higher than
10 mm deep flat panel photobioreactor was used as in the previous the reactor dilution rate (Fig. 5). In the night period biomass growth is
studies under constant light. The photobioreactor, however, was illu- negative and the biomass concentration deceases again. This daily
minated from one side according to a natural 12 h/12 h day night cycle variation in biomass concentration could be simulated well by the same
with a peak light intensity of 1340 μmol⋅ m-2⋅s−1. This light regime model that was already used to describe microalgal growth on synthetic

Fig. 5. microalgae cultivation on 3x pre-diluted synthetic urine

with 2331 mgN⋅L−1(total nitrogen, TN) under a 12 h/12 h day/
night cycle at HRT of 2.6 days. Orange triangles: experimental
results of dry weight biomass concentration Cx at steady state;
Orange solid line: model simulations of steady state biomass
concentration; Yellow dash-dot line: model simulation of steady
state nitrogen concentration CN; blue dash: light intensity Iph.
Other conditions: Chlorella sorokinina, flat panel PBR, 1 cm deep,
single-sided illumination at 1340 μmolph⋅ m-2⋅s−1, T = 38 °C,
pH = 7. Error bars represent the standard deviation of the con-
secutive daily measurements during a 3 days period after having
reached a steady state. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web
version of this article.)

8
K. Tuantet, et al. Algal Research 44 (2019) 101667

urine under constant light. In the beginning of the night, however, Nitrogen limitation can be reduced in case the photobioreactor is
measured biomass concentrations seem to increase rapidly within a 2 - only diluted during day time. This is illustrated by additional simula-
hs interval while the model simulations predict a decrease in biomass tion at a HRT of 2.6 days only diluted for 10 h during the day period
concentrations. We suspect an experimental artefact since there is no albeit at a higher hourly rate (0.038 h−1). Following this dilution
light energy which could support photosynthetic growth in the night. strategy the daily variation in biomass concentration and nitrogen
The biomass concentration during the day-night cycle was based on the concentration can be reduced (Figs. 6 and 7) in comparison with con-
measurement of the optical density at 750 nm (OD750). Using a pre- tinuous (24/24) reactor dilution. The daily averaged nitrogen removal
determined calibration factor the dry weight concentration then was efficiency then can be improved to 93% at the same urine dilution
calculated. Most likely the correlation between OD750 and dry weight factor of 13 (Table 4). Also the photosynthetic efficiency and the areal
concentration is not constant during a day-night cycle. Optical density nitrogen removal rate can be improved because nitrogen limitation can
is based on light scattering which is sensitive to the diameter of the be minimized. The average nitrogen concentration in the effluent and
scattering particles. During day-night cycles it has been established that thus inside the algal culture (ideally mixed situation) is 36 mgN⋅L−1.
cells grow in volume during the day and divide at the end of the day or Obviously, to achieve this improvement in practice, urine collected
the beginning of the night [43,44]. This phenomenon elegantly explains during the night must be stored prior to treatment in a photobioreactor.
the sudden ‘increase’ in the measured biomass concentration in the The areal nitrogen removal rate and the photosynthetic efficiency
beginning of the night and the observation that the measured values can be considerably improved by a further pre-dilution of urine and by
return to the model values the next day. an increase of the dilution rate, which is equivalent to a decrease of the
The model simulations now allow for a more detailed analysis of HRT. While maintaining a daily 10 h dilution period during daylight
photobioreactor operation under natural day-night cycles with the hours (10/24) the areal nitrogen removal rate becomes 2.53 gN⋅ m−2⋅d-
1
purpose of treating diluted urine. The conditions tested in the lab were at an HRT of 1 day. Also the photosynthetic efficiency is high at
performed with 3 times diluted synthetic urine giving a TN con- 0.841 gx⋅mol ph-1. In order to maintain a good nitrogen removal effi-
centration of 2331 mgN⋅L−1.The remaining nitrogen in the effluent was ciency the urine must be pre-diluted with a factor 25. This results in a
calculated based on the model simulations assuming a constant ni- nitrogen removal efficiency of 90.4% (Table 3) and the average effluent
trogen mass fraction in the biomass (8.2% w/w) and is plotted in Fig. 5. nitrogen concentration can be limited to 27 mgN⋅L-1 (Fig. 7).
The daily average values calculated from the model simulations are For completeness it must be discussed that the model used to si-
presented in Table 4. On average 22% of the nitrogen can be removed mulated day-night conditions was slightly modified in comparison to
and 1814 mgN⋅L−1 remains in the effluent. Such high nitrogen levels the one used to simulate constant illumination. In order for the model to
were also confirmed by chemical analyses of the reactor effluent. Since converge to the relatively high biomass concentration measured (Fig. 5)
the nitrogen was in the form of urea and not in the form of ammonium the maintenance factor was reduced from 2.5⋅10−6 to 1⋅10−6
these high nitrogen levels did not pose inhibitory effects to the micro- mols⋅molx⋅s-1. Flexibility of cellular respiration was already measured
algae. by Blanken et al. (2017) [46] who demonstrated that respiration de-
Although nitrogen removal efficiency was low the photosynthetic creased within a few hours after a light to dark transfer. Also in another
efficiency was reasonable at 0.674 gx⋅mol ph−1 and the areal nitrogen study it was shown that in a denser and darker culture the specific rate
removal rate was 2.03 gN⋅ m-2⋅d−1. The only way to improve nitrogen of respiration decreases [47]. We therefore hypothesize that the in-
removal efficiency is to increase urine pre-dilution. In case the urine is clusion of a 12 h night period reduced cellular maintenance. Similarly
pre-diluted with a factor 13 (TN = 538 mgN⋅L−1) the nitrogen removal we assumed that photoacclimation responds to the average specific
efficiency can be increased to 90% under natural day-night cycles growth rate during day time only. The specific growth rate in Eq. 9
(Table 4). Photosynthetic efficiency and areal nitrogen removal rate, therefore is the value of the reactor dilution rate during the 12 h day
however, decrease because of increased nitrogen limitation. The si- period only. At the same time we neglected the build-up of an in-
mulated daily variation in biomass concentration and nitrogen con- tracellular carbohydrate pool (starch) during daytime and its utilization
centration in the microalgal culture inside the photobioreactor are during night time (respiration) as both processes cancel out over a 24 h
presented in Figs. 6 and 7 respectively. At the end of the afternoon cycle. Finally, phosphorus limitation is not included yet. The N:P ratio
dissolved nitrogen levels reach 1.7 mgN⋅L−1, which is very close to the in urine is higher than in the algal biomass produced. This means that in
saturation constant for nitrogen used in the model simulations, 1.4 practice real urine needs to be enriched with phosphorus in case it is
mgN⋅L−1. In reality the saturation constant possibly is lower [45], but required to reach maximal nitrogen removal efficiency.
even in that case nitrogen limitation will still occur since the absolute Although the model can be further improved by including dynamics
amount of nitrogen available in the medium is very low. in cellular maintenance, the inclusion of day-night dynamics in

Fig. 6. Model simulations of steady state biomass concentration

Cx during microalgae cultivation on pre-diluted urine under a
12 h/12 h day/night cycle and different hydraulic retention time
(HRT) and reactor dilution strategies. Orange solid line: HRT 2.6
days, constant dilution, urine pre-dilution 13; Yellow dash: HRT
2.6 days, 10 h daylight dilution, urine pre-dilution 13; green dash-
dot: HRT 1.0 days, 10 h daylight dilution, urine pre-dilution 25;
blue dash: light intensity Iph. Conditions simulated: Chlorella sor-
okinina, flat panel PBR, 1 cm deep, single-sided illumination,
T = 38 °C, pH = 7, undiluted urine with 6993 mgN⋅L−1 (total
nitrogen, TN). (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this
article.)

9
K. Tuantet, et al. Algal Research 44 (2019) 101667

Fig. 7. Model simulations of steady state nitrogen concentration

CN during microalgae cultivation on pre-diluted urine under a
12 h/12 h day/night cycle and different hydraulic retention time
(HRT) and reactor dilution strategies. Orange solid line: HRT 2.6
days, constant dilution, urine pre-dilution 13; Yellow dash: HRT
2.6 days, 10 h daylight dilution, urine pre-dilution 13; green dash-
dot: HRT 1.0 days, 10 h daylight dilution, urine pre-dilution 25;
blue dash: light intensity Iph. Conditions simulated: Chlorella sor-
okinina, flat panel PBR, 1 cm deep, single-sided illumination,
T = 38 °C, pH = 7, undiluted urine with 6993 mgN⋅L−1(total ni-
trogen, TN). (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this
article.)

carbohydrate pools, and a calibrated refinement of photoacclimation, Author contributions

the current model is a good tool to evaluate the possibility to improve
nutrient removal by reactor control as the relative dependency between Kanjana Tuantet was responsible for the conception and design of
reactor productivity and reactor dilution rate did not depend on these the study, for the collection and assembly of data, analysis and inter-
biological uncertainties. The model simulations show that in order to pretation of the data, and drafting of the article.
reach the best effluent quality, while not sacrificing productivity, the Hardy Temmink was responsible for the conception and design of
photobioreactor should be diluted during the day period only. In ad- the study, analysis and interpretation of the data, critical revision of the
dition, maximal areal nitrogen removal and photosynthetic efficiency article for important intellectual content, and obtaining funding for the
can be reached by increasing urine pre-dilution because the photo- study.
bioreactor then can be operated at higher dilution rate and shorter Grietje Zeeman was responsible for the conception and design of the
HRT. The positive effect of short HRT is related to improved light pe- study, critical revision of the article, and obtaining funding for the
netration and low algal biomass concentration preventing the devel- study.
opment of a dark zone where biomass and nutrients are lost due to René Wijffels and Cees Buisman were responsible for the critical
maintenance related respiration [19]. Obviously the biomass con- revision of the article.
centration should be maintained at a level which still guarantees Marcel Janssen was responsible for the conception and design of the
maximal light absorption. Such conditions can also be realized by other study, analysis and interpretation of the data, drafting and critical re-
operating strategies than chemostat cultivation at fixed dilutions rate. vision of the article, and for the final approval.
Turbidostat control [48], or any control mode able to maintain constant
biomass concentration, and thus constant nutrient removal efficiency, Declaration of Competing Interest
are well suited for the control of phototrophic nutrient removal from
wastewaters. No conflicts, informed consent, or human or animal rights are ap-
plicable to this study.

4. Conclusions Acknowledgements

Experiments and model simulations demonstrated that urine can be The project was financially supported by Innowater funding pro-
treated efficiently and at high rate by photoautotrophic cultivation of vided by the Dutch Department of Economic Affairs. The PhD student
microalgae on pre-diluted urine. Nitrogen removal rates of 2.5 was financially supported by the Ministry of Science and Technology,
gN⋅ m−2⋅d-1 can be reached at daily light doses equivalent to Dutch Thailand.
summer conditions. The level of pre-dilution required depends on the
optical path of the algal cultivation system (photobioreactor) and the References
incident light intensity. A shorter optical path and a higher incident
light intensity will lower the requirement for pre-dilution. Urine pre- [1] T.J. Painter, Carbohydrate polymers in desert reclamation: the potential of micro-
dilution is necessary to reach full nutrient removal rate while still algal biofertilizers, Carbohydr. Polym. 20 (1993) 77–86.
[2] W. Mulbry, S. Kondrad, C. Pizarro, Biofertilizers from algal treatment of dairy and
maintaining a relatively short hydraulic retention time (HRT). A short swine manure effluents, J. Veg. Sci. 12 (2007) 107–125.
HRT is required to reach maximal photosynthetic efficiency and, as [3] M. Morales-Amaral, M. del, C. Gómez-Serrano, F.G. Acién, J.M. Fernández-Sevilla,
such, maximal areal nitrogen removal rate. The positive impact of de- E. Molina-Grima, Outdoor production of Scenedesmus sp. in thin-layer and raceway
reactors using centrate from anaerobic digestion as the sole nutrient source, Algal
creasing the HRT is founded in the reduction of the biomass con- Res. 12 (2015) 99–108.
centration improving light penetration in the algal culture. Under nat- [4] J.B.K. Park, R.J. Craggs, A.N. Shilton, Wastewater treatment high rate algal ponds
ural day-night cycles the photobioreactor ideally should be only diluted for biofuel production, Bioresour. Technol. 102 (2011) 35–42.
[5] L. Novoveská, A.K.M. Zapata, J.B. Zabolotney, M.C. Atwood, E.R. Sundstrom,
during the day period as this will limit daily fluctuations of the algal Optimizing microalgae cultivation and wastewater treatment in large-scale offshore
biomass concentration and dissolved nutrient concentrations. As such it photobioreactors, Algal Res. 18 (2016) 86–94.
is possible to obtain a good effluent quality with low nutrient levels [6] L. Christenson, R. Sims, Production and harvesting of microalgae for wastewater
treatment, biofuels, and bioproducts, Biotechnol. Adv. 29 (2011) 686–702.
while limiting nutrient limitation of algal growth in the photo-
[7] S. Van Den Hende, V. Beelen, G. Bore, N. Boon, H. Vervaeren, Up-scaling aqua-
bioreactor. culture wastewater treatment by microalgal bacterial flocs: from lab reactors to an
outdoor raceway pond, Bioresour. Technol. 159 (2014) 342–354.
[8] W. Mulbry, S. Kondrad, C. Pizarro, E. Kebede-Westhead, Treatment of dairy manure

10
K. Tuantet, et al. Algal Research 44 (2019) 101667

effluent using freshwater algae: algal productivity and recovery of manure nutrients [29] H. Qiang, Y. Zarmi, A. Richmond, Combined effects of light intensity, light-path and
using pilot-scale algal turf scrubbers, Bioresour. Technol. 99 (2008) 8137–8142. culture density on output rate of Spirulina platensis (Cyanobacteria), Eur. J. Phycol.
[9] G. Zeeman, K. Kujawa-Roeleveld, Resource recovery from source separated do- 33 (1998) 165–171.
mestic waste(water) streams; full scale results, Water Sci. Technol. 64 (2011) [30] C. Holdmann, U. Schmid-Staiger, H. Hornstein, T. Hirth, Keeping the light energy
1987–1992. constant — cultivation of Chlorella sorokiniana at different specific light avail-
[10] R. Otterpohl, M. Grottker, J. Lange, Sustainable water and waste management in abilities and different photoperiods, Algal Res. 29 (2018) 61–70.
urban areas, Water Sci. Technol. 35 (1997) 121–133. [31] R. Blackburne, V.M. Vadivelu, Z. Yuan, J. Keller, Determination of growth rate and
[11] K. Kujawa-Roeleveld, G. Zeeman, Anaerobic treatment in decentralised and source- yield of nitrifying Bacteria by measuring carbon dioxide uptake rate, Water
separation-based sanitation concepts, Rev. Environ. Sci. Biotechnol. 5 (2006) Environ. Res. 79 (2007) 2437–2445.
115–139. [32] N.G.A.I. Karya, N.P. van der Steen, P.N.L. Lens, Photo-oxygenation to support ni-
[12] K. Tuantet, et al., Microalgae growth on concentrated human urine, J. Appl. Phycol. trification in an algal-bacterial consortium treating artificial wastewater, Bioresour.
26 (2014) 287–297. Technol. 134 (2013) 244–250.
[13] S. Zhang, C.Y. Lim, C.-L. Chen, H. Liu, J.-Y. Wang, Urban nutrient recovery from [33] E. Sforza, E. Barbera, F. Girotto, R. Cossu, A. Bertucco, Anaerobic digestion of lipid-
fresh human urine through cultivation of Chlorella sorokiniana, J. Environ. extracted microalgae: enhancing nutrient recovery towards a closed loop recycling,
Manage. 145 (2014) 129–136. Biochem. Eng. J. 121 (2017) 139–146.
[14] S. Jaatinen, A.-M. Lakaniemi, J. Rintala, Use of diluted urine for cultivation of [34] L. Garcia Alba, et al., Recycling nutrients in algae biorefinery, ChemSusChem 6
Chlorella vulgaris, Environ. Technol. 37 (2016) 1159–1170. (2013) 1330–1333.
[15] B. Piltz, M. Melkonian, Immobilized microalgae for nutrient recovery from source- [35] F. Hadj-Romdhane, et al., Development and validation of a minimal growth
separated human urine, J. Appl. Phycol. 30 (2018) 421–429. medium for recycling Chlorella vulgaris culture, Bioresour. Technol. 123 (2012)
[16] K. Tuantet, et al., Nutrient removal and microalgal biomass production on urine in a 366–374.
short light-path photobioreactor, Water Res. 55 (2014) 162–174. [36] C. Zhu, R. Zhang, L. Cheng, Z. Chi, A recycling culture of Neochloris oleoabundans
[17] M. Cuaresma, M. Janssen, C. Vílchez, R.H. Wijffels, Productivity of Chlorella sor- in a bicarbonate-based integrated carbon capture and algae production system with
okiniana in a short light-path (SLP) panel photobioreactor under high irradiance, harvesting by auto-flocculation, Biotechnol. Biofuels 11 (2018) 204.
Biotechnol. Bioeng. 104 (2009) 352–359. [37] Y. Chang, Z. Wu, L. Bian, D. Feng, D.Y.C. Leung, Cultivation of Spirulina platensis
[18] M. Janssen, Microalgal photosynthesis and growth in mass culture, Adv. Chem. Eng. for biomass production and nutrient removal from synthetic human urine, Appl.
48 (2016). Energy 102 (2013) 427–431.
[19] H. Takache, G. Christophe, J.F. Cornet, J. Pruvost, Experimental and theoretical [38] C. Yang, H. Liu, M. Li, C. Yu, G. Yu, Treating urine by Spirulina platensis, Acta
assessment of maximum productivities for the microalgae Chlamydomonas re- Astronaut. 63 (2008) 1049–1054.
inhardtii in two different geometries of photobioreactors, Biotechnol. Prog. 26 [39] D. Feng, Z. Wu, Culture of Spirulina platensis in human urine for biomass pro-
(2010) 431–440. duction and O2 evolution, J. Zhejiang Univ. Sci. B 7 (2006) 34–37.
[20] H. Takache, J. Pruvost, J.F. Cornet, Kinetic modeling of the photosynthetic growth [40] M. Adamsson, Potential use of human urine by greenhouse culturing of microalgae
of Chlamydomonas reinhardtii in a photobioreactor, Biotechnol. Prog. 28 (2012) (Scenedesmus acuminatus), zooplankton (Daphnia magna) and tomatoes
681–692. (Lycopersicon), Ecol. Eng. 16 (2000) 243–254.
[21] J.C. Ogbonna, H. Yada, H. Tanaka, Kinetic study on light-limited batch cultivation [41] J. Masojídek, J. Kopecký, L. Giannelli, G. Torzillo, Productivity correlated to pho-
of photosynthetic cells, J. Ferment. Bioeng. 80 (1995) 259–264. tobiochemical performance of Chlorella mass cultures grown outdoors in thin-layer
[22] E. Grima, F. Camacho, A mathematical model of microalgal growth in light-limited cascades, J. Ind. Microbiol. Biotechnol. 38 (2011) 307–317.
chemostat culture, J. Chem. 167–173 (2004), https://doi.org/10.1002/jctb. [42] S. Egloff, F. Tschudi, Z. Schmautz, D. Refardt, High-density cultivation of micro-
280610212. algae continuously fed with unfiltered water from a recirculating aquaculture
[23] C. Sorokin, J. Myers, A high-temperature strain of Chlorella, Science (80-.) 117 system, Algal Res. 34 (2018) 68–74.
(1953) 330–331. [43] K. Bišová, V. Zachleder, Cell-cycle regulation in green algae dividing by multiple
[24] W. Blanken, P.R.R. Postma, L. de Winter, R.H.R.H. Wijffels, M. Janssen, Predicting fission, J. Exp. Bot. 65 (2014) 2585–2602.
microalgae growth, Algal Res. 14 (2016) 28–38. [44] L. de Winter, et al., The influence of day length on circadian rhythms of Neochloris
[25] A.M. Kliphuis, et al., Photosynthetic efficiency of Chlorella sorokiniana in a tur- oleoabundans, Algal Res. 22 (2017) 31–38.
bulently mixed short light-path photobioreactor, Biotechnol. Prog. 26 (2010) [45] S.G. Halterman, D.W. Toetz, Kinetics of nitrate uptake by freshwater algae,
687–696. Hydrobiologia 114 (1984) 209–214.
[26] A.D. Jassby, T. Platt, Mathematical formulation of the relationship between pho- [46] W. Blanken, et al., Microalgal biofilm growth under day-night cycles, Algal Res. 21
tosynthesis and light for phytoplankton, Limnol. Oceanogr. 21 (1976) 540–547. (2017) 16–26.
[27] T. de Mooij, Z.R.Z.R. Nejad, L. van Buren, R.H. Wijffels, M. Janssen, Effect of [47] A.M.J. Kliphuis, M. Janssen, E.J.E.J. van den End, D.E.D.E. Martens,
photoacclimation on microalgae mass culture productivity, Algal Res. 22 (2017) R.H.R.H. Wijffels, Light respiration in Chlorella sorokiniana, J. Appl. Phycol. 23
56–67. (2011) 935–947.
[28] A.M. Kliphuis, et al., Metabolic modeling of Chlamydomonas reinhardtii: energy [48] J.H. de Vree, et al., Turbidostat operation of outdoor pilot-scale photobioreactors,
requirements for photoautotrophic growth and maintenance, J. Appl. Phycol. 24 Algal Res. 18 (2016) 198–208.
(2012) 253–266.

11