Process Design and Economics

for the Conversion of Algal

Biomass to Biofuels:
Algal Biomass Fractionation to Lipid-
and Carbohydrate-Derived Fuel
Products
R. Davis, C. Kinchin, J. Markham, E.C.D. Tan,
and L.M.L. Laurens
National Renewable Energy Laboratory

D. Sexton, D. Knorr, P. Schoen, and J. Lukas

Harris Group Inc.

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

Technical Report
NREL/TP-5100-62368
September 2014

Contract No. DE-AC36-08GO28308

 Process Design and
Economics for the Conversion
of Algal Biomass to Biofuels:
Algal Biomass Fractionation to Lipid-
and Carbohydrate-Derived Fuel
Products
R. Davis, C. Kinchin, J. Markham, E.C.D. Tan,
and L.M.L. Laurens
National Renewable Energy Laboratory

D. Sexton, D. Knorr, P. Schoen, and J. Lukas

Harris Group Inc.
Prepared under Task No. BB14.1110

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

National Renewable Energy Laboratory Technical Report

15013 Denver West Parkway NREL/TP-5100-62368
Golden, CO 80401 September 2014
303-275-3000 www.nrel.gov
Contract No. DE-AC36-08GO28308
 NOTICE

This report was prepared as an account of work sponsored by an agency of the United States government.
Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty,
express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of
any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States government or any agency thereof. The views and opinions of authors
expressed herein do not necessarily state or reflect those of the United States government or any agency thereof.

This report is available at no cost from the National Renewable Energy

Laboratory (NREL) at www.nrel.gov/publications.

Available electronically at http://www.osti.gov/scitech

Available for a processing fee to U.S. Department of Energy

and its contractors, in paper, from:

U.S. Department of Energy

Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831-0062
phone: 865.576.8401
fax: 865.576.5728
email: mailto:[email protected]

Available for sale to the public, in paper, from:

U.S. Department of Commerce

National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
phone: 800.553.6847
fax: 703.605.6900
email: [email protected]
online ordering: http://www.ntis.gov/help/ordermethods.aspx

Cover Photos: (left to right) photo by Pat Corkery, NREL 16416, photo from SunEdison, NREL 17423, photo by Pat Corkery, NREL
16560, photo by Dennis Schroeder, NREL 17613, photo by Dean Armstrong, NREL 17436, photo by Pat Corkery, NREL 17721.
 Executive Summary
The U.S. Department of Energy (DOE) promotes the production of a range of liquid fuels and
fuel blendstocks from biomass feedstocks by funding fundamental and applied research that
advances the state of technology in biomass production, conversion, and sustainability. As part
of its involvement in this program, the National Renewable Energy Laboratory (NREL)
investigates the conceptual production economics of these fuels. This includes fuel pathways
from lignocellulosic (terrestrial) biomass, as well as from algal (aquatic) biomass systems.

Over the past decade, NREL conducted a campaign to quantify the economic implications
associated with observed and aspirational performance for the conversion of corn stover to
ethanol through techno-economic modeling. This effort served two important purposes: (1) to
establish a benchmark representing the initial state of technology at the time, and (2) to set
goals for near-term R&D and cost targets, as well as to track progress toward achieving these
targets by periodically updating the models based on the latest research improvements.
Beginning in 2013, NREL began transitioning from the singular focus on ethanol to a broad slate
of products and conversion pathways, ultimately to establish similar benchmarking and targeting
efforts. One of these pathways is the conversion of algal biomass to fuels via extraction of lipids
(and potentially other components), termed the algal lipid upgrading or ALU pathway.

This report describes in detail one potential ALU approach based on a biochemical processing
strategy to selectively recover and convert select algal biomass components to fuels, namely
carbohydrates to ethanol and lipids to a renewable diesel blendstock (RDB) product. The
overarching process design converts algal biomass delivered from upstream cultivation and
dewatering (outside the present scope) to ethanol, RDB, and minor coproducts, using dilute-acid
pretreatment, fermentation, lipid extraction, and hydrotreating. Ancillary areasanaerobic
digestion of spent algal residues, combined heat and power generation, and utilitiesare also
included in the design. Detailed material and energy balances and capital and operating costs for
this baseline process are also documented. This case study techno-economic model provides a
production cost for the fuel products that can be used to gauge the technology potential and to
quantify critical cost drivers. The analysis presented here also includes consideration of life-
cycle implications by tracking environmental sustainability metrics for the modeled conversion
step, including greenhouse gas emissions, fossil energy demand, and consumptive water use.

Following similar methods for design report modeling as conducted over recent years, NREL,
supported by Harris Group Inc., performed a feasibility-level analysis for a plausible conversion
pathway to meet a cost goal below $5/gallon gasoline equivalent (GGE). The modeled facility
processes an average of 1,339 dry ash-free ton biomass/day and achieves an overall fuel selling
price of $4.35/GGE in 2011-dollars as determined by modeled conversion targets and nth-plant
project costs and financing, associated with a net fuel yield of 141 GGE/dry ton. These values
are attributed to a high (41%) lipid, late-harvest biomass scenario, which implies aggressive
future targets in achieving simultaneously high cultivation productivity rates during upstream
biomass growth, implicit in the year 2022 target projections stipulated in this design.
Consequently, an alternative scenario was also considered based on lower (27%) lipid, earlier-
harvest biomass, which resulted in fractionally higher costs at $5.04/GGE associated with a
lower overall fuel yield of 116 GGE/dry ton. As this alternate case is based on a lipid content

iii
This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications.
more typically achievable today, a realistic future target case may in fact fall between these two
points as an optimum between aspirational goals for algal growth rate and lipid content.

The analysis also indicates that economics are influenced strongly by feedstock costs,
contributing 70% to overall selling price at $430/ton. The report reiterates prior findings that
seasonality plays a unique and critical role in algal systems, with high variation in upstream algal
cultivation translating to quantified tradeoffs in cost and sustainability metrics depending on how
the facility is designed to accommodate such variability. In addition, the report includes a high-
level discussion on improvements needed to achieve an ultimate target of $3/GGE moving
forward, including unique advantages inherent to this particular processing strategy in enabling
numerous alternative product or coproduct routes.

iv
Figure ES-1. Economic summary for ALU pathway

v
Table of Contents
1 Introduction ........................................................................................................................................... 1
1.1 Background and Motivation .......................................................................................................... 1
1.2 Process Overview .......................................................................................................................... 4
1.3 Techno-Economic Analysis Approach .......................................................................................... 7
1.4 About nth-Plant Assumptions......................................................................................................... 8
1.5 About the NREL Aspen Model ..................................................................................................... 9
2 Design Basis and Conventions ......................................................................................................... 10
2.1 Plant Size: Feedstock Logistics and Cost .................................................................................... 10
2.2 Feedstock Composition ............................................................................................................... 11
2.3 Modeling Basis Parameters ......................................................................................................... 15
2.4 Design Report Conventions......................................................................................................... 15
2.4.1 Units ............................................................................................................................... 15
2.4.2 Product Density and Heating Value ............................................................................... 15
3 Process Design and Cost Estimation Details .................................................................................. 17
3.1 Area 100: Pretreatment and Conditioning ................................................................................... 17
3.1.1 Overview ........................................................................................................................ 17
3.1.2 Design Basis ................................................................................................................... 19
3.1.3 Cost Estimation .............................................................................................................. 21
3.2 Area 200: Fermentation and Distillation ..................................................................................... 23
3.2.1 Overview ........................................................................................................................ 23
3.2.2 Design Basis ................................................................................................................... 24
3.2.3 Cost Estimation .............................................................................................................. 28
3.3 Area 300: Lipid Extraction and Solvent Recovery...................................................................... 28
3.3.1 Overview ........................................................................................................................ 28
3.3.2 Design Basis ................................................................................................................... 29
3.3.3 Cost Estimation .............................................................................................................. 31
3.4 Area 400: Product Purification and Upgrading ........................................................................... 32
3.4.1 Overview ........................................................................................................................ 32
3.4.2 Design Basis ................................................................................................................... 33
3.4.3 Cost Estimation .............................................................................................................. 38
3.5 Area 500: Anaerobic Digestion/CHP .......................................................................................... 39
3.5.1 Overview ........................................................................................................................ 39
3.5.2 Design Basis ................................................................................................................... 40
3.5.3 Cost Estimation .............................................................................................................. 44
3.6 Area 600: Storage ........................................................................................................................ 45
3.6.1 Overview ........................................................................................................................ 45
3.6.2 Design Basis ................................................................................................................... 45
3.6.3 Cost Estimation .............................................................................................................. 45
3.7 Area 700: Utilities ....................................................................................................................... 46
3.7.1 Overview ........................................................................................................................ 46
3.7.2 Design Basis ................................................................................................................... 46
3.7.3 Cost Estimation .............................................................................................................. 49
4 Process Economics ........................................................................................................................... 50
4.1 About Cost-Year Indices ............................................................................................................. 50
4.2 Total Capital Investment (TCI) ................................................................................................... 51
4.3 Variable Operating Costs ............................................................................................................ 53
4.4 Fixed Operating Costs ................................................................................................................. 56
4.5 Discounted Cash Flow Analysis and the Minimum Selling Price of Fuel .................................. 56
5 Analysis and Discussion ................................................................................................................... 62

vi
 5.1 Cost Sensitivity Analysis ............................................................................................................ 62
5.2 Alternative Composition and Processing Scenarios .................................................................... 64
5.3 Sustainability Metrics for Base Model ........................................................................................ 66
5.4 Paths Forward to $3/GGE ........................................................................................................... 71
6 Concluding Remarks .......................................................................................................................... 73
6.1 Summary ..................................................................................................................................... 73
6.2 Future Work ................................................................................................................................ 75
References ................................................................................................................................................. 77
Appendix A. Individual Equipment Cost Summary ............................................................................... 82
Appendix B. Discounted Cash Flow Rate of Return Worksheet .......................................................... 86
Appendix C. Aspen Plus Properties........................................................................................................ 88
Appendix D. Process Flow Diagrams ..................................................................................................... 92

vii
Acronyms
AD anaerobic digestion ISBL inside battery limits (of the
AFDW ash-free dry weight plant)
ALU algal lipid upgrading LCA life-cycle assessment
ANL Argonne National Laboratory LHV lower heating value
ATP3 Algae Testbed Public-Private LHSV liquid hourly space velocity
Partnership MACRS IRS Modified Accelerated
BETO Bioenergy Technologies Cost Recovery System
Office MFSP minimum fuel selling price
BFW boiler feed water MM million (e.g., MMBTU or
BGY billion gallons per year $__MM)
CAPEX capital expenditure MYPP DOE-BETOs Multi-Year
CHP combined heat and power Program Plan
CIP clean-in-place NG natural gas
COD chemical oxygen demand NPV net present value
CSL corn steep liquor NREL National Renewable Energy
DAP diammonium phosphate Laboratory
DB declining balance OPEX operating expenditure
DCFROR discounted cash flow rate of OSBL outside battery limits
return PFD process flow diagram
DCO decarboxylation PNNL Pacific Northwest National
DOE U.S. Department of Energy Laboratory
FAME fatty acid methyl ester PSA pressure swing adsorption
FCI fixed capital investment R&D research and development
FFA free fatty acid RA resource assessment
GGE gallon gasoline equivalent RDB renewable diesel blendstock
GHG greenhouse gas SABC Sustainable Algal Biofuels
HCSD high-carbohydrate (mid- Consortium
harvest) Scenedesmus SCFM standard cubic feet per
HDO hydrodeoxygenation minute
HHV higher heating value SMR steam methane reforming
HLSD high-lipid (late harvest) SOT annual State of Technology
Scenedesmus case
HMF 5-hydroxymethyl furfural TAN total acid number
HRSG heat recovery steam TCI total capital investment
generator TDC total direct cost
HRT hydraulic retention time TEA techno-economic analysis
HTL hydrothermal liquefaction TS total solids
IPCC Intergovernmental Panel on VOC volatile organic compound
Climate Change WWT wastewater treatment
IRR internal rate of return

viii
1 Introduction
1.1 Background and Motivation
The U.S. Department of Energy (DOE) Bioenergy Technologies Office (BETO) promotes the
production of liquid fuels from lignocellulosic and algal feedstocks by sponsoring programs in
fundamental and applied research that aim to advance the state of technology spanning the
supply chain from biomass production through processing and conversion to fuels. Within the
algae platform, these programs include laboratory research to improve biological characteristics
(e.g., algal cultivation productivity, biomass composition, strain robustness) through screening
and synthetic biology; engineering studies of potential systems for growth, dewatering, and
conversion technologies; improvement of laboratory analytical capabilities to accurately
characterize feed and product materials; and support for biomass production test-bed and
processing demonstration facilities. This research is conducted by national laboratories,
universities, and private industry, both individually and through multi-organization consortia
partnerships.

To support the DOE program, the National Renewable Energy Laboratory (NREL) investigates
the process design and economics of numerous biofuel manufacturing pathways in order to
develop an absolute plant gate price for fuels and fuel blendstocks based on process and plant
design assumptions consistent with applicable best practices in engineering, construction, and
operation. This plant gate price is referred to as the minimum fuel selling price or MFSP. The
MFSP can be used by DOE to assess the cost-competitiveness and market penetration potential
of a given biofuel technology in comparison with petroleum-derived fuels and established
biofuel pathways such as starch- or sugar-based ethanol, vegetable oil-based biodiesel, etc.

The techno-economic analysis effort at NREL also helps to direct biomass conversion research
by examining the sensitivity of the MFSP to process alternatives and research advances.
Proposed research and its anticipated results can be translated into a new MFSP that can be
compared to established benchmark cases. Such comparison helps to quantify the economic
impact of core research targets and to track progress toward meeting competitive cost targets. It
also allows DOE to make more informed decisions about research proposals that claim to reduce
costs.

For more than 10 years, NREL has developed design case models and associated reports that
document process and cost targets focused on cellulosic (terrestrial) biofuel pathways, focused
first on ethanol production and more recently on hydrocarbon biofuel production [1-6], based on
the best understanding of the technology and equipment costs at the time. Over the past four
years, DOE-BETO expanded its focus to include algal biofuel pathways in light of renewed
industry interest in these pathways and the promising potential across metrics such as high
biomass yields, flexible product slates, and the ability to utilize low-value land and water
resources that need not necessarily compete with food-based crops [7]. For one of the first
analysis activities under this renewed focus, DOE chartered a collaborative Harmonization
Initiative which brought together modeling partners from NREL, Argonne National Laboratory
(ANL), and Pacific Northwest National Laboratory (PNNL) to harmonize their conceptual
models around techno-economic analysis (TEA), life-cycle assessment (LCA), and resource
assessment (RA) respectively, such that the results from each model would carry the same
implications being based on consistent inputs and assumptions. This effort included a workshop

1
to serve as a vetting process for the respective collaborators to present the details of their models
to a group of expert stakeholders in industry, academia, and other national laboratories, in order
to begin validating or improving key modeling assumptions. This ultimately resulted in the
publication of a harmonization report (referred to as such hereafter) documenting model details
and the resulting near-term cost, sustainability, and resource implications for production of 5
billion gallons per year (BGY) of renewable diesel at the national scale spread across a large
consortium of individual unit farms. The harmonization scenario was based on an algal lipid
extraction and upgrading (ALU) process focused on extracting algal lipids and routing all
remaining residual material to anaerobic digestion. Details of the 2012 ALU harmonization
report may be found in [8] and will not be repeated here, save but to emphasize the key result
from the analysis, which was that algae as a feedstock possess a unique challenge given high
seasonal variability in cultivation productivity rates, varying up to tenfold between summer and
winter; such variations exhibit important impacts on TEA and LCA evaluations and must be
considered in such analyses.

The 2012 ALU harmonization effort was subsequently leveraged in a number of ways, first by
DOE-BETO in translating the results to an initial baseline and projecting out-year process and
cost improvement goals [9], and next by being expanded on in 2013 to repeat a similar
harmonization focused on hydrothermal liquefaction (HTL) conversion of the algal biomass [10].
The newer 2013 harmonization scenarios were extrapolated out to a design report focused on
year 2022 targets for the algal HTL conversion pathway [11] (referred to hereafter as the HTL
design report), which established a modeled MFSP target of $4.49/gallon gasoline equivalent
(GGE) for the pathway. The present report presents a comparative TEA analysis for an
alternative approach to achieving a similar MFSP cost, by way of a targeted process to
fractionate algal biomass and selectively convert the major biomass constituents to fuel products,
specifically carbohydrates to ethanol and lipids to renewable diesel blendstock (RDB). In
contrast to HTL conversion, which is a thermochemical-based approach to convert the whole
algal biomass feedstock to a hydrocarbon (primarily diesel-range) intermediate fuel precursor,
the fractionation process considered here is based on biochemical conversion principles whereby
specific components of the biomass are selectively targeted to specific fuel products via
biological and chemical/physical pathways.

As with any technology pathway, there are strengths and challenges for both HTL processing as
well as the present fractionation processing approach. In the case of HTL, the technology
translates to high carbon-retention efficiency to fuels by way of converting biomass carbon
beyond only lipids, but to do so requires operating at high pressure (approximately 3,000 psia as
documented in [11]) and as with any thermochemical processing technique, does not allow for
stringent control over heteroatom partitioning (namely nitrogen and phosphorus) into the organic
biocrude phase amongst other product phases. Early studies have indicated that low-level blends
of HTL biocrude on the order of 10% or less co-processed with petroleum feedstocks allow for
more mild operating conditions similar to standard petroleum hydrotreating [12]. In the case of
the biochemical fractionation approach considered here, our analysis finds that this approach also
allows for high carbon efficiencies by fractionating and then converting both carbohydrates and
lipids, which constitute a large fraction of the biomass for mid- to late-stage harvesting practices;
however, this pathway also relies more heavily on lipid content to maximize fuel energy yields,
which translates to more challenging algal productivity growth rates during upstream cultivation.
Additionally, conditions required for release of fermentable sugars may vary with species and

2
upstream cultivation conditions. Key benefits for this fractionation process may be summarized
as follows:

Uses established and proven technologies for fractionation (dilute acid

pretreatment/hydrolysis), carbohydrate utilization (fermentation), and spent residual
biomass utilization (anaerobic digestion), with all aspects of the process operating at
reasonable conditions; maximum operating pressure for this system is 70 psia, except for
the hydrotreater which operates at 465 psia as typical for mild hydroprocessing.
Is highly amenable to fermentative conversion of sugars to a range of biological products
(ethanol or hydrocarbons), as algal biomass typically contains low levels of pentose
sugars that are not readily metabolized by most organisms without genetic engineering
[13]. The algal biomass is also less recalcitrant to biochemical deconstruction than
lignocellulosic feedstock pathways, and produces relatively clean sugars.
Does not require thermal evaporation (drying) for any aspect of the conversion process,
with efficacy of all unit operations experimentally demonstrated based on a starting 20%
biomass solids content delivered to fractionation [13].
Allows for a high degree of flexibility in product/coproduct options, namely conversion
of sugars to a wide range of fuels or chemicals, as well as leaving a potentially high-value
protein residue that may be converted to additional coproducts.
Selectively converts specific components to specific, easily-characterized products with
mild subsequent upgrading demands (hydrotreating at approximately 500 psig, in the case
of the lipid product).
The focus of this report is to document a plausible pathway model for conversion of algal
carbohydrates and lipids to fuel and blendstock products, with high fractional energy yield to
hydrocarbon products (e.g., renewable diesel) supplemented by additional energy yield to
ethanol as a representative fermentative product from sugars; primarily to demonstrate a means
to achieve modeled MFSP costs under $5/GGE by 2022, but with additional insight provided
towards a path forward to ultimately reduce costs to $3/GGE. This analysis leverages the
decades of experience that NREL has established in biochemical conversion research, primarily
with respect to dilute acid biomass pretreatment to hydrolyze carbohydrates as well as
fermentation of sugars. Likewise, the work also continues NRELs practice of consulting with
vendors through the assistance of an engineering company to assist in design and cost estimation
for critical unit operations. For the present report, NREL worked with Harris Group to provide
engineering support primarily for new unit operations that had not been previously used in other
cellulosic technology designs (such as extraction and algal lipid cleanup/hydrotreating), while
also leveraging vendor-supplied cost estimates supplied by Harris Group for existing unit
operations (such as dilute acid pretreatment, fermentation, distillation, and anaerobic digestion)
[3, 6]. Thus the economics of this conceptual process uses the best available equipment and raw
material costs and an nth-plant project cost structure and financing. The projected 2022 nth-
plant MFSP computed in this report is $4.35/GGE ($4.57/gal RDB and $2.95/gal ethanol) in
2011-year dollars.

Similar to caveats noted in prior NREL design reports, we stress that this design report serves to
describe a single, feasible conversion process and to transparently document the assumptions and

3
details that went into its design. This report is not meant to provide an exhaustive survey of
process alternatives or cost-sensitivity analyses. Furthermore, it is important to note that algal
biofuels as a technology platform are in a more nascent stage of development and understanding
than other more-established biofuel pathways (e.g., from terrestrial biomass), and continue to
suffer from a dearth of available public data on processing performance at sufficient scale. While
the key inputs assumed in this analysis are extrapolated from data gathered through experimental
work conducted at NREL and the recently-completed Sustainable Algal Biofuels Consortium
(SABC) [14], further experimental work and scale-up is required to reduce model uncertainties.
Moving forward, as the science and technology progresses for less well-studied areas of the
process such as lipid extraction or alternative conversion routes of sugars to hydrocarbon
products, the process models and economic tools developed for this report may be updated in a
similar fashion as prior NREL design report iterations have evolved.

1.2 Process Overview

The process described here uses cocurrent dilute-acid pretreatment of algal biomass delivered
after upstream dewatering to 20 wt% solids (outside the scope of this analysis), followed by
whole-slurry fermentation of the resulting monomeric sugars to ethanol, followed by distillation
and solvent extraction of the stillage to recover lipids (primarily fatty acid-based lipids with
inclusion of polar lipid impurities). The process design also includes lipid product purification,
product upgrading (hydrotreating) to straight-chain paraffin blend stocks, anaerobic digestion
and combined heat and power (CHP) generation, product storage, and required utilities. The
process is divided into seven areas (see Figure 1).
Area 100: Pretreatment and conditioning. In this area, the algal biomass is combined
with steam and treated with dilute sulfuric acid catalyst at a high temperature for a short
time to hydrolyze the glucan carbohydrates to monomeric sugars and make the remaining
biomass more amenable to lipid extraction. Ammonia is then added to the whole
pretreated slurry to raise its pH to ~5 for fermentation.
Area 200: Fermentation and distillation. After flashing the pretreatment hydrolysate back
to 20% total solids (TS), the slurry is cooled and inoculated with the fermenting organism
Saccharomyces cerevisiae. Enzymatic hydrolysis is not required, as hydrolysis (by acid)
was already accomplished in Area 100. Fermentation proceeds in batch mode for 1.5
days, after which point most of the sugars (primarily the hexose sugars glucose and
mannose) are converted to ethanol. The resulting dilute ethanol broth is distilled to near-
azeotropic concentration and then purified to 99.5% using vapor-phase molecular sieve
adsorption. The beer stillage containing all residual solids is routed to lipid extraction;
while water from the rectification column is sent to the water recycle pool. CO2 vented
from the fermentation step is recycled to upstream algal cultivation ponds.
Area 300: Lipid extraction and solvent recovery. The distillation stillage is fed to a
countercurrent liquid-liquid extraction column along with hexane solvent required at a
5:1 mass ratio (solvent versus dry algal solids in the stillage feed). The light product
phase containing primarily solvent and oil is routed to a distillation stripping column to
recover the solvent, with marginal solvent losses to the bottoms product as well as the
extraction aqueous phase. The resulting distillation bottoms product is composed of
>99.5% oil (neutral and polar lipids) and is sent to lipid purification to remove polar
lipids and other impurities. The aqueous phase is routed to anaerobic digestion.

4
 Area 400: Product purification and upgrading. The recovered oil product is sent to a
series of purification steps consisting of degumming, demetallization, and bleaching to
remove phospholipids and other polar lipids, metals, salts, and other impurities with the
addition of phosphoric acid, wash water, silica, and clay. The purified oil, consisting
primarily of fatty acid based lipids, is sent to product upgrading in an on-site
hydrotreating facility (including reactors, fresh and recycle gas compressors, flash
columns, and product fractionation, utilizing purchased hydrogen). The primary product
from the hydrotreating section is a diesel-range paraffinic product suitable as a diesel
blendstock (RDB) with a small naphtha coproduct. The hydrotreating section also
includes a PSA unit in the recycle gas loop to remove CO2 generated during
decarboxylation.
Area 500: Anaerobic digestion/CHP. The residual raffinate product stream from
extraction is combined with the oil cleanup waste stream and sent to anaerobic digestion,
primarily utilized as a means to reclaim carbon via biogas production as well as enable
nutrient recycle coproduct credits intended to be routed back to upstream algal cultivation
(outside the scope of this model). The methane-rich biogas is combusted in a gas turbine
to generate electricity, which is produced in excess of facility power demands and sold to
the grid. Flue gas heat is recovered by generating steam to satisfy process and utility
steam demands. A small amount of supplemental natural gas is co-fed with the biogas to
the turbines to meet facility heat balances. The digester effluent water contains nitrogen
and phosphorus nutrients which garner additional coproduct revenues, as does the solids
digestate cake which is sold for the nitrogen content as a land-application fertilizer.
Area 600: Storage. This area provides bulk storage for diesel, naphtha, and ethanol
products, as well as firewater and chemical inputs.
Area 700: Utilities. This area includes a cooling water system, chilled water system,
process water manifold, and power systems. The steam generation units and steam utility
loops are also included in this section.

5
 CO2 RECYCLE
TO PONDS

RECYCLE WATER

AMMONIA

MINOR
HP STEAM FERMENTATION
NUTRIENTS
ALGAE GROWTH
AND DEWATERING ACID
(not modeled here) WATER FOR VENT
SCRUBBER

A100 A200
ALGAL BIOMASS PRETREATMENT HYDROLYSATE FERMENTATION
AND CONDITIONING AND DISTILLATION
RECYCLE WATER

PURIFICATION
CHEMICALS

PRODUCT
ETHANOL
MAKEUP H2 SOLVENT
MAKEUP

A400
PRODUCT A300
PURIFICATION AND RAW OIL LIPID EXTRACTION STILLAGE

UPGRADING AND SOLVENT

RECOVERY

NAPHTHA AND A600

DIESEL PRODUCTS
STORAGE

LIGHT GAS TO CHP

SUPPLEMENTAL NG

WASTE FROM
PURIFICATION

A700
RECYCLE WATER
+ NUTRIENTS A500 UTILITIES
SPENT RESIDUE
ANAEROBIC
FLUE GAS DIGESTION AND
CHP STEAM

POWER FOR
PROCESS
AD DIGESTATE
SURPLUS
POWER
OVERALL PROCESS:
ALGAL FRACTIONATION PROCESS
JULY 2014 PFD-P120-A000

Figure 1. Simplified flow diagram of the overall process. (Key streams only; see Appendix D for more detailed diagram and PFDs.)

6
1.3 Techno-Economic Analysis Approach
Figure 2 describes the engineering approach used for modeling the conversion of biomass to
biofuels, including process design, process modeling, and economic analysis. This approach was
largely followed for this study as well, albeit under a condensed timeline and with additional
extrapolations around areas with more limited supporting experimental data to date (primarily
the extraction step as well as operability and yield assumptions related to anaerobic digestion of
a material with elevated nitrogen, discussed later). As such, this report is less prescriptive in
some sections than previous reports due to the early stage of understanding for this novel process
pathway and somewhat more preliminary nature of the associated models.

Engineering Companies
Consulting on Process Process Flow Diagrams
NREL- or other
Configuration
DOE-Sponsored
Research

Rigorous Published Literature

Estimates of Other
Material & Energy Balances
Commercial Technology
(Aspen Plus )
Outside Engineering
Studies (e.g., feed handling,
Cost Estimation separations)
Software (e.g., ICARUS )

Capital & Project Cost

Vendor Cost Quotations
Estimation

Engineering Company
Cost Estimations

Discounted Cash Flow

Economic Model

Minimum Fuel Selling Price

Figure 2. NRELs approach to process design and economic analysis

Starting from the general process flow diagram (PFD) shown in Figure 2 and the more detailed
PFDs contained in Appendix D, a process simulation is developed using Aspen Plus software
[15]. This process model computes thermodynamically rigorous material and energy balances for
each unit operation in this conceptual integrated conversion process. The material and energy
balance data from the Aspen simulation are next used to assist in determining the number and
size of capital equipment items. As process conditions and flows change, baseline equipment

7
costs are automatically adjusted in an Excel spreadsheet using a scaling factor. These baseline
cost estimates come from vendor quotes typically based on AACE Class 4 estimates (a favored
procedure for larger or non-standard unit operations and packaged or skid-mounted subsystems)
or from Harris Groups historical cost database (for secondary equipment such as tanks, pumps,
and heat exchangers). Harris Group provided new design and cost estimates based on vendor
quotations for critical areas of the process model for which similar unit operations had not
previously been considered in other analyses, including a liquid-liquid extraction system, solvent
distillation, and oil cleanup operations. Other process steps, such as dilute acid pretreatment
(fractionation) and neutralization, fermentation, ethanol purification, anaerobic digestion, and
hydrotreating, were based on recently updated Harris Group estimates from NRELs 2011
cellulosic ethanol design report and 2013 biological hydrocarbon design report [3,6], referred to
as such hereafter. Final equipment costs for this report are tabulated in Appendix A.
Once equipment costs are determined, direct and indirect overhead cost factors (e.g., installation
costs and project contingency) are applied to determine a total capital investment (TCI). The
TCI, along with the plant operating expenses (also developed using flow rates from the Aspen
model), is used in a discounted cash flow rate of return (DCFROR) analysis to determine a plant
gate price for the refined renewable diesel blendstock and ethanol products, combined together
based on energy content of each stream. This plant gate price is also called the minimum fuel
selling price (MFSP, in $/GGE) required to obtain a net present value (NPV) of zero for a 10%
internal rate of return (IRR) after taxes.

The product of the analysis described above is a techno-economic model that reasonably
estimates a product price for a pre-commercial process. The resultant MFSP is unique for the set
of process conditions simulated and it should be emphasized that a certain percentage of
uncertainty always exists around these chosen conditions, as well as around the assumptions
made for capital and raw material costs. Without a detailed understanding of the basis behind it,
the absolute computed MFSP carries a risk of being taken out of context. While the MFSP can be
used to assess the marketplace competiveness of a given process, it is best suited for comparing
technological variations against one another or for performing sensitivity analyses that indicate
where economic or process performance improvements are needed.

1.4 About nth-Plant Assumptions

The techno-economic analysis reported here uses what are known as nth-plant economics. The
key assumption implied by nth-plant economics is that our analysis does not describe a pioneer
plant; instead, it assumes several plants using the same technology have already been built and
are operating. In other words, it reflects a mature future in which a successful industry of n plants
has been established. Because the techno-economic model is primarily a tool for studying new
process technologies or integration schemes in order to comment on their comparative economic
impact, nth-plant analysis avoids artificial inflation of project costs associated with risk financing,
longer start-ups, equipment overdesign, and other costs associated with first-of-a-kind or pioneer
plants, lest these overshadow the real economic impact of research advances in conversion or
process integration. At the same time, NREL also continues to work on quantifying economic
factors associated with first-of-a-kind implementation. At the very least, these nth-plant
economics should help to provide justification and support for early technology adopters and
pioneer plants.

8
The nth-plant assumptions in the present model apply primarily to the factored cost model used to
determine the total capital investment from the purchased equipment cost and to the choices
made in plant financing. The nth-plant assumption also applies to some operating parameters,
such as process uptime of 90%. These assumptions were agreed upon by NREL and DOE for
this report and reflect our best estimates at the time of publication. It should be emphasized,
however, that these assumptions carry a degree of uncertainty and are subject to refinement.

1.5 About the NREL Aspen Model

While Aspen Plus can be thermodynamically rigorous, such detail is not always warranted in the
simulation, whether for lack of data or introduction of additional complexity for little gain in
accuracy. Some unit operations, namely lipid extraction and oil cleanup, were modeled with a
fixed set of inputs determined by expected performance for the given flowrates and stream
compositions confirmed with vendors. Bioreactors were modeled using experimentally-
determined conversions of specific reactions (e.g., glucose to ethanol) rather than using rigorous
kinetics or rate expressions. This simple stoichiometric model still satisfies mass and energy
balances.

The Aspen Plus simulation uses component physical properties internal to the software as well as
property data developed at NREL or from the literature [16,17]. Similar to other recent NREL
models, the current model does not rely on external property databanks and minimizes the
number of custom-defined components within reason. A discussion of the components and
properties used is given in Appendix C.

9
2 Design Basis and Conventions
2.1 Plant Size: Feedstock Logistics and Cost
Similar to the recently-released algal HTL design report [11], the scope of this analysis begins
with dewatered algal biomass feedstock delivered at 20 wt% solids, thus does not include
upstream biomass production and dewatering, but instead assumes a rolled-up feedstock cost that
accounts for all upstream activities. As noted above and discussed at length in recent
harmonization publications [8, 10], regardless of conversion method, all integrated systems
based on algal biomass processing are constrained by high cultivation variability, which is both
location- and season-dependent [18, 19]. This variability must be considered in TEA and LCA
analysis, as it carries non-trivial impacts on results given the requirements for systems
engineering to accommodate this variability. Specifically, PNNLs resource assessment
modeling under recent harmonization efforts identified suitable candidate locations that are
envisioned to provide an optimum between high cultivation productivity and low net water
consumption (after considering evaporation and precipitation at a given location), based on algal
biomass cultivation in large open raceway ponds. For both generic and specific strains (namely a
Chlorella strain titled NAABB 2412 or DOE 1412 [10, 20]), the RA model predicted swings in
cultivation productivity up to 10:1 between summer and winter seasons, with the most recent
models averaging out slightly more than 5:1 between these peak and minimum seasons as an
average across all selected sites across the U.S. Gulf Coast. While it is possible, and in fact
preferential, for a particular site and algal strain to fluctuate over a lower range of extremes
across the course of a year, the same degree of variability near 5:1 was assumed in this analysis,
to maintain consistency with the HTL design report [11].

The analysis assumes an annual average feed rate of 1,339 dry U.S. ton/day (1,215 metric
tonne/day, ash-free dry weight (AFDW) basis) of algal biomass delivered to the conversion
facility from upstream dewatering. This represents a roughly twofold increase over the 2013
HTL harmonization baseline of 650 ton/day (AFDW overall average) which corresponded to an
annual average 14.6 g/m2/day cultivation productivity rate [10] and thus implies an annual
average productivity target near 30 g/m2/day if all other upstream process and facility size
assumptions were identical to the previously published work. Likewise, the feedstock cost is
assumed at $430/ton (AFDW basis) as consistent with the HTL design report, which is based in
turn on DOEs most recent MYPP report that targets a 2022 cost for algal biomass at $430/ton
after cultivation (contributing $340/ton) and dewatering (contributing $87/ton) [9]. This cost
target also is predicated on an algal cultivation productivity increasing to 30 g/m2/day as well as
a number of cost reductions for cultivation and dewatering relative to the 2010 baseline as
presented in the referenced report. Specifically, the MYPP document presents a 2010 baseline
cost of $1,091/ton of algal biomass, based on the 2012 harmonization report assumptions for
initial benchmark cultivation and dewatering performance. Relative to this benchmark, the 2022
target scenario stipulated that algal cultivation productivity increases from 13.2 to 30 g/m2/day
and removes the use of costly plastic pond liners, while reducing dewatering costs by 50% per
amount of harvested material processed.

When the 1,339 ton/day annual average feed rate basis is combined with seasonal fluctuations as
presented in Davis et al. [10], the resulting seasonal feed rates to the conversion facility are
shown in Figure 3. Two cases are considered: one where the facility is sized to the peak summer
flowrate and remains partially under-utilized for the remainder of the year, and a second where

10
the amount of material produced in excess of the second-highest season (spring) is diverted,
dried, and stored to be used in the winter.

2500
2,229 Seasonal variability
Feed rate dry ton/day (AFDW) Dry/store peak summer capacity
2000
Average = 1,339 ton/day
1,449 1,449 1,449
1500
1,264 1,264
1,195
1000

500 416

0
Summer Fall Winter Spring

Figure 3. Seasonal and annual average feed rates to conversion facility. (Full seasonal variability
without (blue bar) and with (red bar) excess peak capacity storage for use in winter.)

As shown in Figure 3, if the facility were designed to accommodate the raw peak summertime
capacity, installed capital costs would be considerably under-utilized for the remainder of the
year, particularly in the winter season which would also imply a required turndown capacity for
the equipment of roughly 80% (5.4:1 variation) in instances where only a single unit operation is
used rather than multiple units in parallel. In contrast, if excess summertime capacity were
diverted, dried, and stored for later use in the winter, the capital utilization turndown factor
would decrease to a significantly lower 18% (1.2:1) variation. Given more favorable operability
as well as modeled MFSP costs (presented later), the latter approach to divert the excess material
for use in the winter was selected as the base case for the analysis, which is consistent with the
HTL design report approach [11]. Still, the practice of excess summer drying adds both capital
costs for drying equipment as well as operating costs for natural gas (which also brings
sustainability implications), and also adds risks of stored biomass spoilage that may require
additional mitigation measures. Additionally, there is also a risk that the biomass could behave
differently after being dried and reconstituted, but this is not yet known. Design and cost
implications associated with this approach are discussed in Section 3.1.

2.2 Feedstock Composition

The algal biomass composition is based on recent NREL analytical characterizations [13], which
were conducted for two separate fresh water algal strains over three harvesting time points as
originally established under the SABC consortium [14]. Specifically, these were Scenedesmus
sp. and Chlorella sp. provided by Arizona State University from harvests taken in early-, mid-,
and late-cultivation stages, corresponding to low-, mid-, and high-nitrogen starvation states, and
also to high-protein, high-carbohydrate, and high-lipid content biomass, respectively. By timing
the harvest, biomass of different composition was obtained in a controlled fashion [21].
Cultivation time after reaching nutrient deplete conditions depended on final target biomass

11
composition desired, which, depending on season, typically was 3 to 5 days for high
carbohydrate (midpoint harvest) biomass and 6 to 9 days for high lipid (late harvest) biomass.
High protein (early harvest) biomass was obtained by harvesting prior to nutrient depletion.
Although this analysis focuses on fresh water media, the use of brackish or saline water strains
may also be plausible for this technology pathway, although the efficacy for this option has not
been experimentally investigated; at a minimum, biomass ash content would be considerably
higher for saline species which could translate to increased lipid purification demands following
the extraction step, and operation of anaerobic digestion could become challenging at high salt
levels.

More favorable cultivation and processing performance as well as higher theoretical yield
potential favored focusing on the Scenedesmus strain for TEA investigation; thus we consider
this strain for both the mid- and late-harvest time points, but exclude the early-harvest case as it
does not possess substantial levels of fuel precursor components. An abbreviated table for these
specific cases is presented below (Table 1) showing both elemental and component
compositions, excerpted from a more complete table for all strain/harvest combinations
documented in analysis by Laurens et al. [13]. While only the mid- and late-harvest points are
considered for modeling purposes, the measured early-harvest composition is also included to
demonstrate the clear shift in biomass composition that occurs over the harvest regime originally
studied. The measured shift in composition of the biomass across the early, mid and late stages
of cultivation is considered typical for nutrient limited biomass with carbohydrate, protein and
lipid content values consistent with earlier reports for Chlorella and Scenedesmus sp. [22-24].
The ash content reduced from 6.7% to < 3% between the early and late-harvest points. The
protein content is inversely related to the lipid content and was found to drop rapidly from 34%
in the early harvest biomass to 8.9% in the late harvest biomass, while lipid content increased
from ~7% to 41%. The nitrogen content underpins the protein content with a consistent
relationship of a nitrogen-to-protein conversion factor and previous reports have shown a protein
content of <10% for nutrient limited biomass [22]. In Table 1 and hereafter, HCSD represents
high-carbohydrate (mid-harvest) Scenedesmus, and HLSD represents high-lipid (late harvest)
Scenedesmus. Likewise, HPSD represents high-protein (early-harvest) Scenedesmus, but is
only included below in the original compositional analysis data and not considered further in this
TEA evaluation.

12
 Table 1. Algal Biomass Elemental and Component Composition for Early-Harvest (HPSD), Mid-
Harvest (HCSD) and Late-Harvest (HLSD) Scenedesmus. (Left columns are raw data from
experimental measurements, right columns are adjusted values to achieve 100% mass closure for
modeling purposes. Excerpted from [13]. HPSD case was not considered for modeling purposes, but is
included below to highlight the shift in biomass composition over time)
Scenedesmus sp. Scenedesmus sp.
(measured) (input to model)
Early Mid Late Mid Late
Wt% composition (dry basis)
(HPSD) (HCSD) (HLSD) (HCSD) (HLSD)
C 49.3 52.9 56.3 54.0 57.4
H 7.1 8.0 8.5 8.2 8.7
N 8.3 1.8 1.6 1.8 1.6
O 28.7 34.8 31.2 35.5 31.8
S 0.0 0.2 0.2 0.2 0.2
P 0.97 0.22 0.20 0.22 0.20
Total a 94.4 97.9 98.0 100.0 100.0
Ash 6.7 2.3 2.1 2.4 2.1
Fermentable carbohydrates 24.3 46.3 37.9 47.8 38.2
-Starch 6.9 12.2 8.1 47.8 c 38.2 c
-Non-starch glucose 6.8 22.6 18 - -
-Mannose 7.2 11.5 11.8 - -
Other carbohydrates b 3.4 1.6 1.3 5.0 3.9
Protein 34.5 12.8 8.9 13.2 9.0
Lipids (Fuel-relevant lipids as FAME) 6.6 26.5 40.9 27.4 41.2
Non-fuel polar lipid impurities ND d ND ND 2.7 2.1
Cell mass ND ND ND 1.6 3.5
Total 75.5 89.5 91.1 100.0 100.0
a
Measured element balance on dry weight basis; adjusted to AFDW basis for model
b
Non-fermentable carbohydrates
c
For modeling purposes, all fermentable carbohydrates are assumed to start in polymeric starch
glucan form with subsequent hydrolysis to monomeric hexose sugar
d
ND = not determined

As shown in Table 1, a number of adjustments were made to the baseline raw experimental
composition values for use in the models. First, elemental compositions were adjusted to 100%
mass closure on an ash-free dry weight basis. For component compositions, initial measured
mass balance closures of the HCSD and HLSD cases were near 90% (including ash). Rather than
adjusting all components evenly to close the balance to 100%, specific components were
adjusted first based on the level of certainty in analytical measurements. Specifically, the routine
analytical method used to determine lipid content based on total fatty acids did not account for
the non-fatty acid polar portion of the lipids or for hydrocarbon-based compounds that are
present in the biomass. The non-fuel polar fraction of lipids can contribute a sizeable amount to
total lipid mass and consist of for example, the polar headgroups of lipids (e.g. phosphatidic acid
of phospholipids, digalactosyl moieties on glycolipids, etc.). Literature indicates that the non-fuel
polar lipid fraction can particularly make up a considerable portion of the lipids in early-harvest
nutrient replete conditions, with a drastic reduction in polar lipids as the lipid content increases
under nutrient limited cultivation scenarios (e.g. moving towards mid-harvest HCSD and further
to late-harvest HLSD cases) [25, 26]. This guidance was applied in adding a non-fuel polar lipid
impurity fraction to the biomass composition as shown in Table 1, assumed to constitute 10% of
the remaining fatty acid lipid fraction in the HCSD case and dropping to 5% of this fraction in
the HLSD case. Intact lipid composition for these biomass samples is required to definitively

13
measure the polar lipid fractions. Of the polar lipid impurities fraction, a portion is assumed to
be extracted with the rest of the lipids and requires subsequent removal in a lipid cleanup step,
discussed below in Section 3.4.

Other, hydrocarbon-based lipid species such as straight and branched-chain hydrocarbons, sterols
and isoprenoid-based compounds were also found to be present in the biomass but do not convert
to fatty acid methyl esters (FAME) when transesterified according to NREL published methods
[27]); thus, the lipid fraction shown in Table 1 is based on fatty acid lipids as FAME. The
additional non-fatty acid hydrocarbon-based compounds would not be counted in such a
measurement but are still likely to be extracted in a hexane-based lipid extraction protocol, and
ultimately upgraded to additional fuels through hydrotreating. It is possible that these compounds
add a small fraction to total fuel potential lipids, but for the purpose of this analysis this
fraction will conservatively not be considered as it was not characterized in the compositional
data (Table 1) utilized as inputs to the model. It does, however, provide a small margin of
safety in the targeted lipid extraction efficiency yield discussed in Section 3, as a given
extraction yield of FAME lipids would be equivalent to a slightly lower extraction yield of total
(FAME + non-FAME but still convertible) lipids.

Following the polar lipid impurity fraction, the next highest level of uncertainty is around non-
fermentable carbohydrates, which includes species such as uronic acids, amino sugars, deoxy
sugars, etc.; thus, this fraction was conservatively tripled relative to the baseline analytical value
for the modeling basis. Finally, a small 1.5-3.5% fraction (HCSD and HLSD respectively) was
added to account for other non-characterized cell mass such as nucleic acids based on prior
internal analysis. These three additions increased component mass closures to 97%-99%. To
reach 100% mass closure, all components were then subsequently adjusted evenly.

While a large fraction of fermentable carbohydrates were characterized as non-starch glucose

and mannoses carbohydrates, the specific nature of these non-starch carbohydrates is not yet
known; thus for modeling purposes the fermentable carbohydrate fraction was assumed to
initially start off in polymeric glucan form with all fermentable carbohydrates modeled as starch.
This allows the model to calculate heat of reaction for a specified glucan carbohydrate and
subsequently estimate the steam demand and overall pretreatment reactor heat balance required
to yield the stipulated monomeric sugar release.

Although we conduct TEA modeling for both the HCSD and HLSD cases (with results presented
in Section 5.2), the HLSD biomass serves as the official base-case for subsequent discussion and
focus for process and economic results. This case was selected due to its higher net fuel potential
(driven by substantially higher lipid content) and resulting superior economics, and in keeping
with the focus of this technology pathway being algal lipid extraction and upgrading (ALU).
However, it is recognized that this case requires longer cultivation time to reach nutrient
deprivation and associated high lipid levels, which must be considered in understanding
ultimately achievable biomass growth rates and associated feedstock costs from the cultivation
step. This is outside the scope of this study, but will remain a focus area for analysis moving
forward. In reality, a more practical out-year target case may lie somewhere in between the
harvest points (and implicit productivity rates achievable in the future) attributed to the HCSD
and HLSD cases. All modeling scenarios assume a delivered solids content of 20 wt% after
upstream dewatering.

14
2.3 Modeling Basis Parameters
Similar to other biofuel pathways recently published in design reports under DOE-BETO efforts
[6, 11], a number of key modeling assumptions are fixed constant to allow for comparable
modeling outputs. These parameters include:

1. Design and construction time: 36 months

2. Facility start-up time: 0.5 year (6 months)
3. Cost-year dollar basis: 2011 dollars (all cost results presented here will be in 2011-year
dollars)
4. Facility on-stream time: 90% (330 days/year or 7,920 hours/year). Note, this is intended
to represent an nth-plant facility, and may be a higher factor than what is achievable
today, particularly when considering the algal cultivation step. This on-stream factor
assumes year-round operation with approximately one month per year allocated to
facility shut-down, whether due to maintenance, upstream pond upsets (such as culture
crashes, freezes, etc.), or other downtime factors.

2.4 Design Report Conventions

2.4.1 Units
The Aspen Plus model we developed is based on the same set of units as used in NRELs 2013
biochemical hydrocarbon design report model [6], as it was constructed by modifying this model
which also follows a similar process using dilute acid pretreatment and fermentation operations;
these units include kg, kmol, atm, C for materials, and MM kcal (Gcal) for energy. Values in
this report that were retrieved directly from the Aspen model therefore tend to be reported in
these units. Harris Group preferred to use U.S. standards (lb, BTU, F, gal, etc.) when
communicating with equipment vendors. Therefore, certain equipment specifications are cited in
these U.S. units.

Note that in the present report, certain quantities (e.g., yields and costs) are computed and
reported in terms of tons. To avoid ambiguity, tonne will denote a metric tonne (1,000 kg)
and ton will denote a short or U.S. ton (2,000 lb). In general, the U.S. ton is the standard for
this document. Ton also appears in Section 3.7 in the context of refrigeration, but this usage
should be clear from the discussion.

2.4.2 Product Density and Heating Value

The results from this analysis are reported primarily in terms of energy yields in gallons gasoline
equivalent: $/GGE, GGE/yr, GGE/ton, etc. This is done to maintain focus on total fuel (energy)
yield from the relevant precursor components in the starting biomass, whether ultimately
converted to diesel (RDB) or ethanol, and also because ethanol is intended to serve as a
representative end-product of carbohydrate utilization, among other options, including
biological or catalytic conversion of sugars to hydrocarbon fuels and/or products. Additionally,
and more important from a TEA standpoint, the volumetric split between RDB and ethanol
products is evenly divided, varying from 50% (HCSD) to 65% (HLSD) RDB versus ethanol. In
such cases where two products are produced in similar volumes, it is most appropriate to
consider the summative energy-equivalent (GGE) yield for both products, to avoid introducing

15
dramatic sensitivity in the MFSP result if one of the products were instead assigned a coproduct
value (this approach is consistent with a recently published woody biomass pyrolysis design
report which also produces similar volumes of different fuel products [28]). Thus, the main two
products, RDB and ethanol, are adjusted by heating values to a single GGE fuel yield, while a
much smaller coproduct, naphtha, is appropriately assigned a coproduct value and generates a
small amount of coproduct revenue.

To adjust to an energy-equivalent fuel yield, component heating values were set at 23.6 MJ/L
(84,530 BTU/gal, HHV basis) for ethanol based on standard published values [29], and were
calculated (HHV basis) from the Aspen model for the RDB product based on the component
mixture. A higher heating value was selected in this case as it allowed for a more readily
achievable comparison to published values for the RDB product, in order to validate a close
resemblance between Aspen calculations and accepted energy content for similar products.
Indeed, the calculated HHV value for the RDB product was 130,430 BTU/gal, which is roughly
0.3% lower than a published value for hydrotreated renewable diesel at 130,817 BTU/gal [29].
To subsequently translate to a GGE basis, a conventional gasoline heating value of 124,340
BTU/gal (46.54 MJ/kg, HHV basis) was applied [DOE 2012]; this corresponds to a lower
heating value of 116,090 BTU/gal gasoline. Similarly, where densities are required to calculate
product volumes, a standard published ethanol density of 0.789 kg/L at 20 C (68 F) was
applied [3], while the Aspen model calculated the density for the RDB and naphtha products at
0.768 and 0.730 kg/L respectively (also at 20 C).

16
3 Process Design and Cost Estimation Details
The process design described in this study is based upon experimental demonstration results
observed first under the SABC consortium [14], then more recently in follow-up NREL work as
described by Laurens et al. [13], projected out to plausible future improvements in key process
areas. Some areas of the process draw from previous NREL experimental and analysis efforts,
namely dilute acid pretreatment and fermentation operations, but aside from expertise gained in
optimizing such operations and associate process integration related to their use, this process is
unique given the presence of lipids and absence of lignocellulosic components such as lignin and
pentose carbohydrates; this makes for a more ideal process in some respects but adds new
challenges in others, as documented in the following sections. It is important to reiterate that the
targets presented here are merely one set of conditions that would enable achieving the 2022 cost
goal below $5/GGE and will help to inform near-term research directions, but do not necessarily
represent the most optimal possible scenario. Additionally, given the scope and timeframe under
which the present analysis effort was conducted, the report is more concise in some areas (such
as detailed process flow diagrams and stream-level information) relative to earlier NREL design
reports. This section describes the process as modeled and discusses the influence of specific
R&D goals in the decision-making process.

With respect to costing estimates, given the more preliminary scope and nature of the design
models established here, the engineering support provided by Harris Group focused primarily on
the major unit operations (e.g. extraction and distillation columns, oil cleanup and upgrading
facility costs, etc.), and did not obtain vendor pricing for minor supporting equipment such as
pumps, heat exchangers, and agitators. For this minor equipment, Harris Group added a
percentage factor to the provided cost estimate for the respective major operations, which varied
according to process area as discussed below, or NREL estimated costs using Aspen Economic
Evaluator [15] for a number of heat exchangers. Additionally, at the request of Harris Group who
provided all relevant vendor estimates for cost quotations, vendor company names associated
with a particular unit operation will not be provided in this report.

3.1 Area 100: Pretreatment and Conditioning

3.1.1 Overview
The process begins with algal biomass delivered to the conversion facility at 20 wt% solids after
processing through upstream dewatering. As discussed above in Section 2.1, the raw feed rate to
conversion varies seasonally over a range of roughly 5.4:1 between summer and winter, with fall
and spring flowrates both roughly 40% lower than peak summer flows, reflecting the same
degree of variance upstream in the cultivation step. Two options are presented above for
handling this variance: (1) to feed the material straight to the facility with no pre-processing, thus
requiring most capital expenses to be substantially over-designed for the remainder of the year
(particularly in the winter), or (2) to divert, dry, and store the fraction of peak summertime
capacity that is produced in excess of the spring season (the next highest production season), and
to use it during the winter months to reduce the dramatic swings in throughputs and equipment
over-design. Of these two options, the latter scenario to divert a portion of summer production
for use in the winter is selected as the base case for the primary focus of this analysis. This is
done to maintain consistency with the above-referenced algal HTL design report [11] as it allows
for more realistic commercial operability without the need for substantial equipment turn-down

17
ratio capabilities, and also because it ultimately achieves a lower fuel production cost, as will be
presented later. It is recognized that this choice carries a sustainability penalty, namely the added
requirement for natural gas drying of the diverted summertime material, but the overall
greenhouse gas impact is diluted given that this step is only required for three months out of the
year and not utilized for 75% of the year. Moving forward, NREL will work with partners at
ANL to more fully characterize the complete LCA of the integrated system, and if the level of
natural gas for drying in the summer proves to be unacceptable, the alternative option may
instead be selected (we show in Section 5.2 that economics are still reasonable for the case
without summer diversion). Additionally, alternative summer drying techniques may be possible
such as solar drying to reduce the sustainability burden for natural gas drying, but this is not
considered here. Storing dried biomass for 3-6 months also brings a risk for biomass spoilage via
oxidation or microbial degradation, and may require additional mitigation measures to prevent
which are not considered here.

The amount of delivered feedstock material diverted to be dried in the summer equates to 35% of
total summertime feed rate (2,229 ton/day AFDW basis), thus 780 ton/day of dry biomass
(AFDW basis) is diverted away to drying and 1,449 ton/day is sent on to the pretreatment
fractionation step. Consistent assumptions for the drying operation were employed as
documented in the algal HTL design report [11], namely a purchase cost of $905,000 per dryer
(2013-dollars), each accommodating a flow up to 2,520 kg/hr of moisture-free biomass for
drying the material from 80% to 10% moisture. The dryer is natural-gas-fired and operates at
77% thermal efficiency (i.e., heat required for drying versus heat content of the natural gas).
These values are based on a quotation for a rotary drum dryer. Also consistent with the HTL
report, the cost for storing the dried material until the winter is assumed to be accounted for in
the indirect capital cost factors presented in Table 15. Subsequently, during the fall, winter, and
spring seasons, all delivered material is sent straight to the fractionation step at a flow rate of
1,264, 416, and 1,449 ton/day (AFDW basis) respectively, with the 780 ton/day from storage
added in the winter to constitute a total winter flow to fractionation of 1,196 ton/day.

The biomass pretreatment (fractionation) step is a crucial operation to enable successful recovery
and conversion of both the carbohydrate and lipid components. In the context of lignocellulosic
(terrestrial) biomass processing, dilute acid pretreatment breaks down primarily hemicellulose
compounds to their monomeric sugar constituents (namely xylose, mannose, arabinose, and
galactose) by hydrolysis reactions, while making the remaining biomass more amenable to
further deconstruction downstream, such as enzymatic hydrolysis of cellulose to glucose [3, 6].
While such recalcitrant carbohydrates are largely not present here, a similar dilute acid operation
has been shown in literature to enable sugar recovery from algal biomass [30, 31] and is required
to disrupt the algal cell wall, achieve high soluble sugar recovery, and enable effective
downstream solvent extraction. Polymeric glucan carbohydrates (primarily starch) have also
been observed to hydrolyze to monomeric glucose through the dilute acid pretreatment step, and
NREL experimental efforts have found that no amylase or cellulase enzymes are required (e.g.,
for targeting either starch or cellulose carbohydrates) following the pretreatment operation [32],
as the majority of the carbohydrates are seen to be less recalcitrant than cellulose and
hemicellulose components typically present in cellulosic feedstocks after processing through acid
pretreatment. This eliminates the need for costly enzymatic saccharification operations that have
been pursued in literature as an alternative means to algal sugar/ethanol production [33-35].

18
Pretreatment fractionation occurs at a temperature of 130-180C and corresponding bubble-
point pressures, with a residence time ranging from 1 to 10 minutes. Following pretreatment, the
hydrolysate slurry is flash-cooled, vaporizing a fraction of water which is condensed and routed
to the water recycle pool. The hydrolysate is then sent to a conditioning reactor, where
ammonium hydroxide is used to raise its pH to 5.

Figure 4 shows a simplified flow diagram of the pretreatment area.

Algal biomass feed Solvent recovery pre-heat

exchanger

To water
Vertical presteamer recycle pool
CW
High-pressure steam

Horizontal pretreatment reactor NH3

High-pressure steam + acid
Hydrolysate to
fermentation

Blowdown Ammonia
tank conditioning

Figure 4. Simplified flow diagram of the pretreatment and conditioning process

3.1.2 Design Basis

In principle, the acid pretreatment reactor system is the same as that described in earlier NREL
reports for biochemical conversion of corn stover [3, 6] and will not be repeated in such detail
here. In summary, this system includes a feedstock receiving system, followed by a vertical
vessel with a long residence time for steam-heating and potential acid impregnation of the
biomass, followed by the horizontal pretreatment reactor that operates at a higher pressure and a
short residence time. The 26-inch plug screw feeder is a rugged, high-compression screw device
designed to form a pressure-tight plug of material through axial compression. Dilute sulfuric acid
is metered at the discharge spool of each plug screw feeder, shown in Figure 5. Feedstock drops
from the plug screw discharge into a mixing and heating screw, which discharges the feedstock
into the top of the presteamer. High-pressure steam is injected into this vessel to maintain
temperature. The current model assumes operation of the presteamer at 100C such that no
significant reactions occur in this section. It can be used if additional holdup is required for acid
hydrolysis. Feedstock flows downward through the vertical presteamer with uniform temperature
throughout, discharging through a dual screw outlet device to two plug screw feeders. The plug
screw feeders meter feedstock to the horizontal pretreatment reactor, with a horizontal reactor
configuration chosen to allow tighter residence time distribution control than with a vertical
reactor.

While historically NREL has operated the pretreatment step at 30 wt% total solids for processing
corn stover, in this case, the solids content both entering and exiting the reaction step (after flash)

19
are 20 wt%. However, pilot-scale work has already shown effective operation of the unit near
20% solids when processing algal biomass using NRELs 120 L horizontal batch reactor, with
trials conducted to date on up to 25 kg of dry weight biomass at 20% solids loading [36]. Thus,
from an operational standpoint, the lower solids loading is not anticipated to be problematic.
Acid loading in the reactor was set at a target 1 wt% relative to feed slurry liquor (water), based
on assumed improvements relative to early experimental efforts using 2 wt%. This appears to be
a reasonable target reduction, as it translates to an acid loading of 40 mg/g dry biomass,
compared to corn stover processing at lower values closer to 9 mg/g [6]. The effective sulfuric
acid concentration in the pretreatment reactor (estimated at roughly 0.8 wt% after dilution by
condensing steam is accounted for) likely requires maintaining a high-cost Incoloy 825 cladding
metallurgy, rather than switching to lower-cost stainless steel alloys, consistent with prior cost
estimates for this operation. The potential impact of lower-cost metallurgy and other
pretreatment reactor cost savings is addressed in Section 5.1. The reactor pressure is held at the
bubble point for the mixture. Heat losses from the reactor are not accounted for in the energy
balance calculations. The residence time in the pretreatment reactor is nominally 5 minutes;
however, given the seasonally variable feed rates discussed previously for this pathway,
residence time may also fluctuate by a similar 20% (1 minute) between summer and winter
seasons, although the screw speed may also be adjusted to maintain targeted residence time.

A comparison of current (experimental baseline) operational parameters for the pretreatment

reactor compared to the target case is summarized in Table 2.

Table 2. Pretreatment Conditions Applied in Experimental Work to Date [13], Compared to this
Design
Parameter Experimental baseline Present design target
Sulfuric acid loading (wt% of 2% 1%
feed liquor rate)
Residence time 1-5 minutes 5 minutes
Temperature 145C 150C
Pressure 4.4 atm 4.6 atm
Total solids loading 20-25 wt% 20 wt%

Table 3 summarizes the reactions and conversions that take place in pretreatment. As noted
previously, although starch carbohydrates have been measured to constitute 20%-25% of total
fermentable carbohydrates in both HCSD and HLSD algal biomass materials (with the remainder
being non-starch glucose and mannose carbohydrates), for modeling purposes, all fermentable
carbohydrates were assumed to initially start out as starch in the feed biomass. This allows for
calculating heat of reaction and subsequently setting high-pressure steam requirements and
pretreatment heat balances based on a defined glucan carbohydrate (with the remaining
carbohydrates not yet characterized as to their compositional nature or recalcitrance), as well as
allows for easier tracking of overall sugar yields in the model. Thus, for this discussion, starch
glucan represents the total fermentable carbohydrate portion for the HLSD biomass base case,
constituting 38.2 wt% of the biomass as shown in Table 1. Likewise, with glucose and mannose
constituting the large majority of monomeric sugars in the biomass (both hexose sugars), all
monomeric sugars were modeled here as glucose for the sake of modeling simplicity. Being the
same composition and molecular weight, glucose and mannose behave the same in the model

20
and have both been experimentally demonstrated to be equally fermentable to ethanol when
using S. cerevisiae [13, 37]. With these points in mind, the model targets a glucan-to-glucose
yield of 90% (i.e., 90% recovery of all fermentable carbohydrates as soluble hexose sugar).
Additionally, hydroxy-methyl furfural (HMF), a common fermentation inhibitor, is assumed to
be produced at the same level as in prior design reports for corn stover [6]. Low levels (0.1-0.2
g/L) of HMF have been measured in hydrolysates from algal biomass pretreatment, but not at
levels that became inhibitory to subsequent ethanol fermentation, and no furfural was detected
[13]. Other inhibitors may also be produced from pretreatment, and additional characterization is
required to further quantify these details moving forward.

Table 3. Pretreatment Hydrolysis Reactions and Assumed Conversions

Reaction Reactant % Converted
to Product
(Glucan)n + n H2O n Glucose a Glucan 90%
(Glucan)n n HMF + 2n H2O Glucan 0.3%
a
Glucan is assumed in the model to represent all initial fermentable carbohydrates and
glucose represents all hexose sugars, e.g., glucose and mannose

Relative to the 90% target for solubilized glucose yield (recovery from initial carbohydrates),
current experimental work has demonstrated up to 77% recovery of sugar from algal
carbohydrates to date when following the same reactor configuration modeled here (pretreatment
prior to extraction) [13]. Thus, the target of 90% represents a reasonable improvement
anticipated to be achieved through continued research and reactor optimization moving forward,
e.g. optimizing acid loading, temperature, and time, or considering alternative mechanical or
alkaline pretreatment steps. The pretreatment reactor is discharged to a flash tank. The pressure
of the flash is controlled to keep the hydrolysate at 20 wt% total solids content. The flash is
condensed by cross-exchange with the solvent recovery distillation preheater downstream, then
cooled and routed to the water recycle pool, without the need for cleanup, as it contains only a
minor fraction (~4 ppm) of HMF. After the flash, the hydrolysate whole slurry containing 20%
total solids and 13% insoluble solids is sent to conditioning, namely neutralization by ammonia
in stoichiometric quantities. Ammonia gas is mixed into dilution water to raise the hydrolysate
pH to 5. The residence time for neutralization is 30 minutes and the dilution cools the slurry from
117C to 100C. The composition of the stream at this point, and other major points throughout
the process, is shown in the stream table information in Appendix D. The material from
conditioning is conveyed to a hydrolysate storage tank.

3.1.3 Cost Estimation

The cost for the rotary drum dryer discussed above, was set based on the HTL design report at a
purchase cost of $905,000 per dryer (2013-dollars), each accommodating a flow up to 2,520
kg/hr of moisture-free biomass [11]. As the facility is already at a slight heat deficit (requiring
marginal natural gas to meet all facility utility heat demands), no excess heat is available to
operate the dryer and summer drying requires additional natural gas during the three months that
the unit operates, delivered to the dryer based on 77% thermal efficiency.

The acid pretreatment reactor design and cost basis was updated in NRELs 2013 biochemical
hydrocarbon design report to allow for a more robust cost estimate with higher flexibility in
operational variables. Namely, an evaluation of pretreatment reactor pricing in terms of

21
metallurgy, reactor conditions, and reaction residence time was conducted by Harris Group based
on input from a vendor. In addition to the base case Incoloy-clad metallurgy used as the baseline
assumption, a less costly duplex stainless steel option was considered as an alternative. Further
details behind the cost study may be found in the 2013 design report [6]. The effort established
correlation equations that allow for capital cost estimation of the pretreatment reactor system as a
function of residence time for the two reactor metallurgy choices:

Total Equipment Cost (2011 $MM); Incoloy-825 = 16.4+7.4m (1)

Total Equipment Cost (2011 $MM); Duplex steel = 16.8+3.9m (2)

Where, m = integer (minutes of resident time/3.3)

The above equations are based on an original design basis of 1,516 dry tonne/day biomass feed
rate to the pretreatment operation, and may be scaled to a different basis following standard
scaling-factor adjustments as discussed in Section 4.2. For the conditions used in this design, the
recommended metallurgy from the vendor continues to be Incoloy-825, with a total installed
capital cost of approximately $43.9MM (2011-dollars) for the system, including feedstock
receiving bin, pre-steaming, pressurized heating, reaction, and flash cooling. This cost, and all
subsequent downstream capital costs discussed in this report for the base case, are based on the
summer storage scenario peak-season feed rate to pretreatment of 1,449 AFDW ton/day in the
summer and spring seasons, after diverting excess summer capacity to the dryer (Sections 2.1
and 3.1.1). The reactor schematic is shown in Figure 5, and is similar to the configuration
presented in earlier reports [3, 6].

In addition to the cost for the pretreatment reactor system itself, a balance of plant factor was
applied to account for other minor equipment supporting the pretreatment fractionation system,
including the neutralization tank, flash tank, and other minor tanks, pumps and agitators. This
balance of plant factor was calculated as 4.6% of the pretreatment system cost based on costing
details presented in the 2013 design report [6]. Finally, two heat exchangers which had not been
employed in earlier analyses were added in this case; one, which cross-exchanges heat between
the pretreatment reactor feed and effluent streams (after steam addition and
pretreatment/neutralization reactions), and a second which condenses the pretreatment flash
steam by exchanging with the downstream solvent distillation pre-heater. Both shell-and-tube
exchangers were costed in Aspen Economic Evaluator [15], as typical for more minor operations
such as these.

Area 100 contributes about $0.50/GGE to the MFSP, including dryers, acid pretreatment, and
conditioning. About 56% of this contribution is attributed to capital cost, of which acid
pretreatment equipment accounts for the large majority (70%) of total capital expenses.

22
 Feedstock
Feedstock

Flush water
Screw feeder

Steam

Vertical
presteamer

Flush water Screw feeder

Acid
Hydrolysate
Horizontal reactor Vent
to fermentation
Pressate Steam

Pressate
reinjection

Figure 5. Horizontal acid pretreatment reactor design scheme

3.2 Area 200: Fermentation and Distillation

3.2.1 Overview
In this process area, released soluble sugars are fermented to ethanol using Saccharomyces
cerevisiae D5A (yeast) which readily utilizes both glucose and mannose as the two primary
sugars released from pretreatment. Fermentation proceeds in batch bioreactors at a total batch
time of 1.5 days, after which point the fermentation broth is sent to ethanol purification
consisting of beer and rectification distillation columns and vapor-phase molecular sieve
adsorption, which ultimately concentrates the ethanol product up to 99.5%. The vapor overhead
from the beer column and fermentation vents, containing CO2, ethanol, and other minor
components, is sent to a scrubbing column to recover the majority of the volatilized ethanol. The
beer column stillage, containing water and all remaining algal components, is routed to lipid
extraction. Figure 6 shows a simplified flow diagram of the fermentation and ethanol purification
process.

Given the lack of lignin, xylan, and other impurity components present in the algal biomass
which are typical for lignocellulosic feedstocks, the produced sugars entering the fermentation
step are relatively clean and readily convertible to a wide range of energy or chemical products,
and represent roughly 38 wt% of the starting biomass following 90% solubilization of
fermentable glucan carbohydrates from upstream pretreatment. While DOE-BETO has begun to
transition from a primary R&D emphasis on ethanol to hydrocarbon fuel and blend stock
products in light of having sufficiently demonstrated R&D progress through 2012 on cellulosic
ethanol technology pathways [38-40], ethanol is reasonably maintained in this particular pathway
as a fuel coproduct alongside renewable diesel and is intended to serve as a representative
energy product from monomeric sugars (with fuel yield translated to a GGE energy basis for
both products). A number of other options exist to instead convert sugars to hydrocarbon fuels or

23
chemicals, via either biological or catalytic routes [41, 42]. Such options could include
commodity fuel precursor molecules such as fatty acids, fatty alcohols, isoprenoids, or paraffins
with cost and yield implications as detailed in [6], or high-value coproducts including organic
acids, diols, and other compounds for which even a small amount of production may garner
significant revenue but must be balanced against market volume limitations (discussed further in
Section 5.4). Moving forward, such options will also be considered, but in this analysis, the
decision was made to focus on ethanol as the product from sugars given favorable in-house data
already established for this step within the integrated algal fractionation process [13].
Nonetheless, this highlights another key advantage of this technology pathway in possessing
process flexibility around utilization of carbohydrates, as well as lipids and protein fractions of
algal biomass.

Fermentation vent CO2 Recycle to ponds

Seed train
Water

Vent
Stillage to scrubber
extraction Vent

Hydrolysate from
pretreatment
Beer
CW well CW

Ethanol + water Ethanol

Product
CW

Broth
Regenerate
Batch fermentation reactor Beer
LP Molecular Sieve
column Steam Adsorption
LP
Steam
Boiler Feed Water
Rectifier (BFW) Economizer #1
To water
recycle pool
BFW

Figure 6. Simplified flow diagram of the ethanol fermentation and purification process

3.2.2 Design Basis

Ethanol fermentation is a well-characterized process that largely follows process and design
details as documented in NRELs 2011 cellulosic ethanol design report [3], with a number of
minor differences. After first being cooled by cross-exchange with the feed stream entering
pretreatment, the neutralized hydrolysate is further cooled by cooling water to the required
fermentation temperature (37C for S. cerevisiae). Ten percent of the hydrolysate stream is split
off to inoculum seed growth to initiate fermentation, utilizing consistent design assumptions for
the seed train system as described in [3] by analogy from previously modeled Z. mobilis
propagation and ethanol fermentation. In summary, two parallel inoculum seed trains, each train
consisting of five reactors in series, operate in batch mode with a 24-hour batch time and an
additional 12-hour turnaround time. The seed reactors are cooled with chilled water from Area
700 to maintain the temperature at 37C. The first vessel is inoculated with a seed culture from
the lab. Its broth is used to inoculate a larger reactor, and so on. After five iterations, the cell
mass from the last vessel is sufficient to inoculate the production vessel. However, at the design
feed rate utilized here (1,449 ton/day in the summer case), a considerably lower amount of
ethanol is produced than in the 2011 design case (3,263 gal/hr in the summer design case versus
7,257 gal/hr in the 2011 ethanol design case). Thus, the individual vessel sizes used for each step

24
of the seed train were assumed to scale accordingly. At 45% of the basis ethanol production rate,
the seed train utilized in the current design was assumed to initiate in a 17-gal vessel (rather than
40 gal) as the first step, eventually scaling up to 86,000 gal (rather than 200,000 gal) in the final
step. Table 4 summarizes the seed train design specifications.

Table 4. Seed Train Specifications

Inoculum level 10 vol % of production vessel size
Batch time 24 h
Vessel turnaround time 12 h
Number of trains 2
Number of vessel stages 5
Maximum vessel volume 86,000 gal (326 m3)
Diammonium phosphate (DAP) loading 0.67 g/L broth (whole slurry)

Table 5 gives the reactions and conversions used to describe the microorganism growth and
sugar metabolism in the seed vessels. In contrast to Zymomonas mobilis, a bacterial organism
frequently utilized for fermentative conversion of cellulosic sugars to fuels [43-45], S. cerevisiae
typically consumes a larger fractional amount of sugar diverted toward biomass cell growth [46].
Thus, while the 2011 ethanol design report assumed 4% conversion of glucose to Z. mobilis in
the seed train, translating to 0.4% diversion of glucose to cell mass in this step (given the split of
10% of total sugars sent to the seed train), the present design increases this fraction to 40%, i.e.,
4% diversion of total sugar to S. cerevisiae biomass. This is likely a conservative assumption for
such a fractionally higher amount of sugar diversion to cell growth when still fundamentally
based on anaerobic fermentation. NREL has also investigated fermentation of algal sugars using
Z. mobilis, and this may also be a viable candidate, although the use of S. cerevisiae is preferred
as this organism also readily converts the mannose sugars hydrolyzed from the algal biomass
while Z. mobilis does not. Lipids are assumed to be preserved across the fermentation stage, and
available for subsequent extraction; in fact, additional lipids may be produced during yeast (S.
cerevisiae) propagation and increase downstream extraction yields, but this has not been
experimentally characterized and is not considered here. Unlike previous models based on
lignocellulosic hydrolysate fermentation, the present analysis assumes sufficient nitrogen for
organism propagation is present in the algal biomass hydrolysate (primarily protein) constituents,
and does not require an external supply of nitrogen from corn steep liquor (CSL) or other
sources; this is consistent with early experimental observations to date. Although the same also
likely holds for phosphorus, the model conservatively assumes DAP is still required for organism
growth, delivered to the seed train at a concentration of 0.67 g/L and to the fermentation stage at
0.33 g/L. This small DAP rate is ultimately reclaimed as additional DAP from the anaerobic
digester and a fraction is available for recycle to the ponds.

Table 5. Seed Train Reactions and Assumed Conversions

% Converted to
Reaction Reactant
Product
Glucose 2 Ethanol + 2 CO2 Glucose 55%
Glucose + 0.3704 Protein + 0.018 DAP 6 S. cerevisiae + 2.4 H2O a Glucose 40%
a
For element balance closure purposes, S. cerevisiae was modeled as Z. mobilis. Sufficient nitrogen for cell
propagation is assumed to be present in the algal biomass (protein) constituents, but the model assumes
supplementary phosphorus is still required for biomass growth, delivered as DAP.

25
Sugar may be lost to side products by contaminating microorganisms in addition to being
converted to cell mass and ethanol product. Sugar contamination losses were set consistent with
the 2011 report basis, namely modeling a diversion of 3% sugar (glucose) losses to lactic acid
upstream of the fermentors.

The main fermentation step assumes the same vessel design as the 2011 ethanol design report,
namely parallel stirred-tank anaerobic batch reactors sized at 1-MM gal capacity each (950,000
gal working volume). The fermentation batch time is modeled as 1.5 days (36 hours). The
fermentation temperature of 37C is maintained using chilled water from the coolant loop (Area
700). Agitation is also set consistently with the previous ethanol design basis at a basis of 0.03
hp/1,000 gal (6 W/m3), a fairly minimal power demand attributed to merely maintaining a
reasonable suspension for anaerobic fermentation. Four 1-MM gal fermentation vessels are
required in the peak (summer/spring) design case, and operate at fractionally lower working
volume below the 95% design basis for lower seasonal throughputs. Table 6 provides a summary
of fermentor design parameters utilized in this design.

Table 6. Guiding Bioreactor Design Basis Assumptions

Temperature 37C (99F)
Initial fermentation solids level 20% total solids (7% soluble, 13% insoluble)
Reactor size 1 MM gal (3.8 MM L) operating at 95% working volume
under peak (summer/spring) seasonal throughputs
Residence time 1.5 days (36 h)
Inoculum level 10 vol %
Diammonium phosphate (DAP) level 0.33 g/L broth (whole slurry)

The model targets a fermentation ethanol yield at 95% of theoretical, i.e., a metabolic yield of
0.48 g/g glucose. This represents an improvement over 81%-83% yield experimentally
demonstrated to date for algal hydrolysate sugars utilizing upstream dilute acid pretreatment
[13], which may be achieved through continued process integration research (i.e. better tailoring
upstream operations towards downstream conversion needs, tracking and mitigating fermentation
inhibitors), optimizing fermentation conditions including further quantification of optimal
nutrient C:N:P ratios for maximizing ethanol production, or evaluating alternative organisms. As
noted previously, all monomeric fermentable sugars are modeled in this analysis as glucose,
representing both glucose and mannose as measured experimentally. Given that these two sugars
have the same molecular weight and behave similarly during fermentation using S. cerevisiae,
with no discernable lag between consumption of one sugar and another (data not published), this
simplifying assumption is appropriate. While a marginal amount of HMF is assumed to be
produced as a byproduct from the upstream pretreatment step, no discernable inhibition from
HMF during fermentation has been observed to date based on measured HMF levels of 0.1-0.2
g/L and no furfural (another fermentation inhibitor) has been measured in the hydrolysate
material [13]. It is recognized that such inhibition, as well as overall fermentation performance
(yield and rate) in general, is a function of feed and product concentrations entering and exiting
fermentation, and the sugar concentration in the hydrolysate is fairly dilute in the base case
model at 7.4% (69 g/L) sugars, translating to an ethanol concentration in the fermentor broth of
30 g/L. If the hydrolysate sugar concentration were to be increased, for example, by adjusting the
pretreatment flash specification and/or adding a concentration step, levels of inhibitory

26
components could also become more problematic. However, if such concentrations could be
achieved with no negative impact on fermentation performance, this would likely lead to a more
optimal system, primarily by reducing the energy-intensive distillation requirement downstream.
Table 7 summarizes the reactions and conversions assumed in fermentation.

Table 7. Fermentation Reactions and Assumed Conversions

Reaction Reactant % Converted
to Product
Glucose 2 Ethanol + 2 CO2 Glucose 95.0%
Glucose + 0.3704 Protein + 0.018 DAP 6 S. cerevisiae + 2.4 H2O a Glucose 2.0%
a
For element balance closure purposes, S. cerevisiae was modeled as Z. mobilis. Sufficient nitrogen for
cell propagation is assumed to be present in the algal biomass (protein) constituents, but the model
assumes supplementary phosphorus is still required for biomass growth, delivered as DAP.

The fermentation broth or beer has an ethanol concentration of 3.4 wt% (30 g/L) and is
collected in the beer well. The beer well is designed for a residence time of 4 hours to provide
surge capacity between fermentation and distillation. The material is then sent to the beer
column, a distillation column with 32 stages operating at 48% efficiency that removes CO2 and a
small amount of ethanol into the vapor overhead, recovers roughly 98% of ethanol into the vapor
side-draw (sent to the rectification column), and recovers roughly 90% of the water plus all
residual algal components and other solids into the bottoms (stillage) stream. The feed to the beer
column is preheated in an economizer heat exchanger by cross-exchange with the bottoms
stream fed to the lipid extraction section. The ethanol side-draw from the beer column is routed
to the rectification column, a second distillation column with 45 stages operating at 76%
efficiency. The rectification column further purifies the ethanol to near-azeotropic concentrations
(92.5 wt%) in the overhead, leaving 99.9% purity water in the bottoms stream. The bottoms
stream is cooled to 30C by cross-exchange with the high-pressure pretreatment steam
economizer, then is combined into the water recycle pool. Further details on both distillation
columns may be found in NRELs 2011 ethanol design report. Both columns are reboiled using
low-pressure (2.7 atm) steam generated by recovering heat from the AD turbine flue gas. Also
consistent with the original design specifications, the beer column is condensed to 60C using
cooling water, while the rectification column is condensed to 91C using air cooling.

The CO2 vent streams from both the fermentation and seed train operations contain a small
amount of volatilized ethanol, and are routed along with the distillation beer column overhead to
a water scrubber. The scrubber is specified to recover 99% of the volatilized ethanol in the vapor
feed stream by varying the scrubbing water flow rate. The scrubber overhead product contains a
high concentration (over 97 wt%) of CO2, and makes an ideal candidate for recycling back to
upstream cultivation ponds as it contains very low levels of inert components which otherwise
represent a parasitic power draw on transporting the gas back to the ponds. Thus, this stream is
destined for recycle to the cultivation ponds, and the CO2 present in the vent stream is added to
the CO2 in the gas turbine flue gas for recycle coproduct credits (discussed further below). The
liquid product (water plus recovered ethanol) is routed to the beer well along with the fermentor
product broth.

27
The overhead vapor from the rectification column is fed to a molecular sieve adsorption unit,
which is a system of columns packed with beds of adsorbent. Water is selectively adsorbed to the
beds as the vapor flows through. The molecular sieve system is a packaged unit (further details
discussed in the 2011 design report) and ultimately removes 95% of the water and a small
amount of ethanol, leaving 99.5% pure ethanol vapor, which is condensed by cross-exchange
with the regeneration condensate, then cooled by cooling water and pumped to storage. The
regeneration condensate is recycled to the top of the rectification column. The feed entering the
molecular sieve unit is slightly heated using additional low-pressure steam.

3.2.3 Cost Estimation

All cost estimates for the pertinent equipment included in the fermentation and distillation
section were scaled from original vendor-supplied costs presented in the 2011 ethanol design
case [3]. For fermentation, this includes (1) the seed train equipment, scaled to lower seed reactor
sizes (noted above) using a 0.6 scaling factor per the methodology discussed in Section 4.2, (2)
fermentation vessels, maintaining a consistent 1-MM gal fermentor size as per the original
vendor-supplied design, and (3) accompanying agitation/cooling equipment. The majority of the
equipment in the fermentation utilizes 304SS as the material of construction. Aside from this
explicitly costed equipment, other minor equipment in the fermentation section was costed by
applying a balance of plant factor of 8.6% relative to the primary fermentation equipment costs,
which account for minor pumps, heat exchangers (primarily the cooling water-supplied
hydrolysate feed cooler), and intermediate tankage, based on a similar factor from NRELs 2011
ethanol design report [3]. For distillation/purification, the earlier cost estimates include costs for
the beer and rectification columns, as well as accompanying reboilers and condensers, the beer
column feed/effluent exchanger, the vent scrubber, and the molecular sieve package. The
distillation section utilizes primarily 316SS as the material of construction.

3.3 Area 300: Lipid Extraction and Solvent Recovery

3.3.1 Overview
This is a key process area targeting the extraction of the algal lipid fraction for subsequent
cleanup and upgrading to RDB, as the primary contribution to overall GGE fuel yield. In this
section, the stillage product from the beer column is routed to a liquid-liquid extraction system
which uses hexane solvent to extract the lipids at high solvent-loading in a multi-stage counter-
current extraction column. The extracted light oil phase contains solvent, lipids (both fatty acid
lipids and polar lipid impurities), and a small amount of water, and is routed to a stripping
column to recover the solvent, leaving a high purity (~99.7% total lipids) lipid stream that is sent
to subsequent lipid cleanup. The aqueous product is sent to anaerobic digestion. Figure 7
presents a simplified flow diagram of the lipid extraction section.

An important distinction of this process pathway is a demonstrated improvement in extractability

of the lipid material when placing this step downstream of dilute acid pretreatment, relative to
the reverse order of extraction followed by pretreatment. Specifically, analysis by Laurens et al.
presents a compelling contrast for FAME-lipid extractability depending on the sequence of
process operations, whereby FAME extractability (FAME measured in extract relative to FAME
in initial biomass) ranged between 6%-8% when the biomass was sent directly to solvent
extraction first, but improved by an order of magnitude to 77%-93% when the biomass was first
processed through upstream pretreatment [13]. While based on a limited amount of experimental

28
data, this is a very encouraging result, and validates NRELs previous assumptions in earlier
harmonization modeling analyses stipulating a high extraction efficiency [8], while also negating
the need for the costly and energy-intensive cell disruption homogenization step assumed in that
prior work to be required as part of the extraction process (based in turn on much earlier work
from the Aquatic Species Program [47]). It should also be reiterated that this extraction process
is done on a wet solids basis, without the need for further evaporative drying, thus dispelling the
notion that lipid extraction must always be done on a dry-solids basis.
Hexane

Hexane +
Algal Oil
Solvent
Recovery
Hexane makeup
Column

Pretreatment
Stillage from flash vapor
beer column

HP
Extraction Steam
Column
Oil product to purification
and hydrotreating
Spent biomass to
BFW
anaerobic digestion
Boiler Feed Water
(BFW) Economizer #2

Figure 7. Simplified flow diagram of the lipid extraction and solvent recovery process

3.3.2 Design Basis

The beer column stillage contains roughly 84% water, 2% impurity/byproduct components
introduced from upstream operations (primarily lactic acid from assumed contamination losses in
fermentation, yeast biomass from ethanol production, plus HMF and ammonium sulfate formed
as a byproduct and neutralization product from pretreatment respectively), and 14% residual
algal components. This stream is cooled from 125C to 50C by cross-exchange with the beer
column preheater, and then fed to the top of the liquid-liquid extraction column. Harris Group
obtained a vendor design and accompanying budgetary estimate for this critical unit operation,
specified as a reciprocating counter-current Karr column, which was quoted by the vendor to be
capable of achieving greater than 95% oil extraction utilizing hexane solvent. Additionally, the
column designed here is described to be well-suited to handle emulsions given the columns
agitation action (the column is agitated rather than static), which is another important design
feature as this process presents a risk of forming stable emulsions between the aqueous and lipid
phases, depending on specific conditions and chemistry. However, the column is also limited to a
fairly low capacity (the summer/spring design capacity in fact requires 16 parallel columns) and
is also a substantial capital expense. Design and cost details for this piece of equipment are
presented in the next section.

Hexane solvent is fed to the bottom of the extraction column at a 5:1 weight ratio (hexane versus
total algal solids, excluding other solids components introduced from upstream processing). This
represents a very large amount of solvent circulated through the column, but was deemed

29
reasonable based on extrapolation from bench-scale extractions and to ensure sufficient solvent
contact is achieved in the column. This did not appear to be an unreasonable assumption from
the viewpoint of Harris Group or the extraction equipment vendor either, but will require further
experimental demonstration, both to confirm efficacy at this solvent ratio as well as subsequently
high solvent recovery and reuse. Hexane was selected for consistency with the experimental
work that also used this solvent system with good results [13] and because it is an ideal choice
both from a processing standpoint (hexane is easily recovered from lipids downstream with low
miscibility) and a selectivity standpoint (hexane is a non-polar solvent and ideally targets
extraction of neutral fatty acid lipids over polar lipids which must be removed later anyway). The
model targets 95% extraction efficiency/recovery of the fatty acid lipid fraction and 33%
extraction of polar lipid components, with a lower recovery of the polar impurity fraction
expected given the use of a non-polar solvent such as hexane. This is still anticipated to err on
the conservative side, as phospholipids and other polar components have been measured on part-
per-million (ppm) levels in the hexane extract for mid- and high-lipid biomass samples (data not
published), substantially less than a 33% yield into the extract phase. Translating to NRELs
standard analytical method definitions [27], this equates to 95% extraction of FAME-equivalent
lipids as defined above. Additional non-FAME (non-fatty acid) lipid components such as
hydrocarbons and sterols have also been measured in the biomass, which would primarily
partition into the hexane extract phase and ultimately get converted to additional fuel products
through hydrotreating. As this fraction was not characterized in the dataset utilized for this
analysis (Table 1), it was conservatively not considered in the model. However, this adds a small
margin of safety to the 95% extraction target stipulated here, as the net total lipid yield in the
extract would be equivalent to a marginally lower extraction of, for example, 90% FAME lipids
combined with 90% of non-FAME (non-fatty acid but still extractable and convertible) lipids if
this fraction were present at 2.5% of total biomass. Additionally, 2% water carryover into the
solvent/oil phase is assumed.

A note is also warranted here regarding the choice for process configuration. While clear benefits
around lipid extractability are enabled by placing extraction after pretreatment (discussed above),
several options exist regarding stream routing downstream of the pretreatment fractionation
operation. Earlier cursory TEA modeling for the system [13] followed the same configuration as
utilized experimentally, namely including a solid-liquid centrifuge separation step immediately
following pretreatment neutralization, which split the material into a soluble liquid (sugar)
stream sent to fermentation and a high-solids (30%-35% insoluble solids) stream sent to
extraction in parallel. While sensible for experimental validation of the technology potential, it is
not an optimal configuration in the context of process integration as it leads to considerable
product yield losses as well as operability issues at commercial scale. Without the use of a
substantially more costly solid-liquid separation system, including a washing step, a centrifuge
separation would lead to losses of 25% or more of the soluble sugars carried over into the solids
phase given a centrifugation limit of 30%-35% insoluble solids in this phase. These sugars would
then no longer be available for fermentation, but instead would ultimately go to anaerobic
digestion. Furthermore, the specified extraction column design is limited to a maximum 20%
insoluble solids content in the feed stream, which would require re-dilution if separated solids at
>30% concentration were fed to extraction from an upstream separation step. In the present
design, the whole slurry is first taken through fermentation to maximize sugar retention and
ethanol yield, then the whole stillage is subsequently sent to lipid extraction at 14% insoluble
solids and does not require any further dilution for operability purposes in the extraction column.

30
The process for whole-slurry fermentation also adds additional capital expenses by way of higher
total fermentation tank volume relative to fermentation of only the liquor phase, but these costs
are relatively small compared to other process operations and are outweighed by the benefit of
higher ethanol yield using the whole-slurry approach. Additionally, a small amount of additional
lipids may be present in the yeast (S. cerevisiae) biomass produced during fermentation which
may then be available for additional lipid yields during extraction; however, this has not been
characterized and is not included here.

Following the extraction operation, the oil/solvent product is sent to a stripping distillation
column to recover the solvent. This column was also specified by Harris Group and designed as
a packed column with an overall height of 20 ft. A total of three columns operate in parallel,
primarily to accommodate high turndown ratios up to the original 5:1 basis if required (e.g., the
non-summer-storage option). In the base case scenario, including diversion of excess summer
capacity and associated reduction in turndown ratio to 1.2:1, a single distillation column would
likely be required, which would likely bring economy-of-scale advantages. However, as the cost
for the stripping columns alone does not represent a large fraction of overall system capital cost,
the savings would likely be marginal. The feed to the distillation column is preheated from 52 to
66C by cross-exchange with the pretreatment flash vapor condensate exchanger described
previously. The reboiler is driven by high-pressure steam (665 psig, 46.3 atm) generated from
AD turbine flue gas heat recovery.

The extraction column bottoms product or raffinate stream contains water, residual biomass
solids, and a small amount of lipid and solvent losses equating to 67%, 5%, and 0.5% of polar
lipids, fatty acid lipids, and solvent respectively, relative to the column feed rates. This stream is
routed to anaerobic digestion. Fresh hexane solvent is mixed with the solvent recycle to make up
for losses to the raffinate and recovered lipid streams. All key process assumptions for the
extraction area, as well as comparisons to pertinent experimentally demonstrated values to date,
are summarized in Table 8.
Table 8. Area 300 Guiding Design Basis Assumptions and Comparisons to Values Experimentally
Demonstrated to Date [13]
Parameter Baseline Experimental Target
Solvent loading (solvent/biomass ratio, dry wt basis) 4.4-5.8 5.0
3-step extraction (extract, Counter-current
Extraction method dry/recover solvent, agitated LLE
re-extract) column
Insoluble solids to extraction 20% 14%
FAME lipid extraction yield a 77-93% b 95%
Polar lipid impurity extraction yield ND 33%
Solvent carryover losses (% makeup required) NA 0.54%
a
FAME lipids as defined in [27]
b
Values shown are for Scenedesmus strain; Chlorella strain extraction efficiencies were lower

3.3.3 Cost Estimation

Cost estimates for the primary unit operations in this section were provided by Harris Group
based on input from vendors. Namely, four Karr extraction columns were originally specified to
accommodate a total lipid feed rate of 5,894 kg/hr to the extraction system. Under the present
scenario based on the maximum summer/spring capacity design, this would translate to 16
equivalent columns operating in parallel. Each column is designed with a 6-ft-diameter agitated

31
zone, 9-ft-diameter expanded ends, and an overall height of 60 ft. Materials of construction are
304L for the column shell and 304SS for column internals. Each column requires 40 hp (30 KW)
of power. The resulting purchase cost was quoted as $1.98 MM per column in 2013-year dollars,
which translates to a total installed cost of $63.5MM for the extraction system, or roughly 30%
of the total ISBL (inside battery limits) equipment installed costs, thus representing a significant
expense for the overall facility; however, the vendor and Harris Group expressed confidence that
this particular column design, while costly, is well-suited for this particular application with
>95% expected performance for extraction efficiency (including the ability to handle emulsions
if they were to form, as noted previously).

Harris Group also supplied cost estimates for the solvent stripping recovery columns and
associated reboilers (designed with three each, as noted above). The costs were quoted as
$714,000 and $150,000 for the distillation column and reboiler respectively, based on purchase
costs per unit in 2009-dollars. The solvent recovery column is a packed tower with dimensions of
10-ft-diameter x 20-ft-height, with all equipment utilizing 304SS metallurgy. The extraction feed
heat exchanger cost is already included with the previously-quoted cost under the distillation
section, e.g., the beer column feed-effluent (stillage) heat exchanger, and the solvent recovery
column preheater exchanger cost was already discussed above in Section 3.1.3 (cross-exchange
with pretreatment flash condensate).

3.4 Area 400: Product Purification and Upgrading

3.4.1 Overview
Area 400 first purifies the recovered lipid product to remove impurities which would be
problematic for the subsequent catalytic upgrading step. Lipid cleanup consists of a series of
degumming, demetallization, and bleaching steps to remove gums, metals, and other impurities
with the use of phosphoric acid, wash water, silica, and clay. These operations are similar to
those utilized in the biodiesel industry, and although this system incurs non-trivial costs, it was
deemed necessary to include, at a minimum until further experimental data becomes available on
the hydroprocessing operation, to validate or refute its use in the context of integration with
upstream operations (e.g., introduction of salts through pretreatment and neutralization and
additional changes incurred through fermentation). The cost impact attributed to removing the
lipid purification steps is considered below in the Sensitivity Analysis section. The resulting
purified neutral lipid material is then sent to upgrading in a hydrotreater for purposes of
deoxygenation and saturation, primarily yielding a diesel-range RDB product with a small yield
of naphtha coproduct, as well as off-gas which is combusted in the biogas turbine. The process
schematic for the Area 400 operations is depicted in Figure 8.

32
 Makeup H2
Light gases to CHP
Purification Three-Step Lipid
chemicals Purification

Oil product
from extraction Purified Oil Product Product
Hydrotreating Storage

Waste to
Anaerobic Digestion Water to recycle pool

Figure 8. Simplified flow diagram of product purification and upgrading

3.4.2 Design Basis

As modeled, the bottoms product exiting the solvent recovery stripping column contains 98%
neutral (fatty acid) lipid, 1.7% polar lipid impurity components, 0.3% hexane, and a trace
amount of water. While phospholipids are of primary concern within the polar lipid fraction for
their propensity to form gums and deactivate catalysts [48, 49], in reality the extracted polar lipid
fraction has also been characterized to contain phytol (derived from chlorophyll) and nitrogen
species (equivalent to ~300-400 ppm nitrogen), and the extracted lipids from Scenedesmus were
also measured to contain >35 ppm metals, primarily sodium, iron, copper, phosphorus, and
calcium, attributed to both upstream salt formation and trace nutrients in the algal biomass (data
not published). In light of this information, Harris Group and the vendors they contacted thought
it prudent to include a lipid cleanup step to remove these impurities for downstream catalyst
protection in the hydrotreater using commercially available technology common to similar
operations used in biodiesel processing.

A general block diagram for the purification system is shown in Figure 9. As shown, gums
(including phospholipids) are removed in a degumming step with the addition of phosphoric acid
followed by centrifugation with wash water. Subsequently, demetallization is performed by
adding silica, forming a slurry, and then filtering out the spent silica. This step is followed by
bleaching with clay to remove additional metals and impurities, which again involves forming a
slurry with clay followed by filtration. Cakes formed from filtering the silica and clay from the
oil, as well as the material removed from degumming, are routed to anaerobic digestion, with the
small phosphoric acid content of this stream assumed to be available as additional phosphorus
for subsequent recycle to the cultivation ponds. The vendors Harris Group spoke to noted that the
phospholipid content in the product oil is expected to be less than 10 ppm, possibly as low as 2 to
3 ppm, based on the analytical characterization of the extracted oil furnished to the vendor.
Additionally, they expect metals content to be less than 5 ppm for potassium (K), calcium (Ca),
magnesium (Mg), and sodium (Na). Testing is required to specifically determine the expected
performance of these systems. It should be noted that when treating vegetable oils, a
neutralization step is performed during degumming wherein lye is added to remove free fatty
acids. Here, free fatty acids are valuable feedstocks for hydrotreating and therefore this step is
not included in the concept presented here. The process takes place at 110C (230F), with the
incoming lipid material first cooled to this temperature after exiting the solvent recovery column
by cross-exchange with a second high-pressure pretreatment steam economizer.

33
 Figure 9. Block diagram schematic for lipid purification system

A summary of the operating conditions and chemical requirements for the lipid cleanup step is
presented in Table 9. The resulting purified lipid product is modeled as essentially 100% neutral
lipids (modeled here as the triglyceride triolein, although diglycerides, monoglycerides, and free
fatty acids are also present), cooled down to 93C after passing through the various cleanup
operations including mixing and removing wash water.

Table 9. Operating Conditions and Material Requirements for Lipid Purification

Degumming Step
Total lipid feed rate design basis 22,576 kg/hr (summer/spring design case)
Feed temperature 110C (230F)
Phosphoric acid dosing (wt% of feed) 0.19%
Wash water demand (wt% of feed) 10%
Demetallization Step
Silica dosing (wt% of feed) 0.1%
Bleaching Step
Clay dosing (wt% of feed) 0.2%
Purified lipid product rate design basis 22,124 kg/hr (summer/spring design case)
Total gums/waste product to AD 12.5% of feed (polar lipids + added materials)

Following lipid cleanup, the material is upgraded in a hydrotreater to refine the oxygenated
intermediate material into saturated paraffin components suitable for blending as a diesel
blendstock. The design and cost estimates applied for this section are largely consistent with the
hydrotreating facility described in NRELs 2013 design report focused on hydrotreating free
fatty acids [6], but will be repeated again here. In summary, Harris Group consulted with a
number of vendors, including catalyst suppliers and process licensors, to obtain budgetary
estimates for a hydrotreating facility that may be expected to process the feed as modeled here
into deoxygenated diesel-range paraffins. A number of cost estimates were provided for such a
facility, but the vendors declined to provide specifics without confidentiality restrictions
regarding associated process and design conditions (e.g., reactor sizing dictated by space
velocities, operating temperatures and pressures, hydrogen partial pressure, etc). Thus, to
estimate reasonable process conditions for modeling purposes, a combination of literature and
Harris Group estimates were applied. Namely, a report by Marker et al. describing experimental
conditions and yields for hydroprocessing vegetable oils was leveraged which remains one of the
most thorough reports on the subject to date, and was jointly authored by UOP, NREL, PNNL,
and Michigan Technological University [50]. This report described relatively mild hydrotreating
temperature and pressure conditions at 350C (660F) and 35 atm (500 psig), but a high

34
hydrogen feed ratio at 6,000 standard cubic feet/barrel of feed (this translates to a hydrogen
partial pressure of approximately 480 psig at reactor inlet). As confirmed by Harris Group, high
hydrogen feed is necessary in part to provide a quench to control the large exotherm associated
with deoxygenation. A reasonable liquid hourly space velocity (LHSV) associated with this
process would likely be in the range of 0.9-1.4 hr-1 [50], which was confirmed by an estimated
value of 1.2 hr-1 from Harris Group.

A schematic diagram of the cleanup and hydrotreating process operations is shown in Figure 10.
This diagram includes an assumption of a two-stage hydrotreater, with the first stage including a
guard bed for metals, as well as a possible mechanism for handling the high acid (TAN) content
as measured for extracted lipids from Scenedesmus. Additionally, a cost estimate was also
included for a pressure swing adsorption (PSA) unit in the hydrotreater gas recycle loop,
primarily to remove CO2 formed by decarboxylation reactions in the hydrotreater, as well as
propane produced from the glycerol backbone of the triglyceride lipids (discussed below). The
design and cost estimates for the PSA unit were also provided by vendor input. In the Aspen
model, the gas recycle feed stream to the PSA consists primarily of H2, CO2, CO, and propane
(after cooling the reactor effluent and knocking out the majority of the produced water, which is
routed to the water recycle pool). Given this gas composition, the PSA vendor provided a cost
estimate for a patented PSA unit which utilizes a zeolite-based sorbent to purify the recycled
hydrogen to roughly 97 wt%, with the product containing approximately 0.1% CO2, 1% CO, and
2% propane, relative to the recycle feed stream to the PSA composed of 32% H2, 47% CO2, 2%
CO, and 19% propane (mass basis), attributed to the hydrotreating product yields discussed
below. Given this composition, particularly the presence of propane, the PSA system incurs a
greater hydrogen loss than the NREL 2013 cellulosic design report (91% H2 recovery versus
>95% in the cellulosic case), requires a much greater power demand (2.3 MW in the peak
summer/spring design case versus 35 KW in the cellulosic case), and is significantly more costly
(discussed in the next section). This difference is because in the earlier cellulosic pathway report,
the targeted product was exclusively free fatty acids which do not contain the three-carbon
glycerol backbone present in lipid triglycerides, which is rejected as propane from the
hydrotreater and makes for a more complicated hydrogen purification operation. The overall
economic sensitivity to the assumed PSA hydrogen recovery is evaluated in Section 5.1. The
PSA tailgas, containing approximately 4% H2, 26% propane, 67% CO2, and 3% CO (mass
basis), is split to provide the necessary heat to the hydrotreater furnace (thus avoiding external
natural gas to drive this operation), with the remainder sent to the biogas turbine (A500).

35
 MAKEUP H2
COMPRESSOR
Makeup H2 from RECYCLE
off-site SMR COMPRESSOR
H2

PSA Tailgas
Natural gas Off gases to CHP
Natural gas for startup
and as needed
PREHEATING
FURNACE
HYDROTREATING
REACTOR
Phosphoric acid, HYDROTREATING
clay, silica, wash GUARD BED/TAN
water REDUCTION PSA UNIT
Product gas

Algal oil from PARTIAL

CONDENSER
extraction and LIPID Offgas
solvent recovery PURIFICATION
SEPARATION Naphtha

Waste to AD CW

Liquid
product PRODUCT
DISTILLATION
COLUMN

Renewable Diesel
Blendsock (RDB)
Water Water to water
recycle pool
Dashed box represents hydrotreating ISBL system

Figure 10. Schematic diagram of hydrotreating operations included in vendor design estimate

Oxygen removal from the lipid feed may occur by way of hydrodeoxygenation (HDO) or
decarboxylation (DCO) reactions in the hydrotreater. The relative degree of one reaction
pathway over another may be tailored by catalyst and reaction conditions, with a tradeoff
exhibited between higher hydrogen consumption to reject oxygen as water via HDO (this carries
important implications particularly on the LCA/sustainability metrics of the system when
sourcing hydrogen from conventional fossil sources) and decreased carbon yields to reject
oxygen as CO2 via DCO (with decreased product yield translating directly to cost impacts). For
this work, hydrotreater product yields were based on the previously-cited report by Marker et al.
[50], which presents an optimized experimental case favoring DCO over HDO by a factor of
over 2:1 (72% of the reaction occurs via DCO and 28% via HDO), translating to high CO2 yields
relative to H2O and lower fuel yields than would be possible with higher HDO ratios. This is a
similar fractional split favoring the DCO pathway as presented in other sources [51-53]. The
associated hydrogen consumption was set at 1.7 wt% of the oil feed rate, also consistent with
other data favoring DCO [51, 52]; however, additional hydrogen is lost from the system via the
PSA as noted above, with all hydrogen consumption/losses met with fresh makeup hydrogen.
Sustainability implications for sourcing hydrogen from off-site natural gas steam methane
reforming are presented in Section 5.3. The experimental literature data from which the modeled
yields were derived indicated no heavier-than-diesel residual material (vacuum gas oil or VGO)
was present in the hydrotreater product stream [50], thus avoiding the need for a second residual
hydrocracking unit in this design. The resulting product yields are summarized in Table 10.

36
 Table 10. Product Yields from Hydrotreating (derived from [50])
Product Yield, wt% of oil feed
Total reactor product yield 101.7%
H2 consumption -1.7%
Reactor product composition:
Fuel-range hydrocarbons 80.0%
Diesel-range 78.5%
Naphtha-range 1.5%
CO2 12.1%
CO 0.6%
H2O 1.7%
Off-gas (propane) 5.6%
Total 100.0%

The product yields summarized above were slightly adjusted from earlier NREL models for algal
lipid upgrading [8, 54], to more precisely match the original literature data source [50] and to
ensure 100% element balance closure across this step (based on the representative triolein
component modeled for the neutral lipid fraction). Given the rigor of the original literature
dataset from which this model is derived, as well as highly favoring DCO mechanisms for
oxygen rejection, this is a reasonable, if not conservative, set of assumptions around
hydrotreating process yields and compositions. Nonetheless, it will be important to validate or
refine these assumptions based on experimental data specific to hydroprocessing the extracted
algal oil from the HLSD strain/harvest basis assumed here. NREL has recently begun
experimental efforts in this regard, with the primary objective of better quantifying
hydroprocessing conditions, yields, and product characteristics associated with algal lipid
upgrading, but data are not yet available at the time of this writing to leverage in the design
report modeling effort. The assumptions employed here will be revisited as such data become
available.

The diesel-range product from the hydrotreating operation is composed primarily of straight-
chain paraffins with high cetane value, making it potentially valuable as a diesel-range
blendstock. It is recognized that to produce a stand-alone final fuel product meeting ASTM
diesel specifications, the straight-chain paraffinic product from hydrotreating would require
further isomerization to address likely cloud point issues [14]; however, the high cetane and low
sulfur content expected to be exhibited by the RDB product is expected to make such a product
valuable for diesel blending purposes, and the removal of oxygen during hydrotreating suffices
to meet DOE requirements consistent across other technology pathways for a finished product
[9]. NREL experimental efforts will also work to better understand product quality and tradeoffs
in further processing requirements brought about by adding a hydroisomerization step after
deoxygenation.

The product stream exiting the hydrotreating reactor is split into gas, liquid, and aqueous phases,
in which the gas phase is sent to the PSA for subsequent hydrogen recycle (as described above)
and the aqueous phase (primarily water) is sent to the water recycle pool. The liquid product is
fractionated in a distillation column, whose design and cost is included with the rest of the
vendor-supplied hydrotreating facility quotation. The distillation column is steam-stripped using
low-pressure steam produced from the turbine flue gas heat recovery steam generator. The RDB
product fraction exiting the distillation column, modeled as paraffins primarily in the C13-C20
range, is diverted to an Aspen calculator block to calculate the higher heating value of the

37
product, for purposes of translating to a GGE yield (see Section 2.4.2). The naphtha product
fraction, modeled as paraffins primarily in the C6-C12 range, is produced as a small coproduct.
The column off-gas is combined with the PSA tail gas and sent to the AD biogas turbine. Table
11 summarizes the key process assumptions discussed above for the hydrotreating step.

Table 11. Summary of Hydrotreating Process Assumptions

Oil feed rate (at standard conditions)
154,040 gal/day (583 m3/day)
design capacity (summer/spring capacity)
Average reaction temperature 660F (350C)
Pressure 500 psig (35 atm)
Hydrogen partial pressure 480 psig (34 atm)/6,000 SCF/bbl
Reactor LHSV ~1.2 hr-1 (estimated) a
Reaction H2 consumption 1.7 wt% of feed
Hydrogen loss across recycle PSA 9%
Makeup hydrogen rate
5.4 MM SCFD (145,311 Nm3/day)
design capacity (summer/spring capacity)
RDB product rate
145,758 gal/day (552 m3/day)
design capacity (summer/spring capacity)
Naphtha coproduct rate
3,403 gal/day (13 m3/day)
design capacity (summer/spring capacity)
a
Reactor LHSV is not used in the present model (total hydrotreater facility cost was used based on
vendor feedback, which implicitly includes hydrotreater sizing which was not provided by vendor)

3.4.3 Cost Estimation

The bleaching, demetallization, and degumming cost estimates were furnished by vendors via
Harris Group, and translate to an installed cost of ~$7MM based on the summer/spring design
capacity. A marginal power demand of roughly 0.01 KWh/kg oil feed rate was applied, also per
vendor feedback. Variable operating costs for the purification chemicals (and all chemicals for
the facility) are discussed in Section 4.3. The original vendor quote also included the requirement
for steam to heat the oil material to at least 85-95C (185-200F) from a lower temperature, but
the present case sends the oil product directly to the cleanup stage from the solvent recovery
bottoms; thus instead is first cooled down to a slightly higher temperature (110C) by further
preheating the feed water for the pretreatment high-pressure steam generator in a second
economizer following the rectification column bottoms cooler. The cost for this heat exchanger
was scaled from earlier NREL model costs for a plate-and-frame heat exchanger [3].

Consistent with NRELs 2013 biochemical hydrocarbon design report, the vendor-supplied
hydrotreating facility cost is an inside battery limits (ISBL) cost estimate, and includes costs
for reactors, makeup and recycle gas compressors, fired heater, separation vessels, and
distillation. Additionally, while catalyst replacement cost is not explicitly specified, one vendor
noted that catalyst would be an insignificant part of the total cost estimate for this particular
system, at a typical catalyst lifetime of two years and the given lipid feed specification. Based on
a number of vendor-supplied cost estimates, combined with Harris Groups expertise on the
subject as well as typical industry costs for petroleum hydroprocessing, an original cost estimate
of $23MM (installed cost) was utilized as consistent with NRELs 2013 design report focused on
free fatty acid hydrotreating, which translates to $30MM installed in the present analysis based
on the design summer/spring capacity. The uncertainty range associated with this cost estimate
was quoted as -50%/+100%, thus is most appropriately viewed as an order-of-magnitude cost for
processing this type of lipid/fatty acid feedstock. The overall cost impact associated with this
given range of uncertainty in the hydrotreating cost estimate is evaluated in Section 5.1. Even

38
with such a relatively high margin of uncertainty in the hydrotreating cost estimate, we feel it is
more prudent to leverage this information than attempting a cost estimate on our own (e.g. using
cost estimation software) as the vendor estimates bring valuable insight as to important
considerations such as reactor design, number of stages, and operating conditions. Given the
relatively mild hydrotreating conditions listed in Table 11, which are in line with a relatively
high-quality material as would be expected to be present from purified fatty acids/lipids (e.g.
<0.05% nitrogen, <0.01% sulfur, ~10% oxygen based on internal analytical characterizations),
both Harris Group and NREL thought it appropriate to utilize such a baseline cost estimate here.

The PSA unit was provided as a separate cost item, as this is not a typical unit operation in
conventional hydrotreating systems, but is required here primarily to remove CO2 and CO ( via
the DCO reaction which produces these products) as well as propane (via rejection of the 3-
carbon glycerol backbone from triglycerides) formed during hydrotreating. The cost estimate for
this operation was provided by a PSA vendor. As noted above, the design basis for this operation
was based on a patented zeolite sorbent system, allowing for reasonable hydrogen recoveries and
high rejection of the other components, albeit at considerable expense given the difficulty of
removing propane. Indeed, the PSA system installed cost was estimated at nearly $12MM, or
roughly 40% of the cost of the rest of the hydrotreating facility. This is a substantial cost increase
for this unit relative to the 2013 design report which excluded propane formation. Harris Group
also evaluated the cost for an amine system as an alternative CO2 removal option, and found that
this alternative may be more cost-effective relative to the PSA option, but this also is likely to
exhibit a higher heat demand for amine regeneration, as well as require the addition of a purge
stream to mitigate propane buildup in the recycle loop. As more definition becomes available for
the intermediate lipid product characterization and hydrotreating specifics, the options for
hydrogen purification and/or purge in the recycle gas line may be revisited. The power demand
for the PSA unit in this case was estimated as 0.10 KWh/kg oil feed to the hydrotreating section.

3.5 Area 500: Anaerobic Digestion/CHP

3.5.1 Overview
The aqueous product (raffinate) exiting the lipid extraction column is combined with the waste
stream from the lipid purification step, then cooled and routed to anaerobic digestion (AD). The
AD step is used in part to provide a degree of wastewater treatment/ sludge stabilization, but of
equal importance in the context of a fully integrated process with algae cultivation, AD is used as
a means to recover nutrients (primarily nitrogen and phosphorus, but also other minor nutrients
not considered in the model) for recycle to the cultivation step, as well as to recover residual
carbon in the form of biogas for heat and power benefits. While AD alone is not envisioned to
necessarily provide sufficient treatment of the facility wastewater for subsequent discharge or
recycle to the conversion process, microalgae cultivation itself is also an effective water
treatment mechanism [55]. Thus, similar to earlier NREL designs, including that utilized in the
previously referenced harmonization report, AD is included as a means for spent biomass
utilization and a precedent for enabling these important sustainability benefits (nutrient recycle
and energy recovery), with the effluent intended for recycle to the algal growth ponds
(maintaining cultivation outside the scope of this work). AD organisms break down the incoming
organic carbon into biogas consisting primarily of methane and CO2, as well as nitrogen and
phosphorus fixed in the residual material into forms that may be assimilated by cells upon
recycle to algal cultivation. Sulfate is also converted to H2S, which partitions into the biogas.

39
While a high fraction of N and P nutrients are assumed to be recovered into the liquid effluent
stream, the remaining fraction primarily allocates to a high-solids digestate cake.

The biogas stream is routed to a gas turbine, which combusts the material to generate electricity
and heat. Sufficient power is generated to provide a net-export coproduct credit, when
considering strictly the conversion facility boundaries here (i.e. negating upstream power
demands for cultivation and dewatering). The turbine flue gas heat is also recovered to generate
steam for meeting facility heating and live steam demands. This combined heat and power
(CHP) benefit is another key sustainability driver that has been previously shown to provide
important LCA advantages in an integrated algal processing pathway [8, 56, 57], justifying the
continued use of AD/CHP in this case as well. The flue gas exits the turbine at high temperature
and is passed through a series of three heat recovery steam generator exchangers, which generate
high-pressure steam for live injection into the pretreatment system (Area 100), high-pressure
steam for heating purposes in a closed-loop utility, and low-pressure steam for closed-loop
heating, respectively. After considering all facility steam and heating demands, overall heat
balances leave a small heat deficit which must be satisfied with supplemental natural gas co-fired
in the turbine. This aspect must be taken into consideration in work that proposes higher value
coproducts from the raffinate. Both the economic and LCA impacts from such a scheme demand
careful attention to the tradeoffs implicit in each proposed coproduct. The flow diagram for the
AD/CHP system is shown in Figure 11.

HP and LP steam
Power

Heat Recovery
Steam Generators
Supplemental Flue gas
natural gas Flue gas
to ponds
Off-gases from Biogas Turbine
hydrotreating Generator Feed water

Biogas
Waste from
purification
Centrifuge
Spent residues from AD effluent Recycle to
extraction column ponds
CW

Anaerobic Digestate cake

Digestion (N fertilizer coproduct)

Figure 11. Simplified flow diagram of the AD/CHP system

3.5.2 Design Basis

The AD system fundamentally employs the same process assumptions as documented in detail in
the 2012 algae harmonization report. This includes a typical 35C operating temperature
(achieved by cooling the AD feed stream down from 52C in a cooling water exchanger,
although thermophilic operation at 50-60C is also possible), and 20-day hydraulic retention time
(HRT) based on peak summer/spring capacity [58]. Applying the same guidance used in prior
efforts as applied in the harmonization report, this translates to a volatile solids loading factor of
roughly 3.3 g/L-day (4.4 g/L-day total solids basis) using standard industrial AD design criteria

40
[58], associated with a total feed solids content of approximately 9%. Based on peak
summer/spring capacity flowrates, the resulting total AD volume required was calculated to be
32.1 MM gal (121,500 m3), requiring two digester units each sized at 16.1 MM gal, assuming 27
MM gal maximum digester size as applied in the 2013 design report [6]. Relative to the
harmonization baseline which otherwise forms the design basis for this step, the present model
applies a higher solids loading rate (4.4 g/L-day versus 2.4 g/L-day originally) due to a
proportionately higher total solids feed concentration (9% versus 5%). Volatile carbon
destruction to biogas was maintained at 48%, consistent with the harmonization basis. While this
translated to a biogas methane yield of 0.3 L CH4/g TS in the harmonization model, it is reduced
to 0.2 L/g TS in the present analysis (roughly 0.27 L/g volatile solids), due to the introduction of
additional non-digestible solids in upstream processes (e.g., salts) as well as lower carbon
content of the digestible solids attributed to removing the majority of both lipids and
carbohydrates upstream, rather than only lipids. Consistent with the harmonization basis, the
biogas is composed of 67% CH4 and 33% CO2. Finally, AD heat and power demands were set at
0.22 KWhthermal/kg total solids and 0.085 KWh/kg total solids respectively as documented in the
harmonization report. The AD operational and yield assumptions are summarized in Table 12.

Although the original assumptions around operating conditions (HRT and solids loading factor)
and biogas yields as established in the harmonization baseline were largely derived from
literature values on AD of spent algal residues with often incomplete or otherwise non-applicable
datasets, experimental studies from NREL and Washington State University (WSU) utilizing
lipid-extracted algae confirmed these assumptions with very close agreement, namely identical
methane yield at the same HRT and roughly 15% higher total solids loading rate [59]. While this
validates the original AD assumptions for lipid-extracted algae and the present assumptions are
sensible in relation to the earlier basis (e.g., lower methane yield given less carbon available for
digestion), the present assumptions carry a higher degree of uncertainty, particularly with respect
to operability of the AD system, given a lower carbon-to-nitrogen (C/N) ratio in this case. As the
AD feed here has been extracted of both lipids and carbohydrates, the residual material is
enriched in protein and thus will have a lower C/N ratio than earlier model scenarios; currently
estimated C/N ratio as modeled is roughly 8.9 (molar basis). It is known that AD systems exhibit
operability issues as the C/N ratio decreases due to nitrogen inhibition [60], but a specific limit is
not well-quantified for algal biomass residues. NREL/WSUs recent experimental work
confirmed that no issues were observed for AD of lipid-extracted algal biomass, thus negating
this concern up to a point, but further research is needed to confirm this remains the case in the
present scenario.

The total slurry feed rate to the AD unit in the summer/spring design case is 243,500 kg/hr or
21,480 kg/hr total solids. Of the total solids, roughly 75% are digestible, with non-digestible
solids identified as ammonium sulfate salt, ash, fermentation organism (S. cerevisiae), clay and
silica materials used in lipid purification, and hexane. As noted above, the model is based on a
freshwater algal strain and subsequent processing; if a saline scenario were considered, salt
levels could translate to more challenging AD operation. Aside from methane and CO2 in the
biogas, H2S is also present as a product of sulfur breakdown. This is subsequently converted to
SO2 after combustion in the gas turbine, present in the flue gas at approximately 280 ppmv. As
the flue gas is not vented to the atmosphere, but rather is destined for recycle to the ponds
upstream (mixed with fresh flue gas from coal power plant operations which also contains SO2),
no desulfurization is included to remove sulfur after the turbine and SO2 is assumed to be

41
absorbed into the cultivation ponds. However, sulfur control may still be needed prior to the
turbine for metallurgy corrosion mitigation purposes; this is not considered further here as the
sulfur content will be highly dependent on source of cultivation CO2, i.e. purified CO2 versus
bulk flue gas, with the measured sulfur content of the biomass presented in Table 1 based on
carefully controlled growth conditions and media composition which may not reflect commercial
systems.

Beyond the biogas fraction, consistent assumptions were employed for both the AD liquid
effluent and solids digestate cake product streams as described in the harmonization report,
namely 80% N and 50% P recovered in the effluent stream (relative to digestible N and P
entering AD, excluding ammonium sulfate salt) with the remainder allocated to the digestate
cake which exits the system at 30 wt% insoluble solids after passing through a centrifuge. A
small amount of flocculant may be intermittently required to reach this solids content in the
digestate, but the cost would be marginal and this is not considered here. Of the nitrogen, 5% is
assumed subsequently lost to volatilization, translating to 76% net nitrogen recovery in the
effluent. The solids digestate material has been identified as a key contributor to fugitive
methane emissions when tracking this parameter [61], and results in a nearly negligible
coproduct credit using the assumptions described below. As an alternative, the digestate cake
fraction may be instead rerouted to the cultivation ponds along with the AD liquid effluent
fraction, where early observations have indicated it substantially biodegrades given the highly
aerobic conditions of the open ponds [62]. This option is not considered here to maintain
consistency with prior work and as it is not yet known if it would negatively impact cultivation
over a long term, but will be considered in future analysis.

The AD effluent stream is intended to be recycled back to the algal cultivation step upstream,
thereby reducing the amount of makeup N/P nutrients required for biomass growth. As such, the
stream is assigned a coproduct value after translating the elemental components to equivalent
diammonium phosphate (DAP) and ammonia, with DAP and NH3 coproduct values equal to
each components purchase cost as applied upstream. The digestate cake stream is also assigned
a coproduct value, equating to $500/tonne of bioavailable nitrogen (assuming 40% of the total
nitrogen in the digestate cake is bioavailable), also consistent with the harmonization report and
as originally described by [55]; this translates to roughly $4/tonne total solids. It is justified to
include coproduct credits here for AD digestate, nutrient recycle to cultivation, and CO2 recycle
to cultivation (discussed below), because these recycle credits were not included in the original
$430/ton biomass cost basis discussed above (documented in DOEs MYPP report, table B-4
[9]). Consequently, rather than reducing the biomass feedstock cost to account for these recycles,
the equivalent cost savings are instead counted here as coproduct credits to (marginally) reduce
the final fuel MFSP. The cost impact to final fuel MFSP attributed to removing any of these
coproduct credits is considered below in the Sensitivity Analysis section (see Section 5.1).

42
 Table 12. Anaerobic Digestion Assumptions
Metric Basis Source
Operating temperature 35 C [58]
Hydraulic retention time 20 days (design basis) [58]
AD heat demand 0.22 KWh/kg total solids (thermal) [8]
AD power demand 0.085 KWh/kg total solids [8]
Volatile solids loading rate 3.3 g/L/day Extrapolated from
(Total solids loading rate) (4.4 g/L/day) [58]
Total solids in feed 9% Per Aspen model
CH4 yield in biogas 0.2 L/g TS (48% carbon destruction) Consistent with [8]
Biogas composition 67% CH4 / 33% CO2 (vol basis) [8]
N recovery in effluent 76% net (80% - 5% volatilization loss) [8]
P recovery in effluent 50% [8]
Bioavailable N in digestate cake 40% [55]
Digestate cake insoluble solids 30% [6]

The biogas is next routed to a gas turbine generator, after mixing with hydrotreating off-gases
including the PSA tail gas and distillation overhead off-gas. A gas turbine is employed here
rather than alternatives such as an internal combustion engine, as it provides high electrical
efficiencies (at large scale as is the case here) and avoids high fugitive methane emissions
typically encountered with internal combustion engines [61]. The biogas is first compressed to
12.2 atm (165 psig) and mixed with injection air at a ratio to maintain combustion temperature
(prior to turbine expansion) near 1155C, as is typical for the specified unit [63]. In the previous
harmonization baseline model, the turbine had been specified to let down to a slightly elevated
pressure of 5 psig (relative to standard gas turbines exhausting near atmospheric pressure), in
order to overcome hydraulic losses associated with flue gas cooling for steam generation, recycle
to the cultivation ponds, and sparging through a membrane into a 1.5m sump. The present
analysis also assumes the turbine flue gas is destined for recycle to the ponds in the same
manner, but given additional steam generation demands in the present system (e.g. for live steam
injection to the pretreatment reactor as well as low-pressure steam for ethanol distillation), the
flue gas exhaust pressure was further elevated to 7 psig (1.5 atm). This difference of 2 psig
incurs a loss of 17% in total amount of power generated, but also increases the exhaust
temperature by 22C (translating to more heat available for steam generation). After accounting
for parasitic losses for biogas and air compression, the turbine electrical efficiency is roughly
20%, compared to 25% in the harmonization baseline at the lower exhaust pressure, or 33% as
typical for standard gas turbines exhausting to atmospheric pressure. Thus, while incurring a
substantial penalty in overall power generation potential, the assumptions employed here are
reasonable in order to allow for all flue gas cooling and recycle requirements in the broader
context of a fully integrated process with upstream cultivation/dewatering connections. Given
that the turbine operating conditions were tailored to allow for flue gas recycle to the cultivation
pond step (taking a significant electrical efficiency penalty), it is justified to then include a
coproduct credit for flue gas recycle, as the recycled flue gas translates to a cost savings for the
cultivation step by requiring less makeup flue gas delivered from offsite, which is more costly
than on-site recycle [8].

The flue gas recycle credit was calculated as $33.9/ton CO2, based on back-calculating the
explicit cost per ton of offsite CO2 pipeline delivery in the original harmonization benchmark
model. This is a reasonable value, relative to standard costs for flue gas carbon capture ranging

43
from $35-45/ton [64, 65]. Given the low CO2 content of the flue gas (7 wt%) as typical for gas
turbines operating with substantial excess air, the parasitic energy penalty for transporting the
flue gas with high inert levels is not ideal, and further analysis is required to better understand if
a more optimum recycle or venting scenario exists. Additionally, more detailed pond dynamics
may dictate that recycle occur intermittently, for example based on variable pond pH, which
would either require compressing and storing flue gas or venting during such intermittent down-
times (a more likely case). The high-concentration CO2 from the ethanol fermentor vent is also
added to this value to determine the total CO2 recycled to the cultivation ponds. The cost impact
to MFSP associated with removing this CO2 recycle credit is evaluated in Section 5.1.

The flue gas exhaust exits the turbine at a temperature of 638C. The exhaust is then used to
generate steam for the various facility heating and live steam requirements in a series of heat
recovery steam generators (HRSGs). The first exchanger cools the flue gas to 490C to raise
superheated steam at 268C which is injected to the dilute acid pretreatment system. The second
exchanger further cools the stream to 430C to raise high-pressure saturated steam at 260C for
closed-loop utility heating serving the solvent recovery distillation reboiler and the hydrotreating
distillation pre-heater. The final exchanger further cools the stream to 140C to raise low-
pressure saturated steam for closed-loop utility heating serving the ethanol distillation reboilers,
the molecular sieve unit, and AD heating demands (this is the largest heat demand of the three
exchangers). Specifics on the steam side of the exchangers and associated costing assumptions
are discussed in Section 3.7. As mentioned previously, while the turbine generates sufficient
electricity to make the conversion facility a net-power exporter to garner a small power
coproduct revenue, the overall facility is slightly heat-limited given the heat raised by flue gas
cooling relative to total heat/steam demands (driven primarily by ethanol distillation).
Consequently, a small amount of supplemental natural gas is imported and co-fed to the gas
turbine to satisfy the marginal heat deficit.

3.5.3 Cost Estimation

The design and cost estimates for the AD system are based on NRELs recent 2013 cellulosic
design report model [6], which in turn was based on updated estimates for AD (and other
operations) provided from an engineering subcontract with wastewater vendor Brown and
Caldwell. Under that subcontract, Brown and Caldwell provided a rigorous review and detailed
design of a number of wastewater treatment operations, drawing on support from technology
suppliers. The AD system designed by Brown and Caldwell consisted of multiple AD vessels
each sized at 27 MM gal, thus the same maximum size criteria was applied for this analysis. Four
such AD units were specified in the original design at a total purchase cost of $25.8MM (2012-
dollars), translating to $6.5MM per individual vessel; this cost also includes supporting
equipment including recycle pumps, biogas blowers, mixers, centrifugation, and a biogas
emergency flare. The original cost was specified to be scaled according to feed chemical oxygen
demand (COD) to the AD system, which was an appropriate basis for scaling in the original
cellulosic biochemical process model. However, in this case COD is a less certain value as
estimated by the Aspen model given the underlying biomass component characterizations.
Consequently, the cost was scaled in the present analysis according to required AD volume,
calculated based on the volatile solids loading rate discussed above and constrained by a
maximum size of 27 MM gal per single AD unit. The design and cost basis employed here is
derived from a low-rate bulk volume flow (BVF) digester system designed for NRELs
cellulosic biofuel model, which originally assumed long retention times on the order of >40

44
days; the high biogas yields and short retention times assumed in this design may require
alternative AD designs, such as smaller (more costly) agitated dual-reactor systems. This will be
a continued area of further investigation.

Added to the resulting overall AD system cost, a digestate centrifuge was also included as 5% of
the cost of the AD system, consistent with the 2013 cellulosic design report. The gas turbine cost
basis was left unchanged relative to the 2012 harmonization baseline, at an installed cost of
$445/KW generated (2006-dollars) [63], which was originally presented in the GTW 2006
Handbook [66].

3.6 Area 600: Storage

3.6.1 Overview
This portion of the plant provides bulk storage for process chemicals and the liquid fuel products.
The tanks explicitly costed include diesel, ethanol, and naphtha storage, as well as water for fire
suppression. Additional tankage for chemical inputs is included as a balance of plant factor. All
assumptions for storage costs and logistics are set consistently with other recent design report
practices, and may not consider more detailed logistics challenges for loading/unloading, rail car
transport, etc. (considered beyond the scope of this stage of analysis).

3.6.2 Design Basis

Table 13 shows the major storage requirements for the present design. Tanks for diesel, ethanol,
and naphtha products were each sized to ensure at least seven days of storage, while the fire
water storage tank was scaled from NRELs 2013 design report according to dry biomass feed
rate (summer/spring capacity case) to the pretreatment reactor.

Table 13. Storage Requirements for Major Tanks

Material Size
RDB product Sufficient to contain >7 days of production: 1 carbon steel tank @ 750,000 gal
Ethanol product Sufficient to contain >7 days of production: 1 carbon steel tank @ 750,000 gal
Naphtha product Sufficient to contain >7 days of production: 1 carbon steel tank @ 25,000 gal
Fire water 4 hours of fire suppression @ 2,500 gpm: 1 glass-lined carbon steel tank @
600,000 gal (scaled down to lower dry feed rate in present design)

Other supplemental tanks and pumps were not considered explicitly here, but were costed based
on a 20% balance of plant factor relative to the total cost of the major tanks listed above, based
on similar minor tankage costs considered in detail in NRELs 2013 design report.

3.6.3 Cost Estimation

The costs for the A700 storage section were based on NRELs prior cellulosic biochemical
design reports [3, 6], and costs are scaled according to new material flowrates to estimate new
tank prices if applicable.

45
 3.7 Area 700: Utilities
3.7.1 Overview
Area 700 contains the utilities required by the conversion facility. Area 700 tracks cooling water,
chilled water, process water, plant and instrument air, the clean-in-place (CIP) system, and steam
generation. In the model, Area 700 also tracks the electricity usage throughout the plant.

The process water manifold in Area 700 provides fresh process water, which is assumed suitable
for all plant users, at a constant pressure to the facility. Water is provided to the boiler for steam
generation, the cooling tower makeup, the CO2 vent water scrubber, and the lipid purification
operations. The plant and instrument air systems provide compressed air for general use
(pneumatic tools and cleanup) and instrument operation. Larger users of compressed air, namely
the biogas turbine, have their own compressors specified. The CIP system provides hot cleaning
and sterilization chemicals to pretreatment and fermentation operations.

3.7.2 Design Basis

The cooling water system is designed for a 28C supply temperature with a 9C temperature rise
in coolers throughout the facility. This is an assumed average rise; the actual cooling water rises
across each exchanger are not explicitly modeled in Aspen. The primary cooling water users in
this process are listed in Table 14. The percentage of cooling duty contributed by each user is
shown in Figure 12. The total cooling water demand is 20 MM kcal/hr, based on average cooling
requirements over the four individual modeled seasons.
Table 14. Cooling Water Users
Chiller Condenser The chilled-water loop requires cooling water to condense the refrigerant. The
cooling water duty for the chiller condenser is set equal to the total load on the
chilled-water loop.
Pretreatment Flash Cooler Cools the pretreatment flash condensate to suitable temperature for mixing
into the water recycle pool.
Fermentation Cooler Cools the hydrolysate feed to fermentation (37C).
Beer Column Condenser Condenses the beer column reflux.
Molecular Sieve Cooler Provides cooling and condensation during regeneration of the molecular
sieve.
Anaerobic Digester Feed Cools the AD feed stream to required operating temperature (35C).
Cooler
Hydrotreating Reactor Cools the effluent stream of the hydrotreater following cross-exchange with
Effluent Cooler the reactor inlet stream.

46
 Total = 20 MMkcal/hr (79 BTU/hr)
Pretreatment
Off Gas
Hydrotreating Chiller Condenser
Reaction Condenser 7%
Effulent Cooler 9%
13%
Anaerobic
Digester Feed
Cooler
17%
Fermentation
Cooler
37%
Molsieve Cooler
16%

Beer Column
Condenser
1%

Figure 12. Average cooling water heat duty distribution between major users

The largest user of cooling water is the fermentation cooler, associated with a large heat duty to
cool the pretreatment hydrolysate down from 68C to 37C after first recovering heat in the
pretreatment feed-effluent exchanger. Other larger cooling demands include the AD feed cooler,
molecular sieve cooler, and hydrotreating reactor effluent cooler, all operating between 35-38C
on the process side. Aspen computes the cooling tower evaporation rate based on a temperature
drop from 37C to 28C. Beyond evaporative water losses from the cooling tower, it was also
assumed that windage would be 0.005% of the total flow to the tower. The tower blowdown was
assumed to be 0.15% of the flow leaving the tower basin.

Chilled water is provided by two Trane centrifugal chillers, each originally specified at 2,350-ton
of refrigerant cooling. Per the chiller spec sheet, the compressor electricity demand for the chiller
was estimated at 0.56 KW/ton of refrigeration. The cooling water demand for the chiller system
was assumed to be equal to the heat removed in the chilled-water loop. The chilled-water loop
provides cooling to the ethanol and seed train fermenters to maintain fermentation temperature at
37C.

Fresh water is assumed to enter the facility at 33C and is split in several ways to meet process
needs. Clean water is used as boiler feed water to create steam for direct injection to the process
(pretreatment reactor), as makeup for utility steam loop blowdown, as makeup water for the
cooling tower, scrubbing water for the CO2 vent water scrubber, and as wash water for use in the
lipid purification unit. The process water tank is designed for an 8-hour residence time. The
process water pump pumps water from the tank into the facility.

47
The plant and instrument air systems provide compressed air for pneumatic tools and cleanup,
instrument operation, and air cooling systems. This does not include major air demands such as
biogas combustion. The plant air compressor was originally sized for 400 SCFM at 125 psig, and
scaled according to dry biomass feed rate to the facility. An instrument air dryer and surge tank
were designed to provide clean dry air at a consistent pressure to the instrument air system.

Steam is used in the plant both as a utility heat source and for direct injection into the process in
the pretreatment reactor system. Heat is recovered from the flue gas exhaust of the biogas
turbine, which is supplemented with natural gas to meet the heating balance needs of the facility
(discussed previously). High-pressure superheated steam (176 psig/ 13 atm, 268C) is generated
for direct injection into the pretreatment reactor system by cross-exchange with the first HRSG
exchanger. Heating demands of other processes are satisfied by two separate steam loops that
create saturated steam at high pressure and low pressure. The high-pressure saturated steam (665
psig, 46.3 atm) is used for the reboiler of the solvent recovery distillation column (Area 300),
and to heat the inlet stream on the RDB product distillation column (Area 400). The low-pressure
steam (39 psig, 2.7 atm) is used for heating the ethanol distillation beer and rectification column
reboilers, the molecular sieve heat requirement, and the heating needs of the anaerobic digester.
Additionally, a small amount of low-pressure superheated steam is also created for steam
stripping of the A400 product distillation column.

About 62% of the electricity generated by the gas turbine is used throughout the plant to power
items such as pumps, agitators and compressors. The surplus is sold to the grid for credit. The
distribution of total plant power utilization among all areas is shown in Figure 13. The largest
power users in the facility are the pretreatment and product purification/upgrading sections, each
accounting for 19% of total power demands relative to total generation; in turn, these are driven
primarily by power requirement for the pretreatment reactor feed system, the hydrotreating PSA
unit, and the hydrotreating makeup hydrogen compressor. Similar to the cooling water values,
the absolute value shown below for total power generated (18.5 MW) is the average of the four
individual seasonal models, as is the allocation to process area. The average annual excess power
produced and sold to the grid as a coproduct is roughly 6.9 MW, or more accurately, 54.9 MM
KWh/year. While power is viewed as a net exported coproduct in the current work scope focused
strictly on the conversion facility, when expanding the scope to a fully integrated process
including upstream cultivation and dewatering, a net power import would still likely be required
given high power demands upstream, e.g. for algal pond paddle wheels [8]. Thus the power
export value indicated here would go towards decreasing overall power demand for the fully
integrated process.

48
 Total Power Generated = 18.5 MW (Annual Average)

to A100
Pretreatment
and Conditioning
19%
Excess Electricity
to Grid to A200
38% Fermentation
and Distillation
7%

to A400 to A300 Lipid

Product Extraction and
Purification and Solvent Recovery
to A500 Upgrading 4%
to A700 Anaerobic 19%
Utilities Digestion/CHP
4% 9%
to A600 Storage
0%

Figure 13. Average distribution of plant electricity utilization by process area

3.7.3 Cost Estimation

All cost estimates for the utility equipment in A700 were maintained consistent with the basis
values used in NRELs 2011 and 2013 biochemical design reports. To summarize, the cooling
tower was based on a cost estimate from a vendor for a fiberglass cooling tower capable of
handling 44,000 gpm; this cost is scaled to the reduced cooling water throughput estimated here.
Harris Group had estimated costs for the cooling water circulation and makeup pumps using their
historical cost database. The material of construction for the cooling water loop is carbon steel.
The cost for the chiller came from a recent quote for a similarly-sized system. Aspen Capital
Cost Estimator [15] was used to cost the steam HRSG equipment including the high- and low-
pressure steam boiler heat exchangers. Harris Group had also used their historical cost database
to estimate costs for the remaining equipment: the process water tank and pump; the
plant/instrument air compressor, dryer, and surge tank; and the CIP system.

49
4 Process Economics
The ultimate purpose for developing such a detailed process design, simulation model, and cost
estimate is to determine the economics of biofuel production. This information is used either as
an absolute cost to assess the products potential in the marketplace or as a relative cost that can
be used to guide research by examining the change in production cost associated with a process
modification or other core research activity.

The total capital investment (TCI) is first computed from the total equipment cost. Next, variable
and fixed operating costs are determined. With these costs, we use a discounted cash flow
analysis to determine the minimum fuel selling price (MFSP) required to obtain a zero net
present value (NPV) with a finite internal rate of return (IRR). This section summarizes the
assumptions made in completing the discounted cash flow analysis, with more details and
supporting description available in NRELs 2011 and 2013 cellulosic design reports for
assumptions which remain unchanged [3, 6].

Our analysis does not take into account any policy factors such as subsidies, mandates, or carbon
credits, because these would be purely speculative. The purpose of this analysis is to demonstrate
the process requirements needed to achieve specific cost targets which are set from the top down,
and to demonstrate how the technology pathway described here is able to achieve such targets on
its own merits (through bottom-up TEA modeling) and, if it cannot, to give stakeholders a sense
of the magnitude of incentive required to make it so.

4.1 About Cost-Year Indices

The cost-year of 2011 was chosen for this analysis to provide more updated and relevant cost
output information relative to the 2007-year basis which had been consistently utilized for a
number of years in prior analyses. This new basis is being applied consistently across all DOE-
BETO platforms for which similar design case target reports are being established during
2013-2014 efforts, and it is expected that performance goals and TEA outputs will remain in
2011-dollars through 2017 to permit comparison of future feedstocks, conversion technologies,
and other alternative scenarios. However, the present equipment costs were obtained in 2012-
dollars or 2013-dollars for new pieces of key equipment which were added or replaced in the
current design model (as part of the subcontract with Harris Group), and in 2009-dollars or 2010-
dollars for unit operations which were previously provided by Harris Group and vendors in
support of NRELs earlier design reports, and which are still maintained in the present model.
Cost-years for chemicals range from 1999 to 2012.

The methods used for determining MFSP in another years dollar value and for scaling capital,
operating, and labor cost estimates to a desired target year remain the same as described in prior
design reports. Thus, the details will not be repeated here, but will be summarized briefly.
Capital costs provided in a year other than 2011-dollars were adjusted using the Plant Cost Index
from Chemical Engineering Magazine [67] to a common basis year of 2011. The final cost index
for a given year is generally not made available until the spring of the following year. Therefore,
for the small number of equipment items that were quoted in 2013-dollars, we assumed the same
Plant Cost Index value from 2012 (all cost quotes which fall in this category were provided in
the first half of 2013). Consistent with methods utilized for all other recent design reports, costs
are not inflated over the 30-year facility lifetime (i.e. inflation is kept constant over the modeled

50
facility lifetime). Similar to capital costs, for chemical costs we used the Industrial Inorganic
Chemical Index from SRI Consulting [68]. Employee salaries were maintained from 2009-
dollars and were scaled using the labor indices provided by the United States Department of
Labor Bureau of Labor Statistics [69]. The general formula for converting nominal values to real
values and for adjusting dollars to a common basis is:

2011 Cost Index

2011 Cost = (Base Cost)
Base Year Index

4.2 Total Capital Investment (TCI)

Section 3 of this report describes the details of the conceptual process design and how the
purchased cost of the equipment was determined. The next step is to determine the installed cost
of that equipment. The installation cost can be determined by performing a detailed study of
everything required to install the necessary equipment and make it operational (e.g., foundation,
piping, and wiring). This type of detail is not warranted at this level of analysis, and a factored
approach in which multipliers are applied to the purchased equipment cost is considered
satisfactory. The methodology and rationale for applying unit-level installation costs remains the
same as described in the 2011 and 2013 cellulosic design reports, and again further detail can be
found there which will not be repeated here. In summary, each type of equipment utilizes a
different installation factor to scale the given direct equipment purchased cost to a final installed
cost, with these factors generally varying between 1.5 and 3.0. A complete listing of the
equipment is provided in Appendix A, along with equipment purchased and installed costs. As
described in prior design reports utilizing similar equipment, Harris Group obtained many
package quotes for recent cost estimates, in which a given unit operation and all of its support
equipment were quoted under one price. The installation factor for such packages can be
relatively low because most of the engineering is already included in the price. Additionally,
equipment furnished as a pre-fabricated skid generally has a lower installation cost relative to
equipment cost.

The purchased cost for a given component reflects a baseline equipment size. As changes are
made to the process, the required equipment size may be different than what was originally
designed. Instead of re-costing in detail, an exponential scaling expression was used:

New Size
New Cost = (Base Cost)
Base Size

In this equation, the scaling exponent n varies depending on the type of equipment to reflect
economy of scale dependencies (more detail on reasonable scaling values for different types of
equipment is provided in NRELs 2011 ethanol report). The basis for scaling is typically some
characteristic of the equipment related to production capacity, such as flow or heat duty. Some
equipment does not follow such a scaling-factor approach, namely when the capacity for a given
operation is exceeded and requires multiple units in parallel (thus losing economy of scale
benefits which are captured in the exponential expression above). A good example of this is the
new pretreatment reactor cost information discussed in Section 3, which has resulted in a
modified cost estimate to utilize multiple discrete sections of the horizontal portion of the

51
pretreatment reactor (e.g. the portion which is in contact with acid) to allow for variations in
residence time through this section of the reactor.

Once the total equipment cost has been determined in the year of interest, we must add several
other direct and indirect costs to determine the total capital investment (TCI). Site development
and warehouse costs are based on the inside-battery-limits (ISBL) equipment costs (Areas 100-
400) and are considered part of the total direct cost (TDC). Beyond ISBL Areas 100-400, the
other process areas are considered outside battery limits (OSBL) including Areas 500-700
(AD/CHP, storage, and utilities). Project contingency, field expenses, home-office engineering
and construction activities, and other costs related to construction are computed relative to the
TDC and give the fixed capital investment (FCI) when summed. Similar to the choice of 10%
IRR, the cost factors applied in estimating FCI are based on assuming nth-plant economics,
particularly for less prescriptive elements such as project contingency factor (see Section 1.4).
The sum of FCI and the working capital for the project is the TCI. Table 15 summarizes these
categories and additional factors. The values assumed for each respective factor were maintained
consistently with those discussed in NRELs 2011 and 2013 design reports.

Table 15. Additional Costs for Determining Total Capital Investment (TCI)
Item Description Amount
Additional direct costs
Warehouse On-site storage of equipment and supplies. 4.0% of the
installed cost of
ISBL equipment
(A100-400)
Site development Includes fencing, curbing, parking lot, roads, well 9% of ISBL
drainage, rail system, soil borings, and general paving.
This factor allows for minimum site development
assuming a clear site with no unusual problems such as
right-of-way, difficult land clearing, or unusual
environmental problems.
Additional piping To connect ISBL equipment to storage and utilities 4.5% of ISBL
outside the battery limits.
Indirect costs
Prorateable costs This includes fringe benefits, burdens, and insurance of 10% of total direct
the construction contractor. cost (TDC)
Field expenses Consumables, small tool and equipment rental, field 10% of TDC
services, temporary construction facilities, and field
construction supervision.
Home office and Engineering plus incidentals, purchasing, and 20% of TDC
construction construction.
Project contingency Extra cash on hand for unforeseen issues during 10% of TDC
construction.
Other costs Start-up and commissioning costs; land, rights-of-way, 10% of TDC
permits, surveys, and fees; piling, soil
compaction/dewatering, and unusual foundations; sales,
use, and other taxes; freight, insurance in transit, and
import duties on equipment, piping, steel and
instrumentation; overtime pay during construction; field
insurance; project team; transportation equipment, bulk
shipping containers and plant vehicles.

52
 Table 16. Project Cost Worksheet Including Total Direct Costs and Total Capital Investment
Process Area Purchased Cost a Installed Cost a
Area 100: Pretreatment and Conditioning $ 42,800,000 $ 65,600,000
Area 200: Fermentation and Distillation $ 10,500,000 $ 19,600,000
Area 300: Lipid Extraction and Solvent Recovery $ 34,900,000 $ 71,500,000
Area 400: Product Purification and Upgrading $ 48,300,000 $ 48,300,000
Area 500: Anaerobic Digestion/CHP $ 20,500,000 $ 21,600,000
Area 600: Storage $ 2,700,000 $ 4,500,000
Area 700: Utilities $ 4,200,000 $ 5,700,000
Totals (excl. Area 100) $ 163,900,000 $ 236,700,000
Warehouse 4.0% of ISBL $ 8,200,000
Site development 9.0% of ISBL $ 18,400,000
Additional piping 4.5% of ISBL $ 9,200,000
Total Direct Costs (TDC) $ 272,600,000
Prorateable expenses 10.0% of TDC $ 27,300,000
Field expenses 10.0% of TDC $ 27,300,000
Home office & construction fee 20.0% of TDC $ 54,500,000
Project contingency 10.0% of TDC $ 27,300,000
Other costs (start-up, permits, etc.) 10.0% of TDC $ 27,300,000
Total Indirect Costs $ 163,500,000

Fixed Capital Investment (FCI) $ 436,100,000

Land $ 1,800,000
Working capital 5.0% of FCI $ 21,800,000
Total Capital Investment (TCI) $ 459,800,000

Lang Factor (FCI/purchased equip cost) 2.7

TCI per annual GGE $7.37/GGE
2011-dollars
HSLD + Store
a
Capital costs are based on peak (summer/spring) design capacity following excess summer
drying/storage scenario

4.3 Variable Operating Costs

Variable operating costs, which include feedstock, other raw materials, and co-product credits,
are incurred only when the process is operating. In this pathway, variable operating costs by
definition also vary as seasonal feed rates and associated equipment throughputs fluctuate, in
contrast to fixed operating costs (and equipment capital costs) which do not. Quantities of raw
materials used and coproducts produced were determined using the Aspen material balance,
based on running individual Aspen models for each season and then taking the average of all
four seasons in setting the net annual operating expenses and coproduct revenues. Table 17
documents the costs and sources of chemicals used in the process and Table 18 summarizes the
variable input/output flowrates on a per-season basis, as well as the resulting variable costs and
coproduct revenues on a per-year and per-GGE-of-total fuel basis.

The cost basis for all material costs in the present model, which were also used in NRELs most
recent 2013 cellulosic design report, were left unchanged and similarly scaled to 2011-dollars.
This includes sulfuric acid, ammonia and DAP (both as chemical inputs as well as coproduct
credits translated from nitrogen and phosphorus balances allocated to the AD effluent stream),
process water, and hydrogen. As discussed in the 2013 design report, hydrogen was assumed

53
here to be purchased as a product from standard natural gas-derived steam methane reforming
(SMR), priced according to a recent DOE Hydrogen Program report at $1.57/kg (2012-dollars)
[70]. Consistent with the 2013 report, given that hydrotreating is not yet well-defined with direct
experimental data for lipid upgrading (but rather is based on literature data), the decision was
made to maintain the assumption around purchased hydrogen rather than including on-site
generation (from AD biogas and/or natural gas reforming), so as not to place too much burden on
an alternative on-site hydrogen production system without first validating the hydrotreating
operation, particularly hydrogen consumption. Additionally, as this would be considered an
OSBL process, using sound literature values for hydrogen price reduces overall uncertainty
relative to building a new hydrogen generation system into the model. Further details behind the
rationale for the hydrogen price assumptions may be found in the 2013 cellulosic design report.
The overall cost sensitivity to the assumed hydrogen price is presented in Section 5.1, using a
reasonable maximum and minimum hydrogen cost range as presented in the referenced DOE
report, namely $1.10/kg-$2.00/kg.

Aside from feedstock cost discussed earlier, other raw material inputs to the present model which
were not utilized in prior analyses include hexane (costed based on an industrial spot price);
phosphoric acid, clay, and silica for lipid purification (costed based on price estimates furnished
by Harris Group); and natural gas. Cost for natural gas, used both for drying excess summer
feedstock for use in the winter as well as for utility supplemental heating, was set at
$5.1/thousand ft3 [11]. This is within the current price range although slightly on the
conservative side relative to recently published prices of $3.89 and $4.66/thousand ft3 in 2012
and 2013 respectively [71].

This design includes a number of coproducts, including AD digestate cake (valued at $500/tonne
bioavailable N as a fertilizer product, as discussed earlier), AD effluent nitrogen and phosphorus
destined for recycle to upstream cultivation (translated to equivalent NH3 and DAP rates, using
coproduct credits equal to these materials costs, as noted above), flue gas and fermentor vent
CO2 also destined for recycle to upstream cultivation (valued at $33.9/ton of CO2, as discussed
above), excess power sold to the grid, and naphtha. The products and recycle streams from AD
being sent back to upstream cultivation are warranted to include for (small) coproduct revenues
here, as they were not originally counted in the $430/ton biomass feedstock cost basis, thus
instead are accounted for here (see discussion in Section 3.5). The electricity export price, taken
to be the average wholesale price as determined by the North American Electric Reliability
Corporation, was left unchanged from prior NREL 2011 and 2013 design reports, set at 5.72
cents/KWh. This corresponds to roughly 84% of the EIA 2011 average industrial retail price for
electricity of 6.8 cents/KWh [72], which is consistent with published values for wholesale
electricity as a fraction of retail prices [73]. While diesel and ethanol products are summed
together according to GGE heating values to form the overall GGE fuel product for reasons
discussed above, naphtha was assigned as a separate coproduct given its much smaller
production rate and likelihood that it may not be appropriate directly as a fuel or blendstock, but
rather may require further processing. The naphtha coproduct value was set as $3.25/gal as
consistent with [11]. The sensitivity impact on overall fuel MFSP attributed to removing any of
these coproduct credits is considered below in Section 5.1.

54
 Table 17. Chemical Costs + Coproducts and Sources
Component Cost (2011$) Source
Inputs
Biomass feedstock $0.2150/lb 2013 MYPP, $430/dry ton AFDW @ 80% moisture [9]
Hexane $0.5394/lb SRI Chemical Economics Handbook [68]
Sulfuric acid, 93% $0.0499/lb Basic Chemical of Omaha via Harris Group [6]
Ammonia $0.2496/lb Terra Industries via Harris Group [6]
Diammonium phosphate $0.5492/lb Ronas Chemicals via Harris Group [6]
Natural Gas $0.1051/lb $5.1/thousand ft3 [11]
Hydrogen $0.6838/lb DOE report, SMR derived H2 [6, 70]
Process Water $0.0001/lb Peters & Timmerhaus [6]
Phosphoric acid (lipid
$0.3600/lb Harris Group cost database
cleanup)
Silica (lipid cleanup) $1.0000/lb Harris Group cost database
Clay (lipid cleanup) $0.3000/lb Harris Group cost database
Coproducts
AD digestate cake $0.2419/lb of N $500/tonne of bioavailable N (40% of digestate N) [6, 55]
Ammonia recycle $0.2496/lb Set equal to input cost per above
Diammonium phosphate $0.5492/lb Set equal to input cost per above
Naphtha $3.25/gal Consistent with algal HTL design report [11]
Carbon dioxide recycle $0.0169/lb $33.89/ton CO2; extrapolated from flue gas pipeline cost in [8]
Power coproduct to grid $0.0572/KWh Consistent with prior design reports [6]

Table 18. Variable Operating Costs. (Includes seasonal flowrates and resulting annual costs; HLSD
summer drying/storage design basis.)
Process Stream Description Summer Fall Winter Spring Cost MM$/yr Cent/GGE
Area Usage Usage Usage Usage ($/ton) (2011$) Fuel
(kg/h) (kg/h) (kg/h) (kg/h) 2011$ (2011$)
Raw materials
A100 Feedstock input a 84,224 b 47,767 15,705 b 54,749 430.00 190.03 304.74
Sulfuric acid, 93% 2,433 2,099 1,985 2,406 99.85 1.95 3.12
Ammonia 786 678 641 777 499.27 3.14 5.04
Natural gas for
8,111 0 0 0 210.27 3.72 5.97
summer drying
A200 DAP 112 97 92 111 1,098.38 0.99 1.58
A300 Hexane 993 856 810 982 1,078.85 8.58 13.75
Phosphoric acid 43 37 35 42 720.00 0.25 0.40
Silica 23 19 18 22 2,000.00 0.36 0.58
Clay 45 39 37 45 600.00 0.22 0.35
A400 Hydrogen 546 471 446 540 1,367.51 5.98 9.59
A700 Process water 85,179 73,448 69,745 84,247 0.29 0.20 0.32
Supplemental natural
949 799 748 937 210.27 1.58 2.53
gas for heating
Subtotal 216.98 347.95
By-products and credits
A400 Naphtha (gal/hr) 142 122 116 140 $3.25/gal 3.35 5.37
A500 AD Digestate N c 71 62 58 71 483.79 0.28 0.44
AD Ammonia 740 638 604 732 499.27 2.96 4.74
AD DAP 322 277 262 318 1,098.38 2.83 4.54
CO2 Recycle 34,005 29,279 27,673 33,618 33.89 9.22 14.78
Grid electricity (KW) 7,637 6,465 6,116 7,528 5.72/KWh 3.14 5.04
Subtotal 21.77 34.91
Total variable operating costs 195.21 313.05
a
Feedstock flowrate is on an ash-free dry weight basis. Values represent raw feed rate of feedstock to the facility
from dewatering, prior to diversion of summer excess to storage.
b
Total amount of summer feedstock diverted to storage for winter usage is 29,475 kg/h.
c
Flowrate of bioavailable N in total AD digestate cake stream

55
4.4 Fixed Operating Costs
Fixed operating costs are generally incurred in full whether or not the plant is producing at full
capacity. These costs include labor and various overhead items. The assumptions on fixed
operating costs were maintained consistently from NRELs prior 2011 and 2013 cellulosic
design cases [3, 6], which in turn were based in large part on NRELs 2002 ethanol design report
[2] and/or Peters and Timmerhaus [74].

Table 19 shows the recommended number of employees and associated salaries. The number of
employees was estimated by considering the likely degree of automation for each area and
adding a reasonable number of management and support employees. Details behind the assumed
number of employees and associated salaries are provided in NRELs 2011 ethanol report.

Table 19. Fixed Operating Costs

Position 2011 # 2011 Cost MM$/yr Cent/GGE
Salary Required (2011$) Fuel (2011$)
Labor and supervision
Plant manager 155,400 1 155,400
Plant engineer 74,000 2 148,000
Maintenance supervisor 60,257 1 60,257
Maintenance technician 42,286 12 507,429
Lab manager 59,200 1 59,200
Lab technician 42,286 4 169,143
Shift supervisor 50,743 4 202,971
Shift operators 42,286 20 845,714
Yard employees 29,600 4 118,400
Clerks and secretaries 38,057 3 114,171
Total salaries 2,380,686 2.38 3.82
Labor burden (90%) 2,142,617 2.14 3.44
Other overhead
Maintenance 3.0% of ISBL 6,147,070 6.15 9.86
Property insurance 0.7% of FCI 3,052,774 3.05 4.90
Total fixed operating costs 13.72 22.01

A 90% labor burden is applied to the salary total and covers items such as safety, general
engineering, general plant maintenance, payroll overhead (including benefits), plant security,
janitorial and similar services, phone, light, heat, and plant communications. The 90% estimate is
the median of the general overhead range suggested in the 2008 PEP Yearbook produced by SRI
Consulting [68]. Annual maintenance materials were estimated as 3% of the installed ISBL
capital cost and property insurance and local property tax were estimated as 0.7% of the fixed
capital investment, based on the 1994 Chem Systems report described in NRELs 2011 ethanol
report.

4.5 Discounted Cash Flow Analysis and the Minimum Selling Price of
Fuel
Discount Rate
For this analysis, the discount rate (which is also the internal rate of return [IRR] in this analysis)
was set to 10% and the plant lifetime was set to 30 years. The discount rate was also used in
previous design reports and was based on the recommendation in Short et al. [75] on how to

56
perform economic evaluations of renewable energy technologies for DOE. His view was that,
In the absence of statistical data on discount rates used by industrial, transportation and
commercial investors for investments with risks similar to those of conservation and renewable
energy investments, it is recommended that an after tax discount rate of 10%be used.

Equity Financing
For this analysis, it was assumed that the plant would be 40% equity financed. The terms of the
loan were taken to be 8% interest for 10 years. The principal is taken out in stages over the 3-
year construction period. Interest on the loan is paid during this period, but principal is not paid
back (this is another nth-plant assumption, which says that this cash flow comes from the parent
company until the plant starts up). This is all consistent with the assumptions used in prior NREL
design reports. Figure 14 illustrates the sensitivity of minimum fuel selling price to the
percentage of equity financing and the after-tax discount rate (the IRR).

$6.00

$5.75 10% IRR

Minimum Fuel Selling Price ($/GGE)

15% IRR
$5.50
20% IRR
$5.25

$5.00

$4.75

$4.50

$4.25

$4.00
0% 20% 40% 60% 80% 100% 120%
Equity %

Figure 14. Sensitivity of MFSP to IRR and % equity (8% interest on a 10-year loan)

Depreciation
To determine the capital depreciation amount for the calculation of federal taxes to be paid we
used the IRS Modified Accelerated Cost Recovery System (MACRS). Within the MACRS
system is the General Depreciation System (GDS), which allows both the 200% and 150%
declining balance (DB) methods of depreciation. This offers the shortest recovery period and the
largest tax deductions. According to IRS publication 946 [76], a biorefinery plant would fall
under Asset Class 49.5, Waste Reduction and Resource Recovery Plants. This class uses a 7-
year recovery period, not including the power plant (i.e., gas turbine) equipment, which has a 20-
year recovery period (Asset Class 49.13). Again, all assumptions employed here are maintained
consistently with prior design reports.

57
Taxes
The federal corporate tax rate used in our analysis is 35%. Income tax is averaged over the plant
life and that average is calculated on a per-gallon basis. The amount of income tax to be paid by
a potential fuel producer varies annually due to changes in the volume of product produced and
the allowable depreciation deduction. In fact, no income tax is paid in the first nine years of
operation because the depreciation and loan interest deductions are greater than the net income.
State taxes are not considered, primarily because the location of the plant has not been
determined and tax rates vary from state to state (from 0% to 12%); this is consistent with
previous design report practice standards and is applied consistently here as well.

Construction Time
The construction time is important to the cash flow analysis because no income is earned during
construction, but huge sums of money are being expended. Construction time assumptions were
left unchanged from the 2011 and 2013 cellulosic design reports. Perry and Green [77] indicates
that small projects (less than $10 million investment) can be constructed in fewer than 18 months
and that larger projects can take up to 42 months. An overview of petroleum refining economics
indicates that large refineries (on the order of a $1.5 billion investment) can be constructed in 24
months [78]. Certainly this algal biofuel process is much smaller than a petroleum refinery, so
using a construction time of 24 months fits within these references, although an important
difference between this type of facility and a refinery is the large number of field-erected vessels.
These are constructed on site and have a longer construction time than if the tanks were delivered
finished. Table 20 summarizes the schedule for construction and the cash flow during that time.
Twelve months are added before construction for planning and engineering.

Table 20. Construction Activities and Cash Flow

Project Project % of
Start End Activity Description Project
Month Month Cost
0 12 Project plan and schedule established; conceptual and basic 8%
design engineering, permitting completed; major equipment bid
packages issued, engineering started on selected sub-packages,
P&IDs complete, and preliminary plant and equipment
arrangements complete.
12 24 All detailed engineering including foundations, structure, piping, 60%
electrical, site complete; all equipment and instrument
components purchased and delivered; all site grading, drainage,
sewers, rail, fire pond, foundation, and major structural
installation complete; 80% of all major process equipment set (all
except longest-lead items); all field fabricated tanks built; and the
majority of piping and electrical materials procured.
24 36 Complete process equipment setting, piping, and instrumentation 32%
installation complete; all electrical wiring complete; all building
finishing and plumbing complete; all landscaping complete; pre-
commissioning complete; and commissioning, start-up, and initial
performance test complete.
TOTAL 100%
Note: The above assumes no utility or process equipment orders placed prior to month seven.
Expenditures based on typical 60 MM gal/yr grain-to-ethanol facility.

58
Start-Up Time
Perry and Green [77] indicates that for a moderately complex plant, start-up should be about
25% of the construction time, or 6 months in this case. While the 2011 ethanol model assumed a
start-up time of 3 months under an nth plant assumption, this is extended back to 6 months in the
present analysis. The start-up period is not completely wasted, however. We expect that an
average of 50% production could be achieved during that period while incurring 75% of variable
expenses and 100% of fixed expenses.

Working Capital
Peters and Timmerhaus [74] defines working capital as money available to cover (1) raw
materials and supplies in inventory, (2) finished product in storage, (3) accounts receivable, (4)
cash on hand for monthly payments such as wages and maintenance supplies, (5) accounts
payable, and (6) taxes payable. The present analysis applies the same basis for working capital as
was used in prior work, namely 5% of fixed capital investment.

Table 21 summarizes the parameters used in the discounted cash flow analysis. Using these
parameters, plus the cost information in Table 16, Table 18, and Table 19, the resulting
minimum fuel selling price of the primary fuel products translated to an energy-equivalent GGE
yield is $4.35/GGE (2011-dollars). Based on individual product heating values, this equates to a
selling price of $2.95/gal for ethanol and $4.57/gal for RDB. It is worthwhile to distinguish this
MFSP result, based on 10% IRR and other economic inputs noted above, from the cost of
production (i.e. breakeven cost to cover capital and operating expenses at 0% IRR), as these
terms are often incorrectly used interchangeably; the associated cost of production for this case is
estimated at $3.68/GGE. Table 22 summarizes the yields and conversion costs for the present
design. According to the methodology of Cran [79], the expected accuracy of the TCI analysis is
+/- 25%. If we apply this uncertainty to the TCI, the impact on the MFSP is +/-$0.26/GGE. The
complete discounted cash flow summary worksheet is shown in Appendix B. The MFSP can be
further broken down into the cost of each process area. Figure 15 illustrates the contribution to
the overall cost by process area and capital, operations, and fixed costs (the bar for feedstock
plus handling reflects the single feedstock cost of $430/dry U.S. ton delivered to the conversion
facility and has not been broken down).

Table 21. Discounted Cash Flow Analysis Parameters

Plant life 30 years
Discount rate 10%
General plant depreciation 200% declining balance (DB)
General plant recovery period 7 years
Steam (turbine) plant depreciation 150% DB
Steam (turbine) plant recovery period 20 years
Federal tax rate 35%
Financing 40% equity
Loan terms 10-year loan at 8% APR
Construction period 3 years
First 12 months expenditures 8%
Next 12 months expenditures 60%
Last 12 months expenditures 32%
Working capital 5% of fixed capital investment
Start-up time 6 months
Revenues during start-up 50%
Variable costs incurred during start-up 75%
Fixed costs incurred during start-up 100%

59
 Table 22. Summary of Yields, Rates, and Conversion Costs
Feedstock rate 1,339 dry U.S. ton/day average annual rate (AFDW basis)
On-line time 7,920 h/yr (330 days/yr, i.e. 90% on-line factor)
Total fuel yield (RDB + ethanol) 141.1GGE/dry U.S. ton feedstock (AFDW)
Total fuel production rate (RDB + ethanol) 62.4 MM GGE/yr
RDB production rate 46.3MM GGE/yr (44.1 MM gal/yr)
Ethanol production rate 16.1 MM GGE/yr (23.7 MM gal/yr)
Total equipment cost $237MM
Total capital investment (TCI) $460MM
TCI per annual gallon $7.37/GGE
Minimum Fuel Selling Price $4.35/GGE
Feedstock contribution $3.05/GGE
Conversion contribution $1.30/GGE

Figure 15. Cost contribution details from each process area (per GGE total fuel)

As shown in Figure 15, feedstock cost dominates overall system economics at the $430/ton price
assumption employed here, accounting for 70% of total MFSP. This is to be expected, when
comparing to terrestrial cellulosic biofuel pathways utilizing feedstock costs closer to $80/ton for
materials such as corn stover or wood [6, 28], and highlights the key challenge facing algal
biofuels being the high cost of biomass production (see [8] for further quantitative details on
estimated cost allocations for upstream steps). Such a result is consistent with the recent algal
HTL design report [11] as well as DOEs MYPP projections [9], where the latter case in fact
estimated cultivation as contributing 80% to MFSP. In this light, cultivation and harvesting
operations will continue to be a critical area of further analysis moving forward, and is
anticipated to be a primary focus for NREL algal TEA modeling in 2015 to expand on earlier
modeling efforts that considered front-end operations.

60
Beyond feedstock costs, the pretreatment fractionation and lipid extraction process areas are the
largest process area contributors to MFSP cost, contributing 50/GGE and 52/GGE
respectively. In both cases, capital costs are the largest driver at roughly 60% of these values, in
turn dominated by the pretreatment reactor system and the extraction column as single line item
costs in each respective area. The next cost driver is the lipid purification and upgrading area,
with primary cost contributions from hydrotreating and PSA capital expenses, offset slightly by
the small naphtha coproduct. Aside from naphtha, the remaining coproducts are all allocated to
the AD/CHP system which overall provides for a net negative 26/GGE contribution to MFSP
after accounting for all capital expenses, associated with coproduct credits from CO2 and N/P
nutrient recycles to upstream cultivation, AD digestate cake, and the exported power. Cost
contributions attributed to fermentation and distillation are fairly small at 13/GGE.

It is noteworthy that even at a feedstock cost of $430/ton (five-fold higher than terrestrial
cellulosic feedstocks), combined with relatively low credits from all coproducts under
$0.35/GGE, the modeled pathway evaluated here still achieves a fuel MFSP of $4.35/GGE. This
is a similar cost as the above-referenced algal HTL design report at $4.49/GGE, enabled in both
cases by a highly favorable fuel yield at 138 GGE/ton AFDW in this design and 130 GGE/ton
AFDW in the HTL case [11] (excluding contributions from naphtha in both cases). In
comparison, cellulosic biofuel pathways have presented design case fuel yield targets of 45
GGE/ton via biochemical conversion of corn stover to hydrocarbons [6], 52 GGE/ton via
biochemical conversion of corn stover to ethanol [3], 87 GGE/ton via pyrolysis of woody
biomass to hydrocarbons [28], and 63 GGE/ton via thermochemical conversion of woody
biomass to mixed alcohols [5]. The ALU fractionation and HTL design cases even surpass the
original DOE 2022 target of 117 GGE/dry ton as presented in the May 2013 MYPP report [9],
which was based only on lipids. Thus, the fuel yields presented here on the order of 1.5 to 3
times higher than the above-noted cellulosic pathways makes up for much of the higher
feedstock cost. While the analysis presented here demonstrates a viable and realistic path to
achieve cost targets below $5/GGE, additional improvements are required to ultimately reduce
MFSP costs towards $3/GGE targets. A brief discussion of possible paths toward this goal is also
included in Section 5.4.

61
5 Analysis and Discussion
5.1 Cost Sensitivity Analysis
A single-point sensitivity was performed using the variables and limits shown in Table 23. The
baseline for all variables used in the design case is described previously in this report.
Reasonable minima and maxima for each variable were chosen to understand and quantify the
resulting cost impact on overall MFSP. Each variable was changed to its maximum and
minimum value with all other factors held constant. The results of the sensitivity analysis are
displayed as a tornado plot in Figure 16.

Table 23. Assumptions Varied in the Sensitivity Analysis

Area Assumption Min Baseline Max
Biomass Delivery Feedstock cost, $/ton 300 430 550
Average feed rate, ton/day 650 1339 2205
Feed solids content, wt% 15% 20% 25%
a
Dryer vs. No dryer No dryer Dryer -
Pretreatment Pretreatment glucose yield 80% 90% 95%
Pretreatment HMF yield 0% 0.3% 2%
Pretreatment acid loading, wt% of liquor feed 0.5% 1% 2%
Pretreatment residence time, min 2 5 10
Pretreatment reactor metallurgy Stainless steel High alloy -
Fermentation Fermentation glucose to ethanol conversion 85% 95%
Fermentation contamination loss 0% 3% 6%
Extraction Extraction lipid yield 85% 95% 98%
Extraction CAPEX -50% 0% +50%
Refining/Upgrading Lipid purification requirement Not required Required -
PSA H2 recovery 85% 91% 99%
H2 price, $/kg 1.2 1.6 2
Hydrotreating CAPEX -50% 0% +100%
Anaerobic Digestion AD volatile solids loading factor, g/L-day 1 3.3 7
Financial/Other Naphtha coproduct credit, $/gal 2.75 3.25 3.75
AD digestate N credit, $/tonne bioavailable N 30 500 700
CO2 recycle credit, $/ton 0 34
AD effluent N/P recycle credit Not included Included
Power coproduct credit, /KWh 5 5.7 7
Total Capital Investment (TCI) -25% 0% +25%
a
Dryer vs. no dryer scenario is still based on base case with excess summer storage; this case assumes storage
without the dryer CAPEX and natural gas OPEX costs in order to demonstrate the cost impact associated with
summer drying.

As shown in Figure 16, biomass feedstock cost translates to the strongest single MFSP influence,
weighing on overall economics considerably more heavily than nearly any other parameter. A
feedstock cost of $550/ton increases MFSP by nearly $1/GGE relative to the $4.35/GGE
baseline, while $300/ton decreases MFSP by a similar amount, which would reduce cost to
$3.42/GGE. The baseline feedstock cost of $430/ton is set consistent with the above-referenced

62
algal HTL design report, in both cases based on DOEs published MYPP target values [9].
Additionally, average feed rate to the conversion facility also exhibits a noticeable impact on
MFSP (although substantially less than feedstock cost), particularly at lower feed rates due to
loss of economy of scale benefits. For the sensitivity analysis considered here, feed rate was
varied from a low of 650 ton/day (roughly 50% of the base case) to a high of 2,205 ton/day
(2,000 metric tonne/day) corresponding to standard cellulosic biomass modeling cases [3, 5, 6].
All feed rates here are based on annual average rates, with seasonal variability maintained
consistent with the base case between individual seasons (see Section 2.1).

The targeted extraction lipid yield and capital cost also make significant impacts to MFSP. The
extraction efficiency is significant primarily because it directly impacts RDB yield which is the
primary fuel product from the integrated process, and thus greatly affects the overall fuel energy
yield (GGE yield) more substantially than similar fractional differences in pretreatment sugar
(glucose) yield. Extraction capital cost was also identified above as one of the most significant
single line items of the overall facility capital cost estimates. Total capital investment (TCI) and
hydrotreating capital costs also significantly impact MFSP, although these two variables are
related, as hydrotreating capital cost is part of TCI. The most significant components of TCI are
the extraction column, pretreatment reactor, and hydrotreating capital cost. The algae dryer also
accounts for a significant portion of TCI, and is presented separately in the tornado plot when
considering removal of the dryer capital cost and natural gas operating cost while maintaining
the assumption of excess summer storage for use in the winter (i.e., primarily to demonstrate the
overall cost impact associated with the summer drying requirements in isolation).

Out of the various coproduct credit inclusions, the primary coproduct with the largest impact on
MFSP is the CO2 recycle credit. If this credit were eliminated, MFSP would increase by roughly
$0.14/GGE. As discussed above it is justified to include this credit as it was not counted in the
original $430/ton biomass cost basis, but further analysis is required to better understand the
most optimal case for recycling the low-concentration CO2 in the turbine flue gas. Feed solids
content exhibits a noticeable, though not drastic, impact on economics over a range from 15-
25%, thus suggesting that there may be room for further upstream optimization to reduce
dewatering costs from the 20% solids basis assumed here. The variation of combined
carbohydrate-to-ethanol yield across pretreatment and fermentation also has a considerable
impact on MFSP. The remaining sensitivity variables each account for less than $0.10 change
in MFSP, which are small in relation to the feedstock parameters given the dramatic contribution
to overall MFSP that feedstock carries (roughly 70% of total MFSP as discussed above).
However, taken in combination, these additional parameters are also non-trivial as they
ultimately may impact overall process economics.

63
 Figure 16. Tornado plot presenting results of the single-point sensitivity analysis on MFSP cost
($4.35/GGE reference case)

5.2 Alternative Composition and Processing Scenarios

While the high-lipid Scenedesmus (HLSD) strain with excess summer drying and storage was
selected as the base case for the model, here we present the results when instead considering the
alternative biomass composition basis (high-carbohydrate Scenedesmus HCSD, Section 2.2)
and/or the alternative processing scheme where raw seasonal feed rates from upstream
cultivation are fed directly to the conversion facility without diverting, drying, and storing excess
summer capacity for use in the winter (Section 2.1). Results for both MFSP and total fuel yield
for all four scenarios are shown in Figure 17. Similar to the HLSD + Store base case, all other
scenarios are based on individual Aspen models for all four seasons, setting maximum design
capacity and associated capital and fixed operating costs according to maximum seasonal
throughput, and taking the season-average results for variable operating costs and
product/coproduct yields.

64
 $6 160
141 GGE/ton $5.24
$5.04 140

Total Fuel Yield (GGE/ton AFDW)

$5
$4.59
$4.35 116 GGE/ton 120
$4
MFSP ($/GGE)

100

$3 80 MFSP, $/GGE
60 Yield, GGE/ton
$2
40
$1
20

$0 0
HLSD + Store HLSD No HCSD + Store HCSD No
Store Store

Figure 17. TEA modeling results for alternative composition and seasonal processing scenarios.
(See Sections 2.1-2.2 for definition of scenarios; base case = HLSD + Store.)

As shown in Figure 17, eliminating the summer excess storage approach (and associated dryer
capital cost and natural gas operating cost) in favor of sending the algal feed straight to the
pretreatment reactor at higher summer versus winter variance results in an overall increase in
MFSP on the order of 6% for the HLSD case and 4% for the HCSD case. This penalty is due to
less efficient utilization of installed capital costs for three out of four seasons in the No Store
case, with strong under-utilization of capital in the winter season at 80% throughput turndown
versus 20% in the base case including summer storage (assuming that the equipment as installed
would still operate effectively at such a high turndown). However, the no store alternative
scenario also would improve overall system sustainability by elimination of the drying natural
gas demand in the summer months (see Section 5.3 for the comparison in sustainability metrics).
As noted previously, if sustainability as measured by greenhouse gas profile for the overall
integrated process (including upstream cultivation and dewatering) proves to be too detrimental
in the summer storage approach during subsequent LCA evaluation of the full system,
switching to the no storage approach may still be a viable option, as it does not substantially
increase cost. However, this would require further consideration for equipment operability to
ensure the unit operations that are not operating with several units in parallel would be able to
accommodate such high turndown ratios. It is also likely that the unfavorable winter effect would
be less detrimental in select locations of the U.S., as well as equatorial regions outside the U.S.

Alternatively, modeling for the HCSD biomass composition basis resulted in a more substantial
increase in MFSP relative to the HLSD basis, at a 14-16% cost penalty. This is primarily due to
the associated overall fuel energy yield (ethanol + RDB, GGE/ton) decreasing by roughly 18%,
given the lower energy content of ethanol relative to RDB, which increases to roughly 50% of
total production by volume in the HCSD basis versus roughly 35% in the HLSD case. This will
be a critical point to consider moving forward in understanding requirements to ultimately
achieve economic viability, as the HCSD basis represents an earlier harvest point for the biomass
than the HLSD basis, and thereby a higher biomass productivity rate (g/m2/day) in cultivation.

65
An ultimately achievable combination of biomass cultivation productivity and lipid content may
in fact lie between these two scenarios, resulting in both yields and MFSP costs within the range
bounded by these points. Moving forward, when reconsidering a fully integrated model,
tradeoffs between cultivation productivity and composition will be important to understand and
quantify. This is a key issue that the recently-formed ATP3 Consortium (of which NREL is a
partner) is working to address, namely to provide publicly available data focused on productivity
rates, compositional analysis, and cultivation characteristics for algal pond test-bed sites located
across the U.S., based on long-term cultivation trials across all seasons using consistent methods
[80]. The resulting data will be highly valuable in validating and improving models for algal
biofuels, given a continued lack of such data in the public domain at sufficient scale and
duration.

5.3 Sustainability Metrics for Base Model

An important aspect of evaluating biofuel processes is the quantification of life-cycle resource
consumption and environmental emissions. Life-cycle assessment (LCA) provides a framework
from which the environmental sustainability of a given process may be quantified and assessed.
This section presents the salient sustainability metrics of the current conceptual process at the
conversion stage. Although sustainability metrics are presented here in keeping with other recent
design report practices for cellulosic feedstock pathways [6], it is imperative to note that the
results presented here are strictly confined to the conversion steps of the process, which is
otherwise inseparably integrated from upstream cultivation and dewatering steps. Although it is
reasonable to consider sustainability metrics for biorefinery pathways processing terrestrial
cellulosic feedstocks, algal processes are more inherently integrated between feedstock
production and logistics (cultivation and dewatering) operations and downstream conversion
steps, including numerous recycles and heat/power integration between the process boundaries.
As such, presentation of sustainability metrics attributed to only the conversion stage for algal
processes do not carry as much meaning as for a fully integrated facility and carry a danger of
being misinterpreted. A complete LCA of the fully integrated process, incorporating the details
from this model as well as pertinent upstream operations, is required to better understand and
quantify the complete sustainability profile of this technology pathway.

Direct biorefinery emissions (i.e., CO2, NO2 and SO2), water consumption, and other process-
related metrics were derived from the conversion process model described above. SimaPro
v.8.0.2 software [81] was used to develop and link units quantifying life-cycle impacts as
previously documented by Hsu et al. [82]. GHG basis values for natural gas, purchased hydrogen
(derived via natural gas steam methane reforming) and the U.S. average electricity grid mix were
applied consistently with the basis from Argonne National Laboratorys GREET model software
[83]. Ecoinvent v.2.0 [84] and the U.S. Life Cycle Inventory (LCI) [85] processes were used to
fill the data gaps. The Ecoinvent processes were modified to reflect U.S. conditions and the U.S.
LCI processes were adapted to account for embodied emissions and fossil energy usage. The
material and energy flows of the conversion step capture the impacts of input raw materials, and
outputs, such as emissions, wastes, and coproducts predicted by the process model, as shown in
Table 24. This LCI table and subsequent discussion of sustainability metrics considers both the
base case (including summer excess capacity drying and storage) and the alternative case without
summer storage, both for the HLSD biomass basis. Additionally, the LCI data presented in Table
24 is translated to an average hourly basis, i.e., dividing the annual rates by the operating factor

66
of 7,920 hours/year; the hourly values shown here (such as GGE/hr and kg/hr) do not reflect
actual hourly rates for any season of the year, but rather are the average of the hourly rates across
all four seasons.

Table 24. Input and Output Data for Modeled Conversion Facility (HLSD Biomass Basis).
(Note: Hourly rates shown below are based on annual averages over all modeled seasons.)

Summer NG drying and storage scenario: Store No Store

Products
RDB, GGE/hr 5,840 5,822
Ethanol, GGE/hr 2,034 2,027
Naphtha, gal/hr 130 130
Electricity coproduct, KW 6,937 6,860
Resource Consumption, kg/hr
Feedstock (AFDW basis) 50,611 50,611
Hexane makeup 910 908
Acid (pretreatment) 2,231 2,224
Ammonia (pretreatment) 721 718
DAP (ethanol fermentation) 103 103
Supplemental natural gas (utility) 859 854
Natural gas for summer drying 2,028 NA
Hydrogen 501 499
Process water demands 78,155 77,901
Phosphoric acid (lipid purification) 39 39
Silica (lipid purification) 21 21
Clay (lipid purification) 41 41
Output Streams, kg/hr
AD digestate cake (dry basis total flow) 8,325 8,300
AD digestate cake bioavailable N 65 65
AD effluent NH3 678 676
AD effluent DAP 295 294
Recycle water excluding N/P nutrients 233,528 232,836
Air Emissions, kg/hr
Water (H2O) 13,297 13,252
VOC (primarily ethanol) 2 2
Ammonia (NH3) 40 40
Oxygen (O2) 47,900 48,116
Nitrogen (N2) 234,877 235,323
Nitric Oxide (NO) 238 234
Nitrogen dioxide (NO2) 7 7
Carbon monoxide (CO) 2.E-02 2.E-02
Carbon dioxide (CO2), biogenic 25,999 28,331
Carbon dioxide (CO2), fossil 10,707 2,712
Methane (CH4) Trace Trace
Sulfur dioxide (SO2) 199 198

67
In addition to the primary fuel products, the process also produces an excess amount of
electricity as well as coproduct naphtha, AD effluent nitrogen and phosphorus (modeled as
equivalent NH3 and DAP), AD digestate cake, and flue gas/fermentor vent CO2 for recycle to
algal cultivation. It is assumed that the excess electricity would be sold to the grid translating to
coproduct credits for avoided GHG emissions and fossil energy consumption. This exported
electricity is treated as an avoided product using the product displacement method [86].
Coproduct displacement (also termed system boundary expansion) is based on the concept of
displacing the existing product with the new product. The excess electricity coproduct displaces
an equivalent amount of grid electricity, thus avoiding a significant amount of GHG emissions as
well as fossil energy consumption, assuming an average U.S. electricity grid mixture. The GHG
and fossil energy consumption credits attributed to the displacement of an average U.S.
electricity grid mixture is 0.65 kg CO2-equiv/KWh and 7.46 MJ/KWh, respectively, as defined by
GREET [83]. Similar to exported electricity, all other coproducts are also treated as avoided
products using the product displacement method. The naphtha displacement coproduct credit is
0.46 kg CO2-equiv/kg for GHG and 52.6 MJ/kg for fossil energy consumption. Additionally,
pertinent GHG emission and fossil energy consumption credits are also applied to the AD
effluent stream that contains nitrogen and phosphorus as NH3 and DAP-equivalents (also
accounting for nitrogen in DAP). AD digestate nitrogen credits are estimated based on pertinent
information in the Ecoinvent database, in which 1 kg of bioavailable N equates to 6.45 kg
calcium nitrate at 15.5% N content. The unit process inventory takes into account the use of
energy resources needed for the production of calcium nitrate as a byproduct from the
manufacture of NPK fertilizers. The corresponding GHG emission factor and fossil energy
consumption for 1 kg of bioavailable N in the AD digestate cake is determined to be 4.21 kg
CO2-equiv and 64.5 MJ, respectively.

Details of contributions to GHG emissions and fossil energy consumption at the conversion stage
are presented in Table 25, which corresponds to the information in Table 24. Fossil GHG
emissions associated with the conversion stage for the cases with and without summer natural
gas (NG) drying and storage are determined to be 2.87 kg CO2-eq/GGE and 0.93 kg CO2-eq/GGE,
respectively. The difference is largely attributed to the additional NG consumption and fossil
CO2 emissions associated with feedstock drying. Direct CO2 emissions from the conversion
process (shown in Table 24) consist of both fossil and biogenic CO2 (i.e., CO2 absorbed from the
atmosphere and incorporated as algal biomass during the feedstock cultivation step). With its
biomass origin, biogenic CO2 does not contribute to the increase of GHG in the atmosphere [87]
and is not considered in the IPCC global warming methodology [88]. Hence, the contributions to
GHG at the conversion stage are solely from facility fossil CO2 emissions, as well as the fossil
GHG profile associated with the underlying LCI processes (e.g., material inputs/outputs to and
from the facility to support process operations). Fossil CO2 emissions shown in Tables 24 and 25
are attributed to combustion of natural gas and volatilized hexane in the turbine flue gas (via co-
firing the AD biogas with supplemental natural gas for facility heat requirements) and
combustion of natural gas in the summer drying scenario.

It is evident that natural gas, hydrogen, and ammonia consumption, as well as the facility direct
fossil CO2 emissions, contribute the most to the overall GHG emissions estimate. The total GHG
coproduct credit attributed to electricity, naphtha, and AD product outputs is determined to be
negative 0.93 kg CO2-eq/GGE. Table 25 also shows the fossil energy consumption for the

68
conversion process. Fossil energy consumption associated with the conversion stage for the cases
with and without summer NG drying and storage are estimated to be 29.9 MJ/GGE and 15.0
MJ/GGE, respectively. Similar to the GHG emissions, the difference in total fossil energy
consumption for both cases is largely attributed to the additional natural gas consumption
associated with feedstock drying. The total fossil energy credits attributed to exported power,
naphtha, and AD coproducts are comparable for both cases, at roughly negative 15.2 MJ/GGE.

Table 25. Conversion Process GHG Emissions and Fossil Energy Consumption per GGE Total
Fuel Product (RDB + Ethanol)

GHG Emission Fossil Energy Input

kg CO2-eq/GGE MJ/GGE
Summer NG drying and storage scenario: Store No Store Store No Store
Direct refinery emissions 1.36 0.35 0.00 0.00
Hexane 0.12 0.12 7.05 7.05
Sulfuric acid 0.02 0.02 0.33 0.33
Ammonia 0.20 0.20 3.81 3.81
Diammonium phosphate 0.04 0.04 0.75 0.75
Supplemental natural gas utility 0.39 0.39 6.30 6.29
Natural gas for summer drying 0.93 0.00 14.89 0.00
Hydrogen 0.71 0.71 11.64 11.64
Lipid purification chemicals 0.01 0.01 0.12 0.12
Biorefinery infrastructure 0.02 0.02 0.23 0.23
Electricity credits -0.58 -0.57 -6.57 -6.52
Co-product naphtha credits -0.02 -0.02 -2.39 -2.39
Co-product AD credits -0.34 -0.34 -6.26 -6.26
Total 2.87 0.93 29.90 15.05

Table 26 summarizes the key sustainability metrics for the conversion process for the cases with
and without summer natural gas drying and storage. On an energy basis, the GHG emissions at
the conversion stage for the cases with and without summer drying are 32.2 kg CO2-eq/GJ and
10.4 kg CO2-eq/GJ, respectively. Similarly, the fossil energy consumption for the cases with and
without summer drying is 0.33 MJ/MJ and 0.17 MJ/MJ, respectively.

In addition to GHG emissions and fossil energy consumption discussed above, water
consumption (i.e., net water use during the biorefinery operation), total fuel efficiency (yield),
and carbon-to-fuel efficiency are also reported in Table 26. Biorefinery net water consumption
includes, but is not limited to, water that is incorporated into the product, and evaporation
directly from process operations or indirectly from cooling and heating processes (e.g., cooling
tower evaporative losses), but it does not include water that is directly returned to surface water
or groundwater resources. Aside from the water delivered to the conversion process with the
high-moisture feedstock, facility process water consumption demands equate to 1,876 m3/day as
an average across all seasons, allocated to ethanol vent scrubber water (14%), wash water for
lipid purification (3%), boiler water for live steam injection to pretreatment (31%), boiler
makeup water to make up for blowdown losses (3%), and cooling tower makeup water (49%).
Note that this value is prior to accounting for water evaporated during summer drying and
consequently required to re-constitute the material for feed in the winter, which adds an

69
additional 687 m3/day when allocated evenly across a full year, for a total process water
consumption of 2,563 m3/day. Compared to this process water demand, total water rate in the
recycle pool, which is destined for recycle back to the upstream cultivation ponds, is
approximately threefold higher, at 5,539 m3/day averaged across all seasonal models, due to
much of the feed water ultimately ending up in this recycle stream. Consequently, the net water
balance for the conversion process alone is a net negative 2,976 m3/day or 4.2 gal/GGE for the
base case with summer drying; however, this is predicated on the large amount of water entering
into the conversion process (at 80% moisture content) for free when viewed exclusively in the
context of the conversion facility by itself. The true water footprint of the overall process must
also take into account upstream cultivation and dewatering operations, which is where the vast
majority of water losses occur attributed to evaporation from very large open pond surface areas
(although the exact amount is highly dependent on facility location [8]). For a fully integrated
facility including consideration of cultivation evaporative and blowdown water losses, the net
water consumption is likely to be greater than zero, but would just be reduced by the amount as
shown here for the conversion facility negative water balance. For the alternative case without
summer drying, the conversion stage water balance reduces to a net negative 5.1 gal/GGE,
attributed to a process water consumption of 1,876 m3/day.

The total fuel yield is 141.1 GGE per dry ton and the corresponding carbon-to-fuel efficiency is
determined to be 63% (or 64% when also including naphtha), calculated strictly as a ratio of
carbon in the liquid fuel products relative to carbon in the feedstock. When also including non-
algal carbon inputs attributed to natural gas (summer drying and utility heating), hexane makeup,
and hydrogen (sourced from off-site natural gas steam methane reforming at a natural gas
carbon-to-hydrogen molar ratio of 0.44:1 [89]), carbon efficiency decreases to 55% (56%
including naphtha) for the summer storage base case, or 58% for the alternative case without
summer drying and storage. A more detailed accounting of carbon balances is presented below in
Figure 18, tracking seasonal-average carbon flows through the conversion process for the
modeled HLSD + Store base case.

Table 26. Summary of Sustainability Metrics for the Process

Summer NG drying and storage scenario: Store No Store

GHG emissions (kg CO2e/GJ) a 32.2 10.4
Conversion stage process water consumption (m3/day) 2,563 1,876
Conversion stage process water consumption (gal/GGE fuel) 3.6 2.6
Conversion stage net water balance (gal/GGE fuel) b -4.2 -5.1
Total fuel yield (GGE/dry ton) 141.1 141.1
Algal Carbon-to-fuel efficiency (C in fuel/C in biomass) c 63% 63%
Total Carbon-to-fuel efficiency (C in fuel/total input C) d 55% 58%
Net fossil energy consumption (MJ/MJ) a 0.33 0.17
a
GHG and fossil energy values include recycle coproduct credits as documented above
b
Table reports both direct process water consumption (inputs to satisfy facility water demands) as well as
net water balance after also considering total water recycled to upstream growth ponds (counting biomass
feed moisture content as free in the boundaries of this analysis)
c
Algal carbon efficiency defined here as the ratio of carbon in primary fuel products (RDB + ethanol) relative
to carbon in biomass feed; excludes naphtha (extra 1%)
d
Total carbon efficiency defined here as the ratio of carbon in primary fuel products (RDB + ethanol) relative
to total input carbon including natural gas, hydrogen (from off-site natural gas), and hexane; excludes
naphtha (extra 1%)

70
 Figure 18. Carbon balance diagram tracking carbon flows across the modeled process
(HLSD + Store base case, overall seasonal average flows).

5.4 Paths Forward to $3/GGE

The analysis presented here demonstrates a strategy for achieving an overall fuel selling price
near $4.50/GGE, on par with published targets for algal hydrothermal liquefaction processing
[11]. However, additional improvements will be required to further improve economics towards
standard DOE-BETO targets closer to $3/GGE. First and foremost, feedstock cost reductions
would translate to substantial cost savings, as presented above. For example, as shown in Section
5.1, reducing feedstock cost from $430/ton (AFDW) to $300/ton would reduce MFSP by nearly

71
$1/GGE. Such an improvement would require high cultivation productivity or a lower-cost
cultivation system (per acre of cultivation area) relative to the open raceway pond design as
detailed in [8, 55]. Similarly, employing a crop rotation strategy or other mechanism to reduce
seasonal variability, particularly increasing winter season productivity, would lead to substantial
improvements in economics; to a more limited extent in the conversion back-end (as seen by
relatively marginal cost differences between the summer store and no store scenarios), but
considerably more in the cultivation front-end, given the high costs for cultivation and necessity
to maintain efficient utilization of installed cultivation capacity. This is also a critical point in
improving sustainability, namely GHG emissions, of integrated algal biofuel processes [8, 10].

Beyond further improvements in biomass production, on the conversion side, as discussed

previously a key advantage of the process described here is its flexibility in allowing for a plug
and play processing approach for conversion of the carbohydrate, lipid, and protein fractions
individually. Given that the process selectively targets solubilization/extraction and conversion
of each of these biomass constituents rather than conversion of the whole biomass feedstock, a
large number of options may be pursued to alternative products or coproducts. For example,
substantial research efforts are underway to pursue biological or catalytic conversion routes of
sugars to hydrocarbon fuel precursors, or to high-value chemicals [6, 90]. This could lead to
either increased energy (GGE) fuel yield or high coproduct revenues, respectively. One example
of precedent is NRELs 2013 biochemical cellulosic design report, which describes various
coproduct pathway options for conversion of lignin components to value-added chemicals,
including 1,3 butadiene, 1,4 butanediol, adipic acid, and cyclohexane, with a number of these
options providing substantial coproduct revenue to assist in reducing MFSP towards $3/GGE
targets [6]. Coproduct options from oil constituents are possible as well, with Cellana proving a
noteworthy example whose business model in the near term focuses on garnering significant
revenue from isolation of omega-3 fatty acid oils and selling for a substantial fraction of total
product sales [91].

Similarly, a number of alternative options exist for utilization of the protein-rich residue. Rather
than AD, the protein-enriched material may instead be potentially used for animal or aquaculture
feed markets (if sufficiently high-value, e.g., >$350/tonne [8]), bioplastics [92], or a novel route
described by Huo et al. [93] and subsequently investigated by SABC, Sandia National
Laboratory, and others, in which the protein is hydrolyzed and fermented to butanol and higher
alcohols, further increasing total fuel yield potential beyond that considered here. These options
were briefly considered in earlier TEA feasibility studies and were not found to offer substantial
benefits to justify their use in the present analysis, however they could become more beneficial
for higher-protein feedstocks such as the early-harvest Scenedesmus biomass scenario presented
in Table 1. Such options will continue to be investigated moving forward. In any case, the
reliance on coproducts requires careful analysis and understanding of the materials market
volume, to avoid saturating a given coproduct market with one or several biofuel facilities, given
the very large commodity fuel market into which the biofuel industry will expand. Additionally,
similar to the discussion in the algal HTL design report [11], off-site upgrading of the extracted
algal oil at a central location or in concert with petroleum refinery blending may also present
further opportunities for cost reduction.

72
6 Concluding Remarks
6.1 Summary
The present report establishes a plausible case for achieving a modeled cost goal under $5/GGE
of upgraded renewable diesel blendstock and ethanol, coproduced via biochemical fractionation
of algal biomass and selective conversion of carbohydrate and lipid constituents. The pathway is
based on a single set of process R&D metrics and associated cost estimates for the base case,
realizing that process alternatives exist, both in terms of sequence order of operations as well as
different operations altogether to convert carbohydrates, lipids, and even proteins to alternative
fuels or products. The pathway discussed here assumes biomass is fed at 20 wt% solids to a
dilute acid pretreatment/fractionation step, the resulting whole hydrolysate slurry is sent through
a fermentation step to convert solubilized sugars to ethanol, and the resulting whole fermentation
stillage is processed in a wet extraction operation with remaining high-protein residue routed to
anaerobic digestion for purposes of carbon and nutrient reclamation. The integrated process
assumes high solubilization/recovery of both carbohydrates and lipids, and subsequent high
conversion of each of these components to ethanol and RDB respectively. Some operations,
primarily pretreatment and fermentation, build off of NRELs well-established operational
experience and carry a reasonably high level of certainty in ultimately achieving their targeted
performance metrics. Other units, such as lipid extraction and upgrading, are in an earlier stage
of understanding based on in-house experimental data, but early results have been promising in
proving out proof-of-concept and planning for follow-on experimental efforts. More open data
flow between NREL and commercial practitioners of these technologies would greatly improve
our understanding of the current state of technology.

With the combination of process conversion targets stipulated in this design (primarily 90%
sugar solubilization in pretreatment, 95% conversion of sugar-to-ethanol after first assuming
fractional sugar losses to fermentation organism propagation and contamination, 95% lipid
extraction efficiency, and 80 wt% yield of RDB from purified lipids across hydrotreating), the
resulting overall fuel yield was found to be 92.5% of theoretical based on the given algal
biomass composition. Thus, while the process scenario allows for achieving MFSP costs below
$5/GGE, relatively little room exists for further increasing yields from the carbohydrate or lipid
fractions. Thus, additional improvements elsewhere will be required to ultimately reach a
$3/GGE goal. These include (1) further optimization on the front-end, primarily achieving high
algal productivity rates and/or identifying fundamentally lower-cost cultivation systems, (2)
reducing seasonal variability by employing different strains in the winter or other
engineering/design measures, (3) considering off-site centralized lipid upgrading options, or (4)
pursuing higher-value coproduct options likely based either on the protein or carbohydrate
fractions. The current technology pathway discussed here is readily amenable to such alternative
options, and possesses a high degree of plug and play flexibility in pursuing other technology
options for any given biomass component.

The end result of the techno-economic analysis was a predicted minimum fuel selling price
(MFSP) of $4.35/GGE (2011-dollars) at a final upgraded product yield of 141.1 GGE/dry ton of
biomass (ash-free dry weight basis) in the base case. This reflects a $3.05/GGE contribution
from feedstock and a $1.30/GGE contribution from the conversion process. This modeled selling
price is strictly representative of nth-plant assumptions regarding biorefinery design, operation,
and financing, and is not intended to reflect first-of-a-kind or early-entry commercial facilities.

73
The total fuel yields presented here are similar to fuel yield targets for HTL conversion of algal
biomass, and in both cases are distinctly higher than terrestrial cellulosic pathways. This
particular technology pathway achieves these yields via selective conversion of specific
components to specific fuel products under relatively low-pressure/temperature operating
conditions, however requires mid- to high-lipid content biomass to achieve these yields, which
adds a burden to upstream cultivation performance and requires carefully balancing achievable
biomass productivity rates against resulting biomass composition and fuel potential.

The base case considered here assumes processing of a late-harvest, high-lipid algal strain with
diversion, drying, and storage of excess summer capacity for later use in the winter, to minimize
seasonal fluctuations inherent to algal cultivation and thereby fluctuations in equipment
processing capacity. However, the analysis also considered, in detail, a process alternative of
sending the dewatered material straight to the conversion process without the drying step,
substantially increasing the feed rate and downstream throughput variability from a range of
1.2:1 (summer versus winter) to 5.4:1. While maintaining essentially the same overall fuel yield,
the alternative processing scenario increased MFSP costs by roughly 5%, due to a higher penalty
in under-utilization of installed capital cost relative to a smaller savings in eliminating summer
drying equipment and natural gas costs. Alternatively, switching to an earlier-stage harvest
biomass composition with high carbohydrate and lower lipid content increased MFSP costs by
14%-16%, attributed to a concomitant GGE yield reduction of 18%. This implies that overall
cost targets for an integrated process would be more difficult to achieve at the earlier-harvest
biomass composition scenario; however, this must also be considered in reference to the ability
to achieve a given algal cultivation productivity rate and thereby algal biomass cost, as well as
our decision to valorize sugar as a biofuel rather than as a higher value bio-based chemical. In
either case, this analysis sets the stage for continued consideration of these tradeoffs for a fully
integrated process including the front-end cultivation and harvesting steps. It may be likely that a
realistic optimum exists between the two scenarios considered here, i.e. a mid-to-late stage
harvest with a resulting lipid content above 27% but below 41% while maintaining reasonable
cultivation productivity rates by year 2022 and beyond.

In addition to providing an economic analysis, the present report also considers sustainability
metrics for the baseline processing scenario with summer storage, as well as for the alternative
without summer storage, by tracking and reporting on greenhouse gas emissions, fossil energy
demand, and direct consumptive water use at the modeled biorefinery facility (conversion stage
only). The sustainability analysis for the baseline scenario indicates a fossil GHG profile of 2.87
kg CO2-eq/GGE, fossil energy consumption of 29.9 MJ/GGE, and water demand of net negative
4.2 gal/GGE (associated with high moisture content of the incoming feedstock). The alternative
scenario without summer excess capacity drying results in GHG emissions, fossil energy
consumption, and water demand of 0.93 CO2-eq/GGE, 15.1 MJ/GGE, and net negative 5.1
gal/GGE respectively. Thus, the results indicate conflicting optima between economics (favoring
summer storage) and sustainability (favoring no summer storage). However, as these values are
only representative of a partial process focused on only the conversion operations in an otherwise
inherently integrated process, the sustainability metrics values carry somewhat limited meaning
and a full TEA and LCA is required for the overall well-to-wheel integrated process including
upstream biomass production and harvesting operations, in order to present a more complete
understanding of such tradeoffs. This will be a subject of continued modeling consideration
moving forward.

74
It is worthwhile to reiterate that algal biofuels in general, and the proposed fractionation and
conversion process discussed here for processing algal biomass in particular, are in an early stage
of development and understanding, relative to other cellulosic technology pathways as well as
HTL conversion technologies (whether for algal biofuels or otherwise). Thus, the absolute cost
values established here, as well as the timeframe required to achieve these outcomes, inherently
carries a somewhat higher degree of uncertainty given the nascent stage of research as presented
in the public domain and conducted to date at NREL. This said, the economic potential as well as
distinct technology advantages enabled by this approach make a strong case for continued
consideration as a means to achieving a viable algal biofuels industry.

6.2 Future Work

Moving forward, to ultimately achieve cost goals and reduce uncertainty in key areas for the
algal fractionation conversion pathway in the context of a fully integrated process with upstream
cultivation, a number of important bottlenecks, uncertainties, and areas for further development
are summarized below. Some of these points are repeated from NRELs recent technical memo
for the algal lipid upgrading (ALU) process [94]:

Feedstock production and logistics:

o Validate algae growth rates and biomass compositional analysis based on
data from large scale demonstrations: In order to ultimately be able to set
realistic, achievable goals related to algal cultivation, as well as biomass quality
(i.e., composition), a sound understanding of the initial baseline is first required.
While NREL and other partners have done substantial analysis on this subject [8,
10], these two key parameters still largely remain limited to modeled or semi-
empirical data. Among a number of entities working to address this issue, the
recently-formed ATP3 and RAFT programs are focused on this goal as a critical
priority, working to provide publicly available data for cultivation test-bed sites
located in regions across the US. Data made available from these efforts will be
leveraged in NRELs models to better quantify todays state of technology
costs.
o Reduce cost and increase performance for cultivation and dewatering: To
ultimately realize economic viability for algal-derived biofuels regardless of the
technology, substantial improvements in both performance (e.g. cultivation
productivity, harvesting efficiency) and cost (e.g. alternative or lower-cost
designs, removing pond liners) will be required. This will depend to a large extent
on both locality and strain specifics, with high cost sensitivities to seasonal
variability of the selected facility location. Additionally, continued attention must
be afforded to protection of crop yield against culture crashes (losses to grazers
and pathogens) through improving strain robustness and systems design [95].
Conversion operations:
o Confirm efficacy of integrated process: While promising experimental data
have proven the efficacy of the process concepts when conducting the key
conversion steps on separated solids and liquids from pretreatment [13], NREL
has not yet considered an experimental approach consistent with the extrapolated
basis used for modeling purposes, e.g., fermentation of the whole pretreatment

75
 hydrolysate followed by lipid extraction of the whole fermentation stillage. This
approach is anticipated to be equally viable, although may present additional
challenges that must be overcome by continued research to demonstrate the high
yield targets stipulated in the model for an integrated process. Further process
intensification to target increased fermentation productivity and decreased capital
costs may also contribute to economic improvements.
o Further characterize products/intermediates: NREL has a reputable history of
algal biomass characterization capabilities and the biomass component
characterizations employed here are sound; however, less work has been done to
date on characterizing the conversion products. This is particularly of interest for
the extracted algal oil product, as it impacts subsequent requirements and costs
both for lipid cleanup and upgrading.
o Demonstrate upgrading of algal oil: The assumptions employed in the model
described here are based on operating and yield assumptions from literature, for
general fatty acid triglyceride materials. The model will incorporate experimental
data as it becomes available from new NREL efforts focused on algal oil
upgrading, which will better define specific requirements for lipid purification, as
well as conversion both to paraffinic blendstocks produced from deoxygenation
and to finished fuels produced from hydroisomerization.
Coproduct utilization:
o Confirm AD performance: The assumptions currently employed here for the
anaerobic digestion (AD) operation, specifically metrics of biogas yield and
nutrient fixation in the recycle effluent, are based on a compilation of literature
studies, which have largely been validated by NREL experimental work when
considering AD of lipid-extracted biomass. However, for biomass that is also
largely devoid of carbohydrates, as in the process modeled here, additional
operational challenges may be presented for the AD unit which are not yet well-
understood. Further analysis is required to confirm the assumptions utilized here.
o Evaluate alternative coproduct opportunities: Many coproduct opportunities
have yet to be fully explored for economic and LCA trade-offs and should further
be considered going forward. These include options for both alternative
carbohydrate utilization scenarios to hydrocarbon fuels or chemicals (balanced by
additional processing costs), as well as protein utilization options such as high-
value animal or fish feed materials, bioplastics, or protein hydrolysis and
fermentation to alcohols. Other high-value coproduct options may also be
plausible; however, any coproduct strategies must be carefully considered relative
to market volumes given the large-scale commodity fuel market that this process
is intended to contribute to. Additionally, direct use of wastewater as a water
and/or nutrient source may also provide additional economic benefits by way of
wastewater treatment cost offsets.

76
 References
NREL milestone reports cited below cannot be accessed outside of NREL and DOE. Readers
may contact the authors of the specific reference to determine if this information has been made
public since publication of this design report.

1. Wooley, R.J., et al., Lignocellulosic Biomass to Ethanol Process Design and Economics
Utilizing Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis Current and
Futuristic Scenarios. NREL/TP-580-26157. Golden, CO: National Renewable Energy
Laboratory, 1999.
2. Aden, A., et al., Lignocellulosic Biomass to Ethanol Process Design and Economics Utilizing
Co-Current Dilute Acid Prehydrolysis and Enzymatic Hydrolysis for Corn Stover. NREL/TP-
510-32438. Golden, CO: National Renewable Energy Laboratory, June
2002.http://www.nrel.gov/docs/fy02osti/32438.pdf.
3. Humbird, D., et al. Process Design and Economics for Biochemical Conversion of
Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic Hydrolysis of
Corn Stover. NREL/TP-5100-47764. Golden, CO: National Renewable Energy Laboratory,
2011. http://www.nrel.gov/docs/fy11osti/47764.pdf.
4. Phillips, S., et al. Thermochemical Ethanol via Indirect Gasification and mixed Alcohol
Synthesis of Lignocellulosic Biomass. NREL/TP-510-41168. Golden, CO: National Renewable
Energy Laboratory, 2007. http://www.nrel.gov/docs/fy07osti/41168.pdf.
5. Dutta, A., et al. Process Design and Economics for Conversion of Lignocellulosic Biomass to
Ethanol. NREL/TP-5100-51400. Golden, CO: National Renewable Energy Laboratory, 2011.
http://www.nrel.gov/biomass/pdfs/51400.pdf.
6. Davis, R., et al. Process Design and Economics for the Conversion of Lignocellulosic Biomass
to Hydrocarbons: Dilute-Acid and Enzymatic Deconstruction of Biomass to Sugars and
Biological Conversion of Sugars to Hydrocarbons. NREL/TP-5100-60223. Golden, CO:
National Renewable Energy Laboratory, 2013. http://www.nrel.gov/docs/fy14osti/60223.pdf.
7. DOE. National algal biofuels technology roadmap. Rep. DOE/EE0332. Washington, D.C.:
Biomass Program, of Energy Efficiency and Renewable Energy, 2010.
http://www1.eere.energy.gov/bioenergy/pdfs/algal_biofuels_roadmap.pdf.
8. Davis, R., et al. Renewable Diesel from Algal Lipids: An Integrated Baseline for Cost,
Emissions, and Resource Potential from a Harmonized Model. NREL/TP-5100-55431. Golden,
CO: National Renewable Energy Laboratory, 2012.
http://www.nrel.gov/docs/fy12osti/55431.pdf.
9. DOE. Biomass Multi-year Program Plan (MYPP)-May 2013. Washington, D.C.: U.S. DOE,
Bioenergy Technologies Office, Editor, 2013.
http://www1.eere.energy.gov/biomass/pdfs/mypp_may_2013.pdf.
10. Davis, R., et al. Integrated Evaluation of Cost, Emissions, and Resource Potential for Algal
Biofuels at the National Level. Environ. Sci. Technol. (48), 2014; pp. 6035-6042.
11. Jones, S., et al. Process Design and Economics for the Conversion of Algal Biomass to
Hydrocarbons: Whole Algae Hydrothermal Liquefaction and Upgrading. PNNL-23227.
Richland, WA: Pacific Northwest National Laboratory, 2014.
http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-23227.pdf.
12. Osterwalder, N.; Sajkowski, D. Production and Upgrading of Renewable Bio-Crude from
Large Scale Algae Ponds. Proceedings of the 2013 AIChE Spring Meeting and 9th Global
Congress on Process Safety. San Antonio, Texas. AIChE, 2013; 138c.
http://www3.aiche.org/proceedings/Abstract.aspx?PaperID=295693.

77
13. Laurens, L., et al. Acid-catalyzed algal biomass pretreatment for integrated lipid and
carbohydrate-based biofuels production Green Chem. (17), 2015; pp. 1145-1158.
14. Dirks, G.; McGowen, J.; Pienkos, P.T. Sustainable Algal Biofuels Consortium. in 2013 DOE
Bioenergy Technologies Office (BETO) Project Peer Review; 2013, Alexandria, VA.
https://www2.eere.energy.gov/biomass/peer_review2013/Portal/Algae/#.
15. AspenPlus2007, Vers. 7.2. Aspen Technology Inc., Cambridge MA. http://www.aspentech.com.
16. Wooley, R.J.; Putsche, V.; Ibsen, K. Development of an ASPEN PLUS Physical Property
Database for Biofuels Components. 1999; p. 18.
17. Wooley, R.J.; Putsche, V. Development of an ASPEN PLUS Physical Property Database for
Biofuels Components. NREL/MP-425-20685. Golden, CO: National Renewable Energy
Laboratory, 1996. http://biodev.nrel.gov/pdfs/3955.pdf.
18. Wigmosta, M.S., et al. National Microalgae Biofuel Production Potential and Resource
Demand. Water Resources Research (47:3), 2011; pp. W00H04.
19. Venteris, E.R., et al. A GIS Cost Model to Assess the Availability of Freshwater, Seawater, and
Saline Groundwater for Algal Biofuel Production in the United States. Environmental Science
& Technology (47:9), 2013; pp. 4840-4849.
20. Huesemann, M. H., et al. A screening model to predict microalgae biomass growth in
photobioreactors and raceway ponds. Biotechnol. and Bioeng (110:6), 2013; pp. 1583-1594.
21. Laurens, L., et al. Strain, Biochemistry and Cultivation-Dependent Measurement Variability of
Algal Biomass Composition. Analytical Biochemistry (452), 2014; pp. 86-95.
22. Loureno, S.O., et al. Distribution of intracellular nitrogen in marine microalgae: Calculation
of new nitrogen-to-protein conversion factors. Eur. J. Phycol. (39:1), 2004; pp. 1732.
23. Wang, L., et al. A flexible culture process for production of the green microalga Scenedesmus
dimorphus rich in protein, carbohydrate, or lipid. Bioresour. Technol. (129), 2013; pp. 289
295.
24. Zemke, P.E.; Sommerfeld, M.R.; Hu, Q. Assessment of key biological and engineering design
parameters for production of Chlorella zofingiensis (Chlorophyceae) in outdoor
photobioreactors. Appl. Microbiol. Biotechnol. (97:12), 2013; pp. 56455655.
25. Lu, S., et al. Phospholipid Metabolism in an Industry Microalga Chlorella sorokiniana: The
Impact of Inoculum Sizes. PLoS ONE (8:8), 2013; pp. e70827.
26. Kendel, M., et al. Non-methylene Interrupted and Hydroxy Fatty Acids in Polar Lipids of the
Algae Grateloupia turuturu Over the Four Seasons. Lipids (48:5), 2013; pp. 535-545.
27. Laurens, L., et al. Accurate and reliable quantification of total microalgal fuel potential as fatty
acid methyl esters by in situ transesterification. Anal. Bioanal. Chem. (403), 2012; pp.167178.
28. Jones, S., et al. Process Design and Economics for the Conversion of Lignocellulosic Biomass to
Hydrocarbon Fuels: Fast Pyrolysis and Hydrotreating Bio-Oil Pathway. PNNL-23053,
NREL/TP-5100-61178. Richland, WA: Pacific Northwest National Laboratory, 2013.
http://www.pnnl.gov/main/publications/external/technical_reports/PNNL-23053.pdf.
29. Lower and higher heating values of hydrogen and fuels. DOE Hydrogen Analysis Resource
Center, 2012.
http://hydrogen.pnl.gov/cocoon/morf/hydrogen/site_specific/fuel_heating_calculator.
30. Harun, R.; Danquah, M.K. Influence of acid-pretreatment on microalgal biomass for bioethanol
production. Process Biochem. (46:1), 2011; pp. 304-309.
31. Zhou, N., et al. Hydrolysis of Chlorella Biomass for Fermentable Sugars in the Presence of
HCL and MgCl2. Bioresource Technol. (102:21), 2011; pp. 10158-10161.
32. Nagle, N., et al. Biochemical conversion of lipid and carbohydrate enriched algal species-
baseline performance. NREL/SABC Milestone Report to DOE-BETO, December 2011.
33. Choi, S. P.; Nguyen, M.T.; Sim, S. J. Enzymatic pretreatment of Chlamydomonas reinhardtii
biomass for ethanol production. Bioresource Technol. (101), 2010; pp. 5330-5336.

78
34. Shirai, F., et al. Cultivation of microalgae in the solution from the desalting process of soy
sauce waste treatment and utilization of the algal biomass for ethanol fermentation. World J. of
Microbiology & Biotechnology (14), 1998; pp. 839-842.
35. Amin, S. Review on biofuel oil and gas production process from microalgae. Energ. Convers.
Manage. (50), 2009; pp. 1834-1840.
36. Nagle, N., et al. Final SABC report on project progress including measure of fuel yields using a
biochemical or a combined chemical-biochemical approach, data for inputs into techno-
economic models, and identification of critical elements for future yield improvements.
NREL/SABC Milestone Report to DOE-BETO, May 2013.
37. Nagle, N., et al. Increasing bioconversion of algal biomass through high-value fractionation
and recovery: A novel process. Presented at the 4th International Conference on Algal
Biomass, Biofuel & Bioproducts, Santa Fe NM, 6/19/14.
38. Dutta, A. Syngas Mixed Alcohol Cost Validation. Presented at the DOE 2013
Thermochemical Platform Peer Review.
https://www2.eere.energy.gov/biomass/peer_review2013/Portal/presenters/public/InsecureDow
nload.aspx?filename=0521_1100_Dutta_WBS3.6.1.1.pdf.
39. Davis, R. Biochemical Platform Review. Presented at the DOE 2013 Biochemical Platform
Peer Review.
https://www2.eere.energy.gov/biomass/peer_review2013/Portal/presenters/public/InsecureDow
nload.aspx?filename=Peer Review_Biochemical_NREL_Davis_v2-final.pdf.
40. Tao, L., et al. NREL 2012 Achievement of Ethanol Cost Targets: Biochemical Ethanol
Fermentation via Dilute-Acid Pretreatment and Enzymatic Hydrolysis of Corn Stover.
NREL/TP-5100-61563. Golden, CO: National Renewable Energy Laboratory, 2014.
http://www.nrel.gov/docs/fy14osti/61563.pdf.
41. Davis, R., et al. Biological Conversion of Sugars to Hydrocarbons Technology Pathway.
NREL/TP-5100-58054. Golden, CO: National Renewable Energy Laboratory, 2013.
http://www.nrel.gov/docs/fy13osti/58054.pdf.
42. Biddy, M.; Jones, S. Catalytic Upgrading of Sugars to Hydrocarbons Technology Pathway.
NREL/TP-5100-58055. Golden, CO: National Renewable Energy Laboratory, 2013.
http://www.nrel.gov/docs/fy13osti/58055.pdf.
43. Mohagheghi, A., et al. Improving xylose utilization by recombinant Zymomonas mobilis strain
8 b through adaptation using 2-deoxyglucose. Biotechnology for Biofuels (7), 2014; p. 19.
44. Dutta, A., et al. An economic Comparison of Different Fermentation Configurations to Convert
Corn Stover to Ethanol Using Z. mobilis and Saccharomyces. American Institute of Chemical
Engineers (26:1), 2009; pp. 64-72.
45. Mohaghegi, A., et al. Performance of a newly developed integrant of Zymomonas mobilis for
ethanol production on corn stover hydrolysate. Biotechnol. Lett. (26), 2004; pp. 321-325.
46. Huang, W.D.; Zhang, Y.H.P. Analysis of biofuels production from sugar based on three
criteria: Thermodynamics, bioenergetics, and product separation. Energy & Environmental
Science (4:3), 2011; pp. 784-792.
47. Nagle, N.; Lemke, P. Microalgal Fuel Production Processes: Analysis of Lipid Extraction and
Conversion Methods, Extract from Bollmeier, W. et al., Aquatic Species Program Annual
Report, 1989. http://www.nrel.gov/docs/legosti/old/3579.pdf.
48. Kubiks D.; Horek, J. Deactivation of HDS catalysts in deoxygenation of vegetable oils.
Appl. Catal. (394), 2011; pp. 9-17.
49. Bovornseripatai, P., et al. Effect of Biomass Feedstocks on the Production of Hydrogenated
Biodiesel. World Academy of Science, Engineering and Technology (6), 2012; pp. 299-303.

79
50. Marker, T.L., et al. Opportunities for Biorenewables in Oil Refineries. DOEGO15085.
Washington, D.C.: U.S. DOE, Editor, 2005, UOP LLC.
http://www.osti.gov/bridge/servlets/purl/861458-Wv5uum/861458.pdf.
51. Donnis, B., et al. Hydroprocessing of bio-oils and oxygenates to hydrocarbons. Understanding
the reaction routes. Top. Catal. (52), 2009; pp. 22940.
52. Sotelo-Boyas, R.; Liu, Y.; Minowa, T. Production of green diesel by hydrocracking of canola
oil on NiMo/c-Al2O3 and Pt-Zeolitic based catalysts. 2008. Accessed October 2010:
http://www.nt.ntnu.no/users/skoge/prost/proceedings/aiche-2008/data/papers/P134226.pdf.
53. Berenblyum, A. S., et al. Production of Engine Fuels from Inedible Vegetable Oils and Fats.
Neftekhimiya (50:4), 2010; pp. 317-323.
54. Davis, R.; Aden, A.; Pienkos, P.T. Techno-economic analysis of autotrophic microalgae for
fuel production. Appl. Energ. (88), 2011; pp. 3524-3531.
55. Lundquist, T.J., et al. A Realistic Technology and Engineering Assessment of Algae Biofuel
Production. Energy Biosciences Institute, 2011.
56. Frank, E. D., et al. Life-Cycle Analysis of Algal Lipid Fuels with the GREET Model.
ANL/ESD/11-5. Argonne, IL: Argonne National Laboratory, 2011.
57. Woertz, I.C., et al. Life Cycle GHG Emissions from Microalgal Biodiesel A-CA-GREET
Model. Environmental Science and Technology (48:11), 2014; pp. 6060-6068.
58. Wang, L. K., Biosolids treatment processes. Totowa, N.J.: Humana Press, 2007.
59. Jarvis, E., et al. Efficient use of Algal Biomass Residues for Biopower Production with
Nutrient Recycle. Presented at the DOE 2013 Algae Platform Peer Review.
https://www2.eere.energy.gov/biomass/peer_review2013/Portal/presenters/public/InsecureDow
nload.aspx?filename=Jarvis%20WBS%209-2-2-3%20SLIDES%20vers%203.pptx.
60. Schwede, S., et al. Anaerobic co-digestion of the marine microalga Nannochloropsis salina
with energy crops. Bioresource Technol. (148), 2013; pp. 428-435.
61. Frank, E., Argonne National Laboratory, Personal communication, June 2014.
62. Lundquist, T., Cal Poly University, personal communication, June 2014.
63. Dean, J., et al. Analysis of Hybrid Hydrogen Systems. NREL/TP-560-46934. Golden, CO:
National Renewable Energy Laboratory, 2010. http://www.nrel.gov/docs/fy10osti/46934.pdf.
64. Carbon Capture R&D. U.S. Department of Energy Office of Fossil Energy. 2014.
http://energy.gov/fe/science-innovation/carbon-capture-and-storage-research/carbon-capture-rd.
65. David, J.; Herzog, H. The Cost of Carbon Capture. Proceedings of 5th International
Conference on Greenhouse Gas Control Technologies (GHGT-5). Cairns, Australia. CSIRO,
2001; pp. 985-990. http://sequestration.mit.edu/pdf/David_and_Herzog.pdf.
66. Gas Turbine World 2006 Handbook. Southport, CT: Pequot Publishing Inc., 2006.
67. Chemical Engineering Magazine Plant Cost Index. (2011). Chemical Engineering Magazine.
http://www.che.com/pci/.
68. U.S. Producer Price Indexes Chemicals and Allied Products/Industrial Inorganic Chemicals
Index. Chemical Economics Handbook. Menlo Park, CA: SRI Consulting, 2008.
69. National Employment, Hours, and Earnings Catalog, Industry: Chemicals and Allied Products,
1980-2009. (2009). http://data.bls.gov/cgi-bin/srgate.
70. Dillich, S.; Ramsden, T.; Melina, M. DOE Hydrogen and Fuel Cells Program Record. Record
#12024. U.S. DOE, Editor, 2012.
http://www.hydrogen.energy.gov/pdfs/12024_h2_production_cost_natural_gas.pdf.
71. Natural Gas Prices. U.S. EIA, 2014.
http://www.eia.gov/dnav/ng/ng_pri_sum_a_EPG0_PIN_DMcf_a.htm.
72. Electricity Retail Price to Consumers. U.S. EIA, July 2013.
http://www.eia.gov/electricity/data.cfm#sales.

80
73. Electricity Wholesale Market Data. U.S. EIA, 2013.
http://www.eia.gov/electricity/wholesale/index.cfm.
74. Peters, M.S., et al. Plant design and economics for chemical engineers. New York: McGraw-
Hill, 2003.
75. Short, W.; Packey, D.J.; Holt, T. A Manual for the Economic Evaluation and Energy Efficiency
and Renewable Energy Technologies. Golden, CO: National Renewable Energy Laboratory,
1995.
76. How to Depreciate Property. Department of the Treasury, Internal Revenue Service, 2009.
http://www.irs.gov/pub/irs-pdf/p946.pdf.
77. Perry, R.H.; Green, D.W. Perry's Chemical Engineers' Handbook. 7th Edition. New York:
McGraw-Hill, 1997.
78. Gary, J.H.; Handwerk, G.E. Petroleum refining: technology and economics. 3rd edition. New
York: M. Dekker, 1994; pp.xii465.
79. Cran, J. Improved factored method gives better preliminary cost estimates. Chemical
Engineer (88:7), 1981; pp. 65-79.
80. Algae Testbed Public-Private Partnership (ATP3). Accessed 2014. http://atp3.org/
81. SimaPro, Vers.7.3. Product Ecology Consultants, Amersfoort, the Netherlands, 2011.
82. Hsu, D., et al. Life Cycle Environmental Impacts of Selected U.S. Ethanol Production and Use
Pathways in 2022. Environ. Sci. Technol., 2010; pp. 52895297.
83. GREET 2013; all supporting information/documentation available at
https://greet.es.anl.gov/publications.
84. Ecoinvent, Vers. 2.2. Swiss Center for Life Cycle Inventories, Duebendorf, Switzerland, 2010.
85. U.S. Life-Cycle Inventory. Vers. 1.6.0. National Renewable Energy Laboratory: Golden, CO,
2008.
86. Wang, M.; Huo, H.; Arora, S. Methods of dealing with co-products of biofuels in life-cycle
analysis and consequent results within the US context. Energ. Policy (39:10), 2011; pp.5726-
5736.
87. Biorecro, A. B. Global Status of BECCS Projects 2010. Canberra, Australia: Global CCS
Institute, 2010.
88. Fischer, B. S., et al. Issues related to mitigation in the long term context, Chapter 3. Climate
Change 2007: Mitigation. Contribution of Working Group III to the Fourth Assessment Report
of the Inter-governmental Panel on Climate Change. Cambridge: Cambridge University Press,
2007.
89. H2A Production Model. Version 3; Available from:
http://www.hydrogen.energy.gov/h2a_production.html.
90. Bozell, J.J.; Petersen, G.R. "Technology development for the production of biobased products
from biorefinery carbohydrates-the US Department of Energy's Top 10 revisited. Green
Chem. (12:4), 2010; pp. 539-554.
91. Kramer, A., Supporting biofuels through production of high-value co-products. Presented at
the 2013 Algae Biomass Summit; Orlando, FL, September 30-October 3, 2013.
92. Algix, Accessed 2014: http://algix.com.
93. Huo, Y., et al. Conversion of Proteins into Biofuels by Engineering Nitrogen Flux. Nat.
Biotechnol. (29:4), 2011; pp. 346-351.
94. Davis, R.; Biddy, M. Algal Lipid Extraction and Upgrading to Hydrocarbons Technology
Pathway. NREL/TP-5100-58049. Golden, CO: National Renewable Energy Laboratory, 2013.
http://www.nrel.gov/docs/fy13osti/58049.pdf.
95. Smith, V.H.; Crews, T. Applying ecological principles of crop cultivation in large-scale algal
biomass production. Algal Research (4), 2014; pp. 23-34.

81
Appendix A. Individual Equipment Cost Summary
The following table shows abbreviated specifications, purchased cost and installed cost for each
piece of equipment in this process design, with all capital costs set based on the maximum
throughput design case, i.e. summer season flows. Although each piece of equipment has its own
line, many were quoted as part of a package. NREL and Harris Group would like to acknowledge
the equipment vendors who assisted us with the cost estimation effort for this design report.

82
83
84
85
Appendix B. Discounted Cash Flow Rate of Return
Worksheet

86
87
Appendix C. Aspen Plus Properties
The table below is a list of the components used in the Aspen model. Previous versions of NREL
models used custom property databanks created at NREL. Where possible in the new model,
these components have been replaced with components from Aspens native databanks. Property
definitions for the few remaining custom components were moved into the model itself (i.e.,
inside the simulation file) and are discussed here.

88
Component Property Quantity Units Reference
H2O - - - Native Aspen component
ETHANOL - - - Native Aspen component
GLUCOSE - - - Native Aspen component (dextrose)
SUCROSE - - - Native Aspen component; used to represent non-fermentable carbohydrates
HMF MW 126.11 (5-hydroxymethyl furfural) Properties for HMF were estimated within Aspen using
NIST TDE routines. Specify molecular structure, MW, TB and DHFORM.
TB 532.7 K
DHFORM -79774.53 cal/mol
DHVLWT- 80550000 J/kmol
1
TC 731.012 K NIST TDE
PC 5235810 Pa
OMEGA 0.99364671
VC 0.3425 m3/kmol
RKTZRA 0.198177974
LACID - - - Native Aspen component (lactic acid)
NH3 - - - Native Aspen component
H2SO4 - - - Native Aspen component
NH4SO4 - - - Native Aspen component (ammonium sulfate)
DAP - - - Native Aspen component (diammonium phosphate)
NAOH - - - Native Aspen component
OIL - - - Native Aspen component (oleic acid); used to represent polar lipid impurity fractions
O2 - - - Native Aspen component
N2 - - - Native Aspen component
NO - - - Native Aspen component
NO2 - - - Native Aspen component
CO - - - Native Aspen component

89
Component Property Quantity Units Reference
CO2 - - - Native Aspen component
CH4 - - - Native Aspen component
H2S - - - Native Aspen component
SO2 - - - Native Aspen component
GLUCAN - - - Native Aspen component (starch); based on [6] CELLULOS component
PROTEIN Formula CH1.57O0.31N0.29S0.007 Wheat gliadin; all protein component properties left unchanged from [6]
MW 22.8396
DHSFRM -17618 cal/mol From literature value of gliadin Hc
PLXANT/1 -1.00E+20 atm Forces non-volatility
ASH - - - Native Aspen component CaO
ZYMO Formula CH1.8O0.5N0.2 Used to represent S. cerevisiae cell mass based on [3, 6] ZYMO component
MW 24.6264
DHSFRM -31169.39 cal/mol See [6]
BIOMASS Formula CH1.64O0.39N0.23S0.0035 Anaerobic digestion solids residue in digestate cake; based on [6]
MW 23.238
DHSFRM -23200.01 cal/mol See [6]
CASO4 - - - Gypsum, used to represent silica and clay for lipid purification; native Aspen component
TAG - - - FAME lipids. Native Aspen component Triolein
HEXANE - - - Native Aspen component (n-hexane)
H3PO4 Formula H3PO4 Phosphoric acid for lipid purification. Native Aspen component
H2 Formula H2 Native Aspen component

90
Component Property Quantity Units Reference
C9 Formula C9H20 Native Aspen component (n-nonane)
C10 Formula C10H22 Native Aspen component (n-decane)
C11 Formula C11H24 Native Aspen component (n-undecane)
C12 Formula C12H26 Native Aspen component (n-dodecane)
C13 Formula C13H28 Native Aspen component (n-tridecane)
C14 Formula C14H30 Native Aspen component (n-tetradecane)
C15 Formula C15H32 Native Aspen component (n-pentadecane)
C16 Formula C16H34 Native Aspen component (n-hexadecane)
C17 Formula C17H36 Native Aspen component (n-heptadecane)
C18 Formula C18H38 Native Aspen component (n-octandecane)
C19 Formula C19H40 Native Aspen component (n-nonadecane)
C20 Formula C20H42 Native Aspen component (n-eicosane)
ALGCEL - - - Algal cell biomass. Duplicate of component ZYMO.

91
Appendix D. Process Flow Diagrams
High-level stream table information from Aspen Plus modeling output follows, for key streams
associated with each process operation area. The stream table values are associated with the
HLSD + Store base case scenario, for the summer season which is used to set the basis for the
facility design capacity. This is followed by high-level PFDs for the associated process areas.
Space for stream tables was limited; below is a key to lumped components. As the stream table
information focuses primarily on the high-level overall process and does not include every
individual modeled stream within each process area, mass balance closure around a given unit
area may not be 100%.

RDB Paraffins targeted in the C13-C20 range

Naphtha Paraffins targeted in the C6-C12 range

Soluble Fermentable Sugars Glucose, mannose (modeled as 100% glucose)

Furfurals Furfural, HMF

CO/SOx/NOx/H2S CO, SO2, NO, NO2, H2S

Cell Mass S. cerevisiae, algal cell mass (see Table 1), AD solids

Other Insoluble Solids Silica, clay (lipid purification chemicals)

92
 240
Recycle Water
230
CO2 Vent to Recycle

140 VENT
A100 AMMONIA 160

Recycle Water
HP STEAM 170

190
Dry Algae
ALGAE GROWTH Wet Algae in Summer
in Summer ALGAE DRY ALGAE ACID 180 MINOR
AND DEWATERING 150 FERMENTATION 210
DRYING STORAGE NUTRIENTS
(not modeled here)
WATER FOR VENT
110

120 CH4 Dry Algae in Winter

PRETREATMENT SCRUBBER
220

130 AIR AND

Algal Biomass
CONDITIONING
100 115
200
A200
Hydrolysate
FERMENTATION
AND DISTILLATION
PURIFICATION 430 Air
CHEMICALS AND 410
WASH WATER MAKEUP
310
440 MP STEAM SOLVENT

Ethanol Product
MAKEUP H2 420

600
A300
450 A400 Stillage
Raw Oil LIPID EXTRACTION
Recycle Water PRODUCT 300
400 AND SOLVENT
PURIFICATION
RECOVERY
AND UPGRADING
Naphtha

610
FURNACE 460 RDB
A600
FLUE GAS STORAGE
620

Light Gases from Upgrading

510 530 CH4

520 540 Air

Waste from Purification A700
Recycle Water + Nutrients Spent Residue UTILITIES
550 500
A500
560
ANAEROBIC
170 HP STEAM TO AREA A100
DIGESTION AND
Flue Gas
AD DIGESTATE 570
CHP 440 MP STEAM TO AREA A400

OVERALL PROCESS:
VOLATILE N LOSS 580
ALGAL FRACTIONATION PROCESS
JULY 2014 PFD-P120-A000

Figure 19. Overall process schematic for Aspen stream tables

93
Aspen Plus mass balance information for key stream tables:

COMPONENT UNITS 100 110 115 120 130 140 150 160 170 180 190 200 210
Total Flow kg/hr 430,242 147,375 282,867 8,111 154,354 276,968 32,872 786 25,847 2,433 19,960 291,973 112
Insoluble Solids wt% 20.0% 20.0% 20.0% 0.0% 0.0% 0.0% 89.7% 0.0% 0.0% 0.0% 0.0% 11.9% 0.0%
Soluble Solids wt% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 9.2% 100.0%
Temperature C 25 25 25 25 25 100 100 25 268 20 37 37 25
Pressure atm 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 13.0 3.4 1.8 1.0 1.0
Vapor Fraction 0.0 0.0 0.0 1.0 1.0 1.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0

RDB kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Naphtha kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Ethanol kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Water kg/hr 344,194 117,900 226,294 0 0 132,720 3,397 0 25,847 170 19,956 230,210 0
Soluble Fermentable Sugars kg/hr 0 0 0 0 0 0 0 0 0 0 0 21,594 0
Non-Fermentable Carbohydrates kg/hr 3,382 1,158 2,223 0 0 0 0 0 0 0 0 2,223 0
FAME Lipids kg/hr 35,443 12,141 23,303 0 0 0 0 0 0 0 5 23,298 0
Polar Lipid Impurities kg/hr 1,773 607 1,165 0 0 0 0 0 0 0 0 1,165 0
Hexane kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Furfurals kg/hr 0 0 0 0 0 0 0 0 0 0 0 50 0
Lactic Acid kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Ammonia kg/hr 0 0 0 0 0 0 0 786 0 0 0 0 0
Sulfuric Acid kg/hr 0 0 0 0 0 0 0 0 0 2,263 0 0 0
Ammonium Sulfate kg/hr 0 0 0 0 0 0 0 0 0 0 0 3,049 0
Phosphoric Acid kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
DAP kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 112
H2 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
CH4 kg/hr 0 0 0 8,111 0 0 0 0 0 0 0 0 0
Propane kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
O2 kg/hr 0 0 0 0 35,952 3,595 0 0 0 0 0 0 0
N2 kg/hr 0 0 0 0 118,402 118,402 0 0 0 0 0 0 0
CO2 kg/hr 0 0 0 0 0 22,251 0 0 0 0 0 0 0
CO/SOX/NOX/H2S kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0

Carbohydrate (IS) kg/hr 32,845 11,251 21,594 0 0 0 0 0 0 0 0 2,095 0

Protein (IS) kg/hr 7,710 2,641 5,069 0 0 0 0 0 0 0 0 5,069 0
Cell Mass (IS) kg/hr 3,072 1,052 2,020 0 0 0 29,475 0 0 0 0 2,020 0
Ash kg/hr 1,824 625 1,199 0 0 0 0 0 0 0 0 1,199 0
Other Insoluble Solids kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0

94
COMPONENT UNITS 220 230 240 300 310 400 410 420 430 440 450 460 500
Total Flow kg/hr 12,282 9,609 22,755 262,260 993 22,576 2,368 546 1,066 780 1,082 1,227 240,677
Insoluble Solids wt% 0.0% 0.0% 0.0% 13.7% 0.0% 0.0% 2.9% 0.0% 0.0% 0.0% 0.0% 0.0% 5.5%
Soluble Solids wt% 0.0% 0.0% 0.0% 2.5% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 2.7%
Temperature C 33 67 30 125 25 110 25 25 32 232 78 380 52
Pressure atm 5.0 1.2 1.6 2.3 1.0 1.0 1.0 1.0 1.0 2.4 1.7 1.2 1.0
Vapor Fraction 0.0 1.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 1.0 0.0 1.0 0.0

RDB kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Naphtha kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Ethanol kg/hr 0 2 11 125 0 0 0 0 0 0 0 0 125
Water kg/hr 12,282 245 22,738 219,759 0 0 2,258 0 0 780 1,082 128 219,759
Soluble Fermentable Sugars kg/hr 0 0 0 569 0 0 0 0 0 0 0 0 569
Non-Fermentable Carbohydrates kg/hr 0 0 0 2,223 0 0 0 0 0 0 0 0 2,223
FAME Lipids kg/hr 0 0 5 23,293 0 22,124 0 0 0 0 0 0 1,170
Polar Lipid Impurities kg/hr 0 0 0 1,165 0 385 0 0 0 0 0 0 781
Hexane kg/hr 0 0 0 0 993 68 0 0 0 0 0 0 925
Furfurals kg/hr 0 0 0 50 0 0 0 0 0 0 0 0 50
Lactic Acid kg/hr 0 0 0 586 0 0 0 0 0 0 0 0 586
Ammonia kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Sulfuric Acid kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Ammonium Sulfate kg/hr 0 0 0 3,049 0 0 0 0 0 0 0 0 3,049
Phosphoric Acid kg/hr 0 0 0 0 0 0 43 0 0 0 0 0 0
DAP kg/hr 0 0 0 96 0 0 0 0 0 0 0 0 96
H2 kg/hr 0 0 0 0 0 0 0 546 0 0 0 0 0
CH4 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
Propane kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0
O2 kg/hr 0 0 0 0 0 0 0 0 248 0 0 41 0
N2 kg/hr 0 0 0 0 0 0 0 0 818 0 0 818 0
CO2 kg/hr 0 9,362 0 0 0 0 0 0 0 0 0 240 0
CO/SOX/NOX/H2S kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0

Carbohydrate (IS) kg/hr 0 0 0 2,095 0 0 0 0 0 0 0 0 2,095

Protein (IS) kg/hr 0 0 0 5,011 0 0 0 0 0 0 0 0 5,011
Cell Mass (IS) kg/hr 0 0 0 3,039 0 0 0 0 0 0 0 0 3,039
Ash kg/hr 0 0 0 1,199 0 0 0 0 0 0 0 0 1,199
Other Insoluble Solids kg/hr 0 0 0 0 0 0 68 0 0 0 0 0 0

95
COMPONENT UNITS 510 520 530 540 550 560 570 580 600 610 620
Total Flow kg/hr 4,154 2,821 949 334,316 206,658 348,022 28,194 43 9,744 392 17,661
Insoluble Solids wt% 0.0% 16.0% 0.0% 0.0% 0.1% 0.0% 29.7% 0.0% 0.0% 0.0% 0.0%
Soluble Solids wt% 0.0% 0.0% 0.0% 0.0% 1.5% 0.0% 0.9% 0.0% 0.0% 0.0% 0.0%
Temperature C 30 93 25 25 35 140 35 35 20 20 20
Pressure atm 1.7 1.0 1.0 1.0 1.0 1.5 1.0 1.0 1.0 1.0 1.0
Vapor Fraction 1.0 0.0 1.0 1.0 0.0 1.0 0.0 1.0 0.0 0.0 0.0

RDB kg/hr 1 0 0 0 0 0 0 0 0 0 17,608

Naphtha kg/hr 12 0 0 0 0 0 0 0 0 382 0
Ethanol kg/hr 0 0 0 0 0 0 0 0 9,678 0 0
Water kg/hr 25 2,258 0 0 202,636 14,287 19,114 0 49 2 53
Soluble Fermentable Sugars kg/hr 0 0 0 0 0 0 0 0 0 0 0
Non-Fermentable Carbohydrates kg/hr 0 0 0 0 0 0 0 0 0 0 0
FAME Lipids kg/hr 0 0 0 0 0 0 0 0 0 0 0
Polar Lipid Impurities kg/hr 0 385 0 0 0 0 0 0 0 0 0
Hexane kg/hr 0 68 0 0 0 0 0 0 0 0 0
Furfurals kg/hr 0 0 0 0 0 0 0 0 0 0 0
Lactic Acid kg/hr 0 0 0 0 0 0 0 0 0 0 0
Ammonia kg/hr 0 0 0 0 740 0 217 43 0 0 0
Sulfuric Acid kg/hr 0 0 0 0 0 0 0 0 0 0 0
Ammonium Sulfate kg/hr 0 0 0 0 2,786 0 263 0 0 0 0
Phosphoric Acid kg/hr 0 43 0 0 0 0 239 0 0 0 0
DAP kg/hr 0 0 0 0 322 0 0 0 0 0 0
H2 kg/hr 159 0 0 0 0 0 0 0 0 0 0
CH4 kg/hr 0 0 949 0 0 0 0 0 0 0 0
Propane kg/hr 1,212 0 0 0 0 0 0 0 0 7 0
O2 kg/hr 0 0 0 77,868 0 52,247 0 0 0 0 0
N2 kg/hr 0 0 0 256,448 0 256,360 0 0 0 0 0
CO2 kg/hr 2,622 0 0 0 2 24,643 0 0 17 1 0
CO/SOX/NOX/H2S kg/hr 123 0 0 0 2 485 0 0 0 0 0

Carbohydrate (IS) kg/hr 0 0 0 0 0 0 0 0 0 0 0

Protein (IS) kg/hr 0 0 0 0 0 0 0 0 0 0 0
Cell Mass (IS) kg/hr 0 0 0 0 145 0 7,120 0 0 0 0
Ash kg/hr 0 0 0 0 24 0 1,175 0 0 0 0
Other Insoluble Solids kg/hr 0 68 0 0 1 0 66 0 0 0 0

96
 HYDROLYSATE TO
FERMENTATION PFD-002

ALGAL BIOMASS
FEED

PRESTEAMERS

WATER TO WATER
HP STEAM FROM RECYCLE
PFD-005

WATER
SOLVENT RECOVERY COOLER
PRE-HEAT EXCHANGER

PRETREATMENT
REACTORS

SULFURIC ACID

AMMONIA ADDITION TANK

Blowdown tank

AMMONIA

NATIONAL RENEWABLE
ENERGY LABORATORY

Harris Group Inc. ADVANCED BIOFUELS ENGINEERING DESIGN

Engineering for Optimum Performance
ALGAE FRACTIONATION TO FUELS
www.harrisgroup.com
PROCESS FLOW DIAGRAM
Subconsultant:
- A100 PRETREATMENT AND CONDITIONING
Drawn: JAR
- Check: -
Engr: JAR
- Check: - Project No: Drawing:

PFD-001
Rev:
A
Appr:
- PMgr: -

97
 SEED FERMENTORS
CO2 RECYCLE TO PONDS

PROCESS WATER

VENT
SCRUBBER

STILLAGE TO EXTRACTION
PFD-003

HYDROLYSATE FROM
PRETREATMENT &
CONDITIONING
PFD-001 BEER WELL

WATER WATER LP
COOLER COOLER STEAM

BEER
COLUMN

ETHANOL PRODUCT

WATER
CHILLED COOLER
WATER

BATCH FERMENTORS
MOLECULAR SIEVE
LP ADSORPTION
STEAM

RECTIFIER BOILER FEED WATER (BFW)

ECONOMIZER #1

LP
STEAM BFW

TO WATER RECYCLE

NATIONAL RENEWABLE
ENERGY LABORATORY

Harris Group Inc. ADVANCED BIOFUELS ENGINEERING DESIGN

Engineering for Optimum Performance
ALGAE FRACTIONATION TO FUELS
www.harrisgroup.com
Subconsultant: -
PROCESS FLOW DIAGRAM
Drawn: JAR - Check: -
A200 FERMENTATION AND SEPARATION
Engr: JAR - Check: - Project No: Drawing: Rev:
Appr: - PMgr: -
PFD-002 A

98
 Hexane

Hexane +
Algal Oil
Solvent
Recovery
Hexane makeup
Column

Pretreatment
Stillage from Flash Vapor
PFD-001
beer column
PFD-002 HP
Steam

Extraction
Column Oil product to purification
and upgrading
PFD-004
BFW

Boiler Feed Water

Spent biomass to
(BFW) Economizer #2
Anaerobic digestion
PFD-005

NATIONAL RENEWABLE
ENERGY LABORATORY

Harris Group Inc. ADVANCED BIOFUELS ENGINEERING DESIGN

Engineering for Optimum Performance
ALGAE FRACTIONATION TO FUELS
www.harrisgroup.com
Subconsultant: -
PROCESS FLOW DIAGRAM
Drawn: JAR - Check: -
A300: EXTRACTION AND SOLVENT RECOVERY
Project No: Drawing:
Engr: JAR - Check: -
PFD-003 A
Rev:

Appr: - PMgr: -

99
 MAKEUP H2
COMPRESSOR

H2 from offsite SMR RECYCLE

COMPRESSOR
H2

Natural gas PSA Tailgas Off-Gases to

(normally not needed) Note 1 PFD-005

PREHEATING
FURNACE
HYDROTREATING
REACTOR
HYDROTREATING
Phosphoric acid, GUARD BED/TAN
clay, silica, washwater REDUCTION PSA UNIT
Product Gas

Algal oil from

PFD-003 LIPID
Off-Gas
PURIFICATION
SEPARATION PARTIAL
CONDENSER
Waste to CW
PFD-005
Liquid Naphtha to
Product HP Product Storage
STEAM

PRODUCT
DISTILLATION
Stripping COLUMN
Steam

Renewable Diesel
Blendsock (RDB) to
Product Storage
Water
Water to Water
Recycle Pool
Dashed box represents hydrotreating ISBL system

NATIONAL RENEWABLE
Notes: ENERGY LABORATORY
1. Natural gas for startup and as needed
Harris Group Inc. ADVANCED BIOFUELS ENGINEERING DESIGN
Engineering for Optimum Performance
ALGAE FRACTIONATION TO FUELS
www.harrisgroup.com
Subconsultant: -
PROCESS FLOW DIAGRAM
Drawn: JAR - Check: -
A400: PRODUCT PURIFICATION AND UPGRADING
Project No: Drawing:
Engr: JAR - Check: -
PFD-004 A
Rev:

Appr: - PMgr: -

100
 HP steam to pretreatment
LP steam to product
injection, PFD-001
distillation column
PFD-004

High-pressure Low-pressure
steam users steam users

Power

HP-HRSG1 HP-HRSG2 LP-HRSG

Natural gas Flue gas Flue gas recycled
(supplemental) to ponds

BIOGAS TURBINE
GENERATOR
Hydrotreating BFW
off-gas BFW BFW
makeup
ECONOMIZER #1 ECONOMIZER #2
Biogas Boiler feed
Boiler water
water
Oil product from
blowdown
Water product from
Waste from purification ethanol rectifier solvent recovery
PFD-004 PFD-002 column, PFD-003
BFW
makeup
CENTRIFUGE
Spent residues from AD Effluent Water recycled
extraction column Boiler water
to ponds
PFD-003 CW blowdown

ANAEROBIC
Digestate
DIGESTION (N fertilizer coproduct)

NATIONAL RENEWABLE
ENERGY LABORATORY

Harris Group Inc. ADVANCED BIOFUELS ENGINEERING DESIGN

Engineering for Optimum Performance
ALGAE FRACTIONATION TO FUELS
www.harrisgroup.com
Subconsultant: -
PROCESS FLOW DIAGRAM
Drawn: JAR - Check: -
A500: ANAEROBIC DIGESTION AND CHP
Project No: Drawing:
Engr: JAR - Check: -
PFD-005 A
Rev:

Appr: - PMgr: -

101
This report is available at no cost from the National Renewable Energy Laboratory (NREL) at www.nrel.gov/publications.