Final Project Report

UOP, LLC

Pilot-Scale Biorefinery: Sustainable Transport Fuels from Biomass

and Algal Residues via Integrated Pyrolysis, Catalytic
Hydroconversion and Co-processing with Vacuum Gas Oil

Award No. EE0002879

July 20, 2016

The information contained in this Final Project Report is Unrestricted

and can be publicly distributed.

Page | 1
Table of Contents
Executive Summary…………………………………………………………………………………...…3

1. Introduction and Background........................................................................................................ 12

2. Project Objectives, Overview and Successes ........................................................................... 16

3. Project Management, Contracting and Risk Mitigation ............................................................ 24

4. Project Scope Changes ................................................................................................................. 27

5. RTP Operations .............................................................................................................................. 28

6. Upgrader- I ......................................................................................................................................36

7 Upgrader II Process .......................................................................................................................46

8 FCC Co-processing ........................................................................................................................63

9. Commercialization and Regulatory pathway ..............................................................................69

10. Feedstock Characterization and Life Cycle Analysis ................................................................76

11. Project Lessons Learned...............................................................................................................86

Appendices ……………………..……………………..………………………………………………..88

Page | 2
Executive Summary

Beginning in 2010, UOP, along with the Department of Energy and other project partners,
designed a pathway for an integrated biorefinery to process solid biomass into transportation
fuel blendstocks. The integrated biorefinery (IBR) would convert second generation feedstocks
into pyrolysis oil which would then be upgraded into fuel blendstocks without the limitations of
traditional biofuels. The upgrading technology consisted of hydroprocessing of pyrolysis oil to
remove oxygenates and water to create renewable gasoline, diesel and jet fuel blendstocks that
met existing fuel quality standards.

IBR Hawaii Site Operations

The IBR project was constructed alongside an existing petroleum refinery on the island of
Oahu to utilize the hydrogen stream from the refinery for hydroprocessing of the pyrolysis oil.
In the project’s first phase during 2011-2012, UOP designed and constructed a Rapid Thermal
Processing (RTPTM) unit at pilot scale to convert 1 bone dry metric tonne per day into pyrolysis
oil. The project’s second stage consisted of removal of the solids from the pyrolysis oil and
then reducing the metals contained in the oil (UG-I). The third stage consisted of the
hydroprocessing technology to upgrade the pyrolysis oil to fuel blendstocks (UG-II). Design
delays prevented this stage from being constructed at the project site. This full pathway
is shown in Figure i.

Figure i – Original Integrated Biorefinery Project Scope

Page | 3
After the RTP and UG-I components were commissioned in late 2012, the site remained idle
until 2014. Continued challenges with development and design of the UG-II hydroprocessing
unit prompted a project scope change from construction of the upgrading unit at the IBR site to
laboratory plant testing at UOP’s Research facility in Des Plaines, IL.

Site operations restarted in late Summer 2014 to process 70 tons of four different biomass
feedstocks; hardwood, softwood, corn stover and switchgrass. During Fall 2014, the RTP and
UG-I produced over 1000 gallons of pyrolysis oil and successfully lowered the metals content to
below 50 ppm-wt. Metals removal to this level was identified as a key requirement to reduce
pyrolysis oil hydroprocessing catalyst deactivation and thereby minimizing operating costs.
Table i below depicts the metals quantity after three individual process runs.

Table i – MFU Metals Removal Results

Run 2 Run 3
MFU Feed* Run 1 Product Product
Metal wt-ppm wt-ppm wt-ppm wt-ppm

Total
Metals 1016 633 94 247

During the Fall 2014, operational problems with the RTP unit at the IBR site limited the total
amount of biomass processed and prompted a full shutdown in December 2014 to make
modifications to improve reliability.

The RTP was restarted in February 2015 after additional modifications were made to improve
the operations and all hardwood was processed. In March 2015, softwood, switchgrass and
corn stover feedstocks were attempted to be processed through the unit. Several efforts were
made to process these biomass feeds in the unit, but the softwood, switchgrass and corn stover
were found to not meet the feedstock specification for size and shape for the system. With the
project termination looming in summer 2015, the decision was made in April 2015 to cease IBR
site operations.

A summary of successful RTP hardwood processing runs is shown in Table ii. Here, the last
column on the right displays a consistent improvement in total liquid yield per bone dry tonne of
biomass feedstock. This confirms the value of the RTP equipment modifications made during
late 2014 and early 2015. With proper sand and heat control, meeting yield goals is achievable
provided upstream feedstock handling and feeding can be managed smoothly.

Page | 4
 Table ii – RTP Hardwood Processing Summary

Biomass Total BDMT BDMT

Run No. Hours Biomass per day

16 76.4 2.15 0.87

17 136.0 3.63 1.04

20 53.7 1.41 0.88

21 135.5 3.81 0.85

22 48.5 0.75 0.43

24 171.4 3.82 0.67

25 118.1 4.85 1.16

Pyrolysis Oil Upgrading

When the hydroprocessing UG-II development was shifted to laboratory scale, extensive
experimental work was conducted on first and second stage upgrading of pyoil using
conventional fixed bed hydroprocessing technology. While it was possible to demonstrate
1000 hours on stream of first stage upgrading, operational challenges still remain. Recurring
problems with pressure drop buildup in the first stage upgrading experiments made progress
very difficult and remains an unresolved issue that impedes the development of this technology.

Numerous attempts were made to promote the hydroprocessing reaction while suppressing the
polymerization reactions. While some progress was made, it was not enough to get to a stable
operating point that would yield commercially viable operation. For this reason, work on first
stage upgrading was not taken beyond the pilot plant scale (~1-3 kg/hr feed rate).

Upgrading process yields have been estimated based on the experiments. After conversion,
approximately 51% yield by volume forms naphtha, diesel and fuel oil, with more than half of the
weight of the pyoil products converted to water. This is a combination of water that was native to
the pyoil feed and water that is produced when organic oxygen is hydrotreated and converted
into water. The predominant hydrogen consuming reaction is the conversion of organic oxygen
to water.

Page | 5
The 51% hydrocarbon yield is a combination of material that boils in the naphtha (gasoline),
diesel and fuel oil boiling range. There is some uncertainty in what the actual total recoverable
hydrocarbon yield is, due to the fact that some light organic material is soluble in the water
phase. How much of this material is recovered and ultimately upgraded into fuel quality products
is determined by the processing flow scheme and how much upgrading cost can be justified to
recover these products.

Pyrolysis Oil Co-processing

In Spring 2014, due to complexity with hydroprocessing pyrolysis oil to fuel blendstocks, the
project scope was changed to include an alternate pathway for co-processing pyrolysis oil in a
fluid catalytic cracking (FCC) unit. This alternate project pathway is shown below in Figure ii.

Figure ii – Revised Integrated Biorefinery Project Scope

Kapolei, Hawaii UOP Pilot Plants

UGII
RTP UG1 (pilot scale)
(Demo Scale) (Demo Scale) Two
Two Stage
Stage
Biomass 168 GPD RFO 168 GPD RFO) Hydroprocessing
Hydroprocessing Fuel
1-ton/day 0.4 GPD/+1000 HOS
Rapid Metals
Thermal Finishing FCC
Processing Unit (pilot scale)
Unit (MFU) Co-processing
in Circulating Fuel
Riser (2-5%)

The revised scope would incorporate laboratory pilot plant scale co-processing of pyrolysis oil
at UOP’s Research facility in Des Plaines, IL. Initial work on co-processing indicated the
pathway may be a near-term solution for petroleum refiners to integrate renewable fuels into
their existing refinery infrastructure. This would allow for a fully integrated pathway from
biomass feedstocks to partially renewable fuels.

During co-processing, a secondary feed stream consisting of cellulosic-derived pyrolysis oil is

processed in the FCC unit along with the traditional VGO feed, as depicted in Figure iii below.
In this application, pyrolysis oil can represent up to 5 volume percent of the total feed to the unit.
The main function of co-processing is to deoxygenate the pyrolysis oil, much of which boils in
the gasoline and diesel ranges. The two main deoxygenation pathways of pyrolysis oil in the

Page | 6
FCC unit are decarboxylation and dehydration. In addition to gasoline and diesel range material,
pyrolysis oil deoxygenation also produces carbon monoxide, carbon dioxide and water as
by-products.

Figure iii – Pyrolysis Oil Co-processing

The results of the pilot plant co-processing testing using three wt-% pyrolysis oil showed
increased gasoline and light cycle oil (LCO) yields along with reductions in dry gas and liquefied
petroleum gas (LPG) yields. Hydrogen yield also decreased as hydrogenation reactions take
place with the hydrogen deficient components present in the pyrolysis oil. Also, an increase was
in coke yield was observed in the three wt-percent pyrolysis oil experiments. A total of 26 tests
of one hour duration (each) were completed to span eight test conditions (four catalyst-to-oil
ratio targets both with and without pyrolysis oil).

Commercializing the technology of converting pyrolysis oil to transportation fuels by either

upgrading or co-processing in an FCC unit are similar but the co-processing solution offers a
near-term option that fits today’s biofuel mandates. Co-processing offers the ability to integrate
renewable fuels into a refiner’s existing infrastructure and limit the capital expenditures of
downstream blending. Figure iv below shows integration at the refinery which allows for normal
downstream processing of partially renewable fuels to deliver high quality products to the
marketplace.

Page | 7
 Co-processing Pyrolysis oil in FCC Integrates at Refinery

Exploration Refining Terminal Rack

Pyrolysis oil
Integration

Using co-processing of pyrolysis oil to produce fuel blendstocks in the United States would
allow a refiner to generate cellulosic Renewable Identification Numbers (RINs) under the EPA
Renewable Fuel Standard (RFS). One of the main priorities of the RFS program is the
increased production of second generation renewable fuels and co-processing can quickly fill
the need for this part of the program.

Biomass Feedstock Characterization

As part of the project, 24 different biomass feedstocks were characterized and tested for
conversion to pyrolysis oil. A cross section of woody biomass, agricultural residues, grasses and
algae feedstocks was compared for liquid yield, water content and viscosity as well as carbon
and nitrogen content. The results showed clean woodchips, switchgrass and bagasse produce
the highest liquid yield, while corn stover must be clean of dirt and with low ash content to
maximize yields. Pyrolysis of algae yielded highly viscous pyrolysis oil that was more tar-like
than liquid and quickly polymerized when heated. Overall, the testing indicated no “show
stoppers” with varying feedstocks, except for use of algae.

Life Cycle Analysis

As greenhouse gas reduction is one of the primary objectives of the integrated biorefineries,
a Life Cycle Analysis (LCA) was performed using the pathway of biomass feedstock to fully
upgraded fuels. This data was generated using laboratory scale results for hydroprocessing.
Michigan Technological University (MTU) was contracted to perform the LCA. MTU has
extensive experience developing LCA’s and was the ideal partner for this process.

Page | 8
The results of this LCA analysis are depicted below in Figure iii for nine types of biomass
feedstocks. Note; all feedstocks with the exception of wild algae reduced GHG emissions by
60% to less than 36.2 gCO2e per MJ to qualify as cellulosic biofuels under the RFS program.

Figure iii – GHG reductions for Upgraded Fuels

60% reduction level

Independent of this project, a life cycle analysis of co-processing pyrolysis oil in an FCC unit at
a refinery indicates a renewable fuel with less than 25 gCO2e per MJ. This result gives
assurance that co-processed fuels will also qualify as cellulosic biofuels.

Summary

Although the project faced many challenges, many lessons were learned that can be of value
to future projects in the area of biomass conversion and upgrading to transportation fuels.
UOP and its project partners continue to pursue a commercial scale integrated biorefinery
pathway based on the findings from this project. In the meantime, a full pathway to produce
partially renewable fuels through pyrolysis oil co-processing can fill mandates in the United
States and worldwide.

Page | 9
List of Abbreviations
AAS Atomic Adsorption Spectroscopy
ABT Average Bed Temperature
ATR Attenuated Total Reflectance
bbl Barrel
BDMT Bone Dry Metric Ton
BET Brunauer-Emmett-Teller
BM Biomass
BP Boiling Point
BPD Barrels per Day
BSU Bench Scale Unit
CE Cylindrical Extrudate
CED Cumulative Energy Demand
CHN Carbon, Hydrogen, Nitrogen
DA Dry Air
DCS Distributed Control System
dLUC direct Land Use Change
DOE (US) Department of Energy
DSC Differential Scanning Calorimetry
E-Cat Equilibrium Catalyst
EPDM Ethylene Propylene Diene Monomer
EU European Union
FBF Fiber Bed Filter
FCC Fluid Catalytic Cracking
FED Fossil Energy Demand
FTIR Fourier Transform Infrared Spectrometry
g/h Grams per Hour
GC Gas Chromatography
GHG Greenhouse Gas
GPC Gel Permeatio Chromatography
GPM Gallons per Minute
GWP Global Warming Potential
HMI Human-Machine Interface
HOS Hours on Stream
IBR Integrated Bio-Refinery
ICP Inductively Couple Plasma
INL Idaho National Laboratory
IPCC Intergovernmental Panel on Climate Change
IR Infrared (Spectroscopy)
IST Intermediate Storage Tank
KF Karl Fischer (method for water content by titration)
KO Pot Knock Out Pot

Page | 10
LCA Life Cycle Assessment
LCO Light Cycle Oil
LHSV Liquid Hourly Space Velocity
LHV Lower Heating Value
LOI Loss On Ignition
LPG Liquified Petroleum Gas
MCH Methylcyclohexene
MFU Metals Finishing Unit
MJ Megajoule
MS Mass Spectroscopy
MT Metric Ton
MTU Michigan Technological University
MW Molecular Weight
NMR Nuclear Magnetic Resonance (Spectroscopy)
PLC Programmable Logic Controller
ppm Parts per Million
PQT Primary Quench Tower
psi Pounds per Square Inch
psia Pounds per Square Inch - Absolute
psig Pounds per Square Inch - Gauge
PV Process Variable
QL Quadralobe (extrudate)
RFS2 Renewable Fuel Standard
RGB Recycle Gas Blower
RTP Rapid Thermal Pyrolysis
SCFB Standard Cubic Feet per Barrel
scft Standard Cubic Feet
SEM Scanning Electron Microscopy
SQT Secondary Quench Tower
SRU Solids Removal Unit
TC Thermocouple
VGO Vacuum Gas Oil
WABT Weight Average Bet Temperature
WC Weight Check
WF Water Free
XPS X-ray Photoelectron Spectroscopy
XRD X-ray Diffraction

Page | 11
1. Introduction and Background
In 2010, UOP LLC, a Honeywell Company, assembled a comprehensive team to build and
demonstrate a pilot scale biorefinery integrating pyrolysis with hydroconversion to produce
fungible transportation fuels including gasoline, diesel and jet fuel range hydrocarbon from
multiple biomass feedstocks. The project included the collection of high impact feedstocks such
as forestry residue, agricultural waste, and biomass derived from dedicated energy crops such
as switchgrass and algal residues. These materials were to be utilized as feedstocks for a one
tonne per day Rapid Thermal Processing (RTPTM) unit.

RTP is a thermal pyrolysis process in which biomass is rapidly heated in the absence of oxygen.
During this reaction, the biomass is vaporized and then rapidly cooled to generate high yields
of liquid pyrolysis oil. The RTP process utilizes a circulating fluidized bed reactor system with
sand as a solid heat transfer media. RTP typically yields 65 to 75 wt% pyrolysis oil from dried
woody biomass. The RTP unit operates similarly to a UOP designed fluid catalytic cracking
(FCC) unit for petroleum refining. In 2008, Envergent Technologies, a UOP and Ensyn
Corporation joint venture, was formed to commercialize the RTP technology by combining
UOP and Ensyn technical expertise.

The Integrated Biorefinery project intended to use the pyrolysis oil product from the RTP unit
as feedstock for a hydroprocessing unit that uses a fixed bed catalytic reactor technology to
deoxygenate and convert the pyrolysis oil to a hydrocarbon fuel. Hydroconversion would
remove any heteroatoms such as N, S, and O from the feed molecules thereby producing ultra-
clean, renewable fuel. Following hydroconversion, the hydrocarbon product would be distilled
into specific fuel cuts – primarily gasoline but also diesel and jet fuel range hydrocarbon as well.
The basic process is shown below in Figure 1.1.

Figure 1.1 - Overview of Biorefinery Process

Gasoline
Biomass Pyrolysis Hydroconversion Diesel
Jet

The hydrocarbon product would be compatible with existing petroleum-based fuels and could
be utilized in standard transportation fuel infrastructure such as pipelines, tanks and terminals.
This type of “drop-in” renewable fuel would accelerate the blending of renewable fuels without
equipment limitations and assist the petroleum refining industry in meeting compliance
standards around the world.

Page | 12
 The project intent was to demonstrate the technology in a fully integrated facility starting with
biomass feedstock to production of transportation fuel blendstocks. Two major unit operations –
Rapid Thermal Processing (Pyrolysis) and catalytic hydroconversion would be used to convert
the biomass to liquid hydrocarbon fuel which could then be stored and distributed using existing
infrastructure. The flow diagram of the integrated process can be seen in Figure 1.2 below.

Area 1 - Pyrolysis Oil Production

S1
FEEDSTOCK A
(Dried & Sized) RAPID RAW
S9 FEEDSTOCK THERMAL PYROLYSIS
PYOIL PROCESSING OIL
FEEDSTOCK B HANDLING
TREATMENT
UNIT(S) UNIT STORAGE
(Wet & Un-sized) UNIT
1b 1c
1a
S2 S4
S3
Hydrogen Raw Pyrolysis Oil
from Area 2 - PyGasoline Conversion Process
Refinery S5
PyGasoline
PYROLYSIS CATALYTIC PYGASOLINE
Utilities Product
OIL HYDRO- AND
Power STABILISATION CONVERSION PYDIESEL/JET S6
UNIT(S) UNIT STORAGE PyDiesel/Jet
Fuel Gas 2a 2b 2c Product

Steam
Area 3 - By-Product Treatment

Cooling S7
Water WASTE BY-PRODUCT
WATER GAS Fuel Gas
HANDLING PURIFICATION to Refinery
Instrum. UNIT(S) UNIT(S)
Air 3a 3b
S8
Treated Water
from
to Refinery
Refinery

Figure 1.2 – Original Integrated Biorefinery Flow Diagram

The block diagram above is broken down into three areas described as follows:
Area 1 – Pyrolysis Oil Production

Area 2 – PyGasoline/PyDiesel Conversion Process

Area 3 – By-Product Treatment

Each of these sections and the unit operations within is briefly discussed below.

Page | 13
Area 1 – Pyrolysis Oil Production

Unit Operation 1a-Feedstock Handling Unit–Four types of biomass feedstocks were supplied
by project partners. The specific feedstocks were hardwood from Ensyn Corporation and
softwood, corn stover and switchgrass from Idaho National Laboratory. The original plan was to
process 30 tons of each feedstock supplied to the integrated plant.
The feedstock supplied by the partners was dried and sized for feeding directly to the Rapid
Thermal Processing Unit. This is shown as Feedstock A, stream S1 in Figure 1.2. If there was a
need to complete the drying and sizing as part of a commercial scale unit, it would be as shown
in Feedstock B, stream S9.

Unit Operation 1b – Rapid Thermal Processing Unit (RTP) – The prepared (dried and sized)
biomass was stored onsite in large plastic-lined sacks to prevent moisture from reaching the
biomass until it entered the RTPTM feed system. The prepared biomass was introduced to the
reactor by means of an auger or screw conveyor.
Circulating hot sand was used to rapidly transfer the heat required to vaporize the biomass for
the endothermic conversion reaction. The biomass was rapidly converted to a pyrolytic vapor
and a solid char by-product (a powder-like charcoal material) in the RTP reactor. The product
and by-product vapors were separated from the solid char and the sand in a cyclone separation
system which transfers the captured solids (sand, char and ash) to the RTP reheater.
RTP technology includes a proprietary quenching, condensation and separation section which
has been developed to minimize cracking and polymerization reactions and to maximize
recovery of the pyrolysis oil product.
Unit Operation 1c – Raw Pyrolysis Oil Storage – Storage tanks were provided for the raw
pyrolysis oil product exiting the RTP unit to hold this intermediate product for testing, and to
allow for fluctuations in the operation of the downstream unit operations.

Area 2 – PyGasoline/PyDiesel Conversion Process

Unit Operation 2a – Pyrolysis Oil Treatment – As produced, pyrolysis oil from the RTP unit
can contain solid char and heteroatoms that would impact the downstream hydro-conversion
operation. The biorefinery process included two treatment methods to prepare the pyrolysis oil
for further upgrading.
The first was a solids removal process to greatly reduce the amount of solid char and impurities
in the pyrolysis oil. The second step was a process to reduce the quantity of mineral elements in
the pyrolysis oil. A metals finishing unit was constructed to lower the amount of metals in the
filtered pyrolysis oil. Low metals pyrolysis oil is necessary in order to limit the amount of catalyst
deactivation in the catalytic hydroconversion process.

Page | 14
Unit Operation 2b – Catalytic Hydro-Conversion – The original project intent was to take
the RTP product and convert it to hydrocarbon fuel under hydroconversion conditions to fully
deoxygenate the feed molecules with co-production of water and carbon dioxide.
The plan was to hydrodeoxygenate the pyrolysis oil in two stages. In the first stage, pyrolysis
oil and hydrogen are introduced to a reactor containing a hydrogenation catalyst under mild
conditions. The second stage hydroconversion process produces a hydrocarbon product
suitable for blending into the transportation fuel pool. This two stage hydroconversion concept
formed the basis for the proposal. However, due to development and design issues with
hydroconversion technology, this pathway was not completed as part of this project.
Unit Operation 2c–PyGasoline and PyDiesel/Jet Storage: Separate storage tanks for
PyGasoline product, PyDiesel product, and PyJet product blending stock were planned as
part of the integrated plant.

Area 3 – By-Product Treatment

Unit Operation 3a–Waste Water Handling: Original project intent was to include a process for
organic material in the aqueous streams to be catalytically converted into a medium Btu gas
(methane/CO2) that would be recovered for use as fuel gas.
Pacific Northwest National Laboratory (PNNL) has demonstrated the technology at a 1-liter
volume batch scale, and in continuous flow modes of operation from bench-scale continuous
flow to a pilot scale using 20 kg/h slurry processing in a fixed-bed reactor with fully integrated
heat recovery. High conversion of a wide range of organic materials has been demonstrated.
Unit Operation 3b – By-Product Gas Purification: Fuel gas scrubbing is straightforward
using conventional refinery technology.

Page | 15
2. Project Objectives, Overview and Successes

Project Objectives

The purpose of this Project was to demonstrate a technically and economically viable
technology for converting renewable biomass feedstocks to sustainable and fungible
transportation fuels. The project would demonstrate new technology by constructing and
operating a pilot-scale integrated biorefinery in Hawaii that processed one dry ton per day of
biomass using pyrolysis and subsequent upgrading to transportation fuels. Specific objectives
of the project were to:

 Identify acceptable lignocellulosic and algae feedstocks that were sustainable,

currently available and/or whose availability could be increased in the future
 Demonstrate that these feedstocks could be pyrolyzed to an oil suitable for
further processing
 Demonstrate that pyrolysis oil could be successfully upgraded to transportation
fuel blendstocks
 Demonstrate that the resulting ultra-clean gasoline, diesel, and jet fuel range
transportation fuels meet specifications for use in existing infrastructure, such as
pipelines and engines
 Confirm that the economics of the process are acceptable

The anticipated outcomes of the Project were a validated process technology, a range of
validated feedstocks, product property and Life Cycle data, and technical and operating data
upon which to base the design of a full-scale biorefinery. The anticipated long-term outcomes
from successful commercialization of the technology were:

(1) The replacement of a significant fraction of petroleum based fuels with advanced
biofuels, leading to increased energy security and decreased carbon footprint; and

(2) Establishment of a new biofuel industry segment, leading to the creation of U.S.
engineering, manufacturing, construction, operations and agricultural jobs.

Project Scope

The scope of the Project was to design, construct and operate an integrated pilot biorefinery
using a variety of biomass feedstocks expected to be widely available for future commercial-
scale biorefineries. The pilot biorefinery would convert the biomass to transportation fuels,
including gasoline, diesel, and jet fuel range hydrocarbon.

The integrated biorefinery process comprises pyrolysis and catalytic upgrading technologies
and would operate on sustainable feedstocks. Data would be generated for detailed Life Cycle

Page | 16
Assessments (LCA) as well as current availability and outlook for selected waste agriculture
products, pulp, paper, woody biomass, dedicated energy crops, and algae.

The fuels would be analyzed and properties tested for their fit into the existing refinery
infrastructure. A consortium of US refiners and end product users would evaluate product
quality and performance via detailed characterization and engine testing. A plan would be
created to offer the process for commercialization based on the data generated during the
pilot testing.

The original project scope is depicted Figure 1.3 below.

Figure 1.3 – Original Integrated Biorefinery Project Scope

Steam
(flue gas)

During the project, two factors forced a change in the original Project Scope. First, the host
refinery site in Hawaii shut down for an undetermined amount of time. The shutdown would
deny the IBR project a source of hydrogen for hydroprocessing. The refinery remained
mothballed until being purchased by Hawaii Independent Energy.

Second, development and design issues with the UG-II hydroconversion technology prevented
completion within the timeframe of the project. A decision was made in Spring 2014 to not build
the UG-II upgrading technology at the pilot site in Hawaii, but instead continue hydroprocessing
development in UOP’s Research facility in Des Plaines, Illinois.

During Spring 2014, an additional option to complete the full integrated pathway from biomass
feedstock to transportation fuel was included in the project. The technology to co-process
pyrolysis oil along with vacuum gas oil (VGO) in a fluid catalytic cracking (FCC) unit was
proposed. This technology would produce a transportation fuel blendstock similar to the
downstream products of traditional FCC unit operation, but with a partially renewable

Page | 17
component. This solution was added to the project in Spring 2014 as an additional pathway to
produce transportation fuels. The revised scope of the project is shown in Figure 1.4 below.

Figure 1.4 – Revised Integrated Biorefinery Project Scope

Kapolei, Hawaii UOP Pilot Plants

UGII
RTP UG1 (pilot scale)
(Demo Scale) (Demo Scale) Two Stage
Stage
Biomass 168 GPD RFO 168 GPD RFO) Hydroprocessing
Hydroprocessing Fuel
1-ton/day 0.4 GPD/+1000 HOS
Rapid Metals
Thermal Finishing FCC
Processing Unit (pilot scale)
Unit (MFU) Co-processing
in Circulating Fuel
Riser (2-5%)

Project Site

Project partner Tesoro hosted the integrated biorefinery project operations at their refinery
site in Kapolei, Hawaii. This location was selected as the pilot site for access to refinery
infrastructure and local support for renewable fuels initiatives. For the hydroconversion pilot unit,
it was desirable to co-locate the plant at (or near) an existing refinery to maximize synergy with
existing infrastructure (hydrogen access, for example). In addition, refiner validation of the
renewable products along with off-take opportunities made co-location an ideal outcome.
Mid-project, Hawaii Independent Energy purchased the facility and continued to support the
IBR project to completion.

Page | 18
The IBR project site was located on the island of Oahu adjacent to Hawaii Independent
Energy’s refinery at Campbell Industrial Park, as seen in Figure 1.5 below.

Figure 1.5 – IBR site location in Hawaii

The site layout housed the RTP unit, UG-I metals finishing tankage, pyrolysis oil storage tanks,
lab and office trailers and space for the potential development of the UG-II hydroprocessing
equipment. This is seen in Figure 1.6 below.

Figure 1.6 – IBR site layout at Tesoro/HIE Refinery

Page | 19
Project Partners

The integrated biorefinery project was a complex undertaking with multiple aspects that were
similar to a greenfield commercial project and the scope of the project could not have been
achieved without the support of many vital partners. The level of experience each partner
provided was invaluable in bringing this project to completion. The project partners are listed
in Table 1.1.

Table 1.1 - Project Partners

Project Activity Project Partners

Project Director Dept of Energy, UOP

Project Management UOP, Ambitech, Group 70

UOP, Ambitech, Zeton, Group 70,

Construction Management
Tesoro/Hawaii Independent Energy

Start-up & Commissioning UOP, Ensyn

Site Operations UOP, Ensyn

Site Host Tesoro/Hawaii Independent Energy

Research Partner Pacific Northwest National Laboratory

Idaho National Laboratory, Ceres, Cargill,

Feedstock Suppliers
Imperium, Aquaflow, Grays Harbor Paper

Life Cycle Analysis Michigan Technological University

Page | 20
Value Proposition

The value proposition of an integrated biorefinery is to create second generation renewable

fuels that smoothly integrate within the transportation fuel industry.

First generation renewable fuels have limitations due to both feedstock choices and inability to
integrate within existing petroleum fuel infrastructure. First generation fuels must be blended
with petroleum products downstream at a terminal rack and require separate storage and
handling equipment. Plus, first generation oxygenated fuels will always be limited in their
blending levels.

1st Generation Biofuels Integrate Post-Refining

Exploration Refining Terminal Rack

1st Gen
Biofuels

The IBR project demonstrated second generation technology that can integrate into today’s
petroleum refining industry. Pyrolysis oil can be integrated upstream at the refinery to produce
renewable fuels through either hydroprocessing or co-processing in a FCC unit. This allows for
smoother integration with existing infrastructure at a lower cost. In addition, second generation
technology produces fuel blendstocks that are unconstrained by blending limits.

2nd Generation Pyrolysis oil Upgrading Integrates at Refinery

Exploration Refining Terminal Rack
Pyrolysis oil
Integration

The refining industry requires renewable fuels to satisfy both government mandates and
customer demands for clean burning fuels. To be successful, renewable fuels must be
economical within today’s fuel environment and maximize the use of existing refining equipment
and pipeline and infrastructure.

Page | 21
Project Successes and Challenges

As with any developing technology pathway, the IBR project had many successes but also
faced many challenges. Creating a successful path for taking biomass feedstocks all the way to
transportation fuels requires completion of multiple intricate tasks. At the time this project was
proposed, it was an ideal combination of existing RTPTM technology with developing pyrolysis oil
hydroconversion technology. Not unexpectedly, the majority of the challenges were found in
completing the developing hydroprocessing technology.

One of the primary successes of the project was the demonstration of upgrading pyrolysis oil to
100% neat transportation fuel and jet fuel blendstocks in a laboratory scale unit. The project
team was able to demonstrate that two-stage hydroprocessing of pyrolysis oil to traditional fuel
blendstocks is possible. The two-stage hydroprocessing approach is a vital step in the effort to
control the operating economics of pyrolysis oil upgrading.

A second success of the project was completion of 1000 hours on-stream for the first stage
hydrodeoxygenation catalyst. The first stage of upgrading is believed to be the most challenging
step in the hydroconversion process for converting pyrolysis oil to transportation fuels. Critical to
commercial viability is a catalyst system that is stable over years of operation. By completing
1000 hours of first stage testing, important information was gained with respect to catalyst
stability and the impact operating conditions have on catalyst stability.

The feedstock testing of more than 24 different biomass types provided important liquid and
energy yield for each feedstock versus its economic cost. This data will benefit the selection of
feedstock and site location for future facilities. During testing, important limiting factors were
identified when converting algae to pyrolysis oil. While the elemental composition of the algae
would suggest that it is a prime candidate for conversion to pyrolysis oil, the results were highly
viscous pyrolysis oil that resembled a “tar” or “taffy”. The algae-based pyrolysis oil was also
thermally unstable and quickly polymerized when exposed to heat.

The technology to substantially reduce the metals content of the pyrolysis oil was successfully
developed and demonstrated. At the site in Hawaii, the metals finishing unit (MFU) reduced the
metals content of pyrolysis oil from over 1000 ppm to less than 50 ppm. The metals removal
step is a key enabling factor for further upgrading of pyrolysis oil. If not removed, the metals in
pyrolysis oil could potentially contaminate the catalyst in either downstream hydroprocessing or
FCC co-processing units.

Successful testing of pyrolysis oil co-processing in an FCC laboratory plant was a big step
forward to completing the pathway from biomass feedstock to transportation fuel. Co-processing
testing resulted in the production of partially renewable gasoline and diesel fuel blendstocks that
would qualify for renewable fuel mandates. Positive economic results from this testing suggest
that co-processing is a near term solution for producing renewable fuels within an existing
petroleum refinery.

With a project of this complexity, there were many challenges and obstacles to overcome. The
development of the demonstration scale integrated biorefinery was similar in scope to a small

Page | 22
greenfield commercial facility. These challenges resulted in delays to the schedule and
contributed to changes in scope that affected the final outcome of the project.

The primary shortcoming of the project was the inability to successfully hydroprocess pyrolysis
oil to transportation fuel blendstocks at pilot plant scale. Hydroprocessing upgrading was the
large step-change technology at the center of the integrated biorefinery project. The difficulties
encountered in pyrolysis oil upgrading were the primary obstacle that resulted in project delays,
work extensions and project scope changes.

The inability to find a reliable solution to the polymerization of pyrolysis oil in a fixed bed reactor
was a critical setback. First stage hydrodeoxygenation was attempted many times using
different operating conditions and flow configurations and all tests ultimately ended with a rise
in reactor pressure due to plugging from polymerized pyrolysis oil.

Overall, the IBR project produced many important findings for the implementation of second
generation biofuels along with the documentation of many ideas tried and failed. The vital
learnings from this development will assist additional research for completing the pathway from
biomass feedstock to transportation fuel via upgrading.

Page | 23
3. Project Management, Contracting and Risk Mitigation
As the leader in petroleum and petrochemical refinery design, UOP has extensive experience
in project management. UOP’s Modular Plants group has successfully delivered over 200
processing units that included design, procurement, fabrication, logistics and installation
advisory services.

The project team for the integrated biorefinery project reached across multiple levels of UOP
to garner expertise from many different areas. Experienced personnel from Research and
Development, Technical Services, Field Operating Services, Modular Plants and Business
groups worked together to deliver a project that would meet the initial goals. In addition to
UOP staff, many different organizations brought vital expertise to the project. Project partners
included the Department of Energy, Ensyn, Hawaii Independent Energy, Tesoro, Pacific
National Laboratory, Idaho National Laboratory, Ambitech, Group 70, Zeton
and American Pipe and Boiler.

As project leader, UOP was accountable for project execution and successful completion of all
tasks. The development of different tasks of the project was divided between many different
project participants.

Project EPC and Contracting Process

The engineering, procurement, construction (EPC) and contracting was handled by UOP similar
to the process used by the UOP Modular Plants group to execute commercial scale projects.
This process utilized UOP Modular to project manage the detailed design, module fabrication,
shipping and site construction of the Hawaii IBR using multiple partners at off-site locations.

UOP Modular uses a proven project execution procedure for all large capital expenditure
projects. Contracts for design and fabrication of the process modules within the project were
conducted at fixed price and no schedule incentives were utilized. Detailed design and
fabrication packages were issued to fully define the job requirements and served as the basis
for contracting with Zeton and Ambitech.

Detail design and fabrication of the RTP modules was completed by Zeton using a process
design provided by UOP and Ensyn. Ambitech specified and procured the biomass handling
equipment upstream of the RTP modules. The process design of the metals finishing unit (MFU)
was by UOP with detailed design, fabrication and automation by REMCO Engineering. UOP
and Ensyn conducted design reviews for the RTP and monitored project quality assurance and
quality control (QA/QC) during module fabrication. UOP conducted similar design reviews and
monitored QA/QC during design and fabrication of the MFU modules.

Page | 24
The Hawaii project site detailed design was completed by Ambitech, while site construction and
integration of the RTP modules, MFU modules and associated biomass feed handling
equipment was completed by American Pipe and Boiler (APB). APB is a mechanical contractor
on the island of Oahu with refinery construction and maintenance experience and was utilized
for construction and contract maintenance tasks during the lifetime of the project.

Project Contracting Challenges

Issues did arise related to the site construction and operation of the project. The primary issue
for construction, operation and maintenance for the IBR project was the island location.
Compared to many locations in the mainland United States, developing a demonstration scale
project in Hawaii faces many challenges. Hawaii struggles with a lack of heavy industrial
infrastructure compared to the mainland and this presented multiple problems. There is a
general scarcity of experienced contractors able to perform crane operations, site grading and
utility work plus timelines to complete work are often extended.

The main challenge was related to maintenance where it was found that many parts were not
available locally and had to be ordered and shipped by air from the mainland. This increased
maintenance costs and resulted in greater downtime than if parts had been available locally.

The project team was also faced with the challenge of hiring qualified local Process Technicians
for a relatively short term project, but this item was successfully addressed through a
relationship with Leeward Community College, who has an excellent program of study in
Process Technology. This relationship assisted the Kapolei Pilot Plant Site Leader in arranging
interviews and led to the hiring a number of locally trained Process Technicians and
replacements for those who were later hired for full time positions by local industries.

Another challenge that arose was when REMCO Engineering, who fabricated and provided the
PLC for the MFU unit, went out of business in the summer of 2014. This occurred shortly before
commissioning and delayed MFU operations. However, a replacement vendor was identified to
assist with PLC programming issues and this delay did not impact the overall production
schedule.

Overall, the implementation strategy for developing a future integrated biorefinery project would
remain the same, although consideration would be given to placing such a facility in a more
established market location.

Page | 25
Risk Mitigation

A project of this size requires a structured risk identification and management program.
UOP uses a project Failure Mode and Effects Analysis system to rank project risks and
severity. The project team utilized this procedure for the integrated biorefinery project.

Two large potential risks were identified before the project began that ultimately impacted the
project; the possibility that successful upgrading technology would not available within the
project timeline and the possibility that overall schedule slippage could impact project financials.
The ultimate result of both of these incidents occurring resulted in changes to project scope and
cost for certain tasks.

At project conception, risk was managed through a process to limit cost overruns with defined
scope bid requests and fixed price procurement contracts. The project used the same pre-
certified, proven suppliers UOP-Honeywell has used for other large processing projects,
when available.

The risk of project cost and schedule delays was mitigated by the use of a Basic Engineering
Package developed by UOP and Ensyn, along with preliminary detailed engineering provided
by Zeton for the fixed cost of the fabrication of the RTP modules. Ambitech was hired to
complete the preliminary engineering for the remainder of the site operations outside battery
limits (OSBL) and provide a cost estimate using the ICARUS system. Three local Hawaiian
construction companies contributed local labor rates and efficiency factors for calculating all
of the project costs.

Ambitech completed the detailed engineering and purchased the OSBL equipment, including
the feed handling system, motor controllers, and laboratory and control room equipment.
Ambitech also assembled a full bid package for the site construction and created a Request
for Proposal for local mechanical contracting companies.

Of the risks successfully mitigated, the modular engineering bid process resulted in costs that
were within 10% of estimates for RTP module fabrication and site construction. The biomass
material handling system was pre-tested with specification-size material, so the operation would
perform as expected on site. The engineering site layout was designed using 3D CADD system
to ensure proper equipment fit.

For issues on site that occurred during mechanical completion, an experienced construction
manager was utilized to ensure proper work and safety procedures. To ensure familiarity with
the equipment layout and continuity, the operations site manager was onsite during a final
construction phase to assist with equipment integration and testing.

Page | 26
4. Project Scope Changes

There were a number of changes in project scope that occurred during the course of the
project. Some of the changes were related to refocusing efforts, while others addressed clear
roadblocks that required changes in project direction. Table 4.1 below depicts the major
scope changes.

Table 4.1 - Project Scope Changes

Original Scope Project Result

Four biomass feedstocks identified as primary
types for commercialization due to feed
Evaluate and identify multiple types of
properties and availability. Certain feedstocks
biomass feedstock for pyrolysis (RTP)
identified as unsuitable candidates
for RTP conversion.
Development of 1st stage hydroprocessing
Build pyrolysis oil upgrader unit (UG-II) unit incomplete within project timeframe.
at IBR project site Substituted upgrading pilot plant testing
as alternative.
Utilize industry partnerships to evaluate No demonstration scale volumes of fully
upgraded fuel products upgraded fuel produced
Evaluation of upgraded fuels No demonstration scale volumes of fully
by feedstock type upgraded fuel produced
Co-processing pilot plant testing brought into
Pathway for co-processing pyoil in
project to demonstrate near term
FCC unit not in original scope
commercialization pathway
Upgrading pilot unit (UG-II) never
LCA of demonstration scale upgrading to
constructed. Developed LCA based
transportation fuels
on lab scale results

Page | 27
5. RTP Operations

5.1 Introduction

This section discusses the operations of the RTP unit in Kapolei. Plant conditions and operations,
data collected, comparison of various feedstocks, and resulting lessons learned and
recommendations are discussed in this section.

A schematic flow diagram of the RTP unit is shown in Figure 5.1.

Figure 5.1 – RTP Flow Diagram

Figure 5.1 shows the RTP Reactor, the sand Reheater, and the downstream equipment for
separating liquid products from the recycle gas.

5.2 2012 Initial Operations

Initial RTP operations began in April 2012, when the staff arrived at site and started dry-out of the
unit. Initial troubleshooting was done in April and May, during which the torch oil system was
redesigned, and additional external insulation was added to the Reactor and Reheater. In June,
the first attempt to introduce biomass feed into the Reactor was made, and operations continued
until October, with a total of 10 runs conducted. No runs achieved a “lined-out” operation and the
longest run achieved was 24 hours. Many of the issues encountered such as excessive heat loss
were due to the small capacity design of the RTP unit for the demonstration scale throughput and
will not occur in commercial scale units. In October, the decision was made to hold RTP unit
operations until the Upgrader II unit design was complete.
Page | 28
5.2.1 2012 Objectives

The objective of the 2012 site operations was to demonstrate the RTP unit’s ability to process
hardwood biomass on a 1 bone dry metric tonne (BDMT) per day basis, achieving a target liquid
yield of pyrolysis oil of 4 Barrels per BDMT.

In addition, future construction of a hydroconversion unit was planned for the IBR site.
Hydroconversion was to be integrated with the RTP, SRU and MFU units to maximize the final
product yield, as opposed to maximizing the yield of a single process step.

5.2.2 Hardwood Biomass Runs 1 to 10

During 2012 operations, there were 10 RTP runs conducted with hardwood biomass as feed to
the Reactor. The first seven runs lasted less than 10 hours, and changes were made to the plant
during those runs in order to address issues that prevented continuous operations. During runs
8-10, the unit was able to run for an extended period of time, with runs 8 and 10 lasting about
24 hours.

[Link] Operating data

Despite troubleshooting efforts, no runs in 2012 achieved a lined-out operation; runs 8 and 10
reached the longest time on-stream of 23.2 hours. Table 5.1 summarizes the operating conditions
and the pyrolysis oil produced for runs 8-10.

Table 5.1 Operating Conditions and Pyoil Yields for 2012 Runs – Hardwood

RTP Biomass Run No. Total/Avg

Run Variable 8 9 10 Runs 8-10

Hours on Stream 23.2 13.1 23.9 60.2

Biomass Feed

BM Charged (lb) 1,984 1,272 1,948 5,204

BM lb/hr 85.5 97.1 81.5 86.4

BDMT per Day 0.88 0.99 0.83 0.88

Page | 29
5.3 2014-2015 Operations Summary

In June 2014, staff returned to the IBR site to restart RTP operations. In August 2014, start-up was
completed and the plant was released to UOP. The unit operated from September 2014 through
April 2015, during which all of the hardwood biomass was processed. During hardwood operations,
the unit saw many operational changes and equipment modifications, with the longest achieved
run lasting about a week. After hardwood operations were completed, initial attempts were made
to process softwood and switchgrass feedstocks, but after specification issues with the softwood
and switchgrass, RTP operations were terminated on April 14, 2015.

5.3.1 Hardwood Biomass Runs 11 to 25

During 2014-2015 operations, the RTP was able to process all of the remaining hardwood
biomass. In total, 20.4 BDMT of biomass were processed in 15 separate runs over 740 total hours
of run time. The seven longest runs processed 98% of the biomass, with the longest run in
February 2015 lasting 7 days.

[Link] Operating Data

Ninety-eight percent of the biomass was processed in seven of the 15 runs; biomass runs 16, 17,
20-22 and 24-25 were able to achieve lined-out operations over several-day periods. These seven
runs are summarized in Table 5.2.

Table 5.2 Operating Summary for Hardwood Runs

Biomass BDMT BDMT

Run No. Biomass per day

16 2.15 0.87

17 3.63 1.04

20 1.41 0.88

21 3.81 0.85

22 0.75 0.43

24 3.82 0.67

25 4.85 1.16

Total 20.42

Page | 30
5.3.2 Softwood and Switchgrass Runs

In March 2015, the processing schedule included runs for softwood, switchgrass and corn stover
feedstock. These feedstocks were sourced through Idaho National Laboratory (INL) by an
agreement with Pacific Northwest National Laboratory (PNNL). INL has successfully processed
various biomass feedstocks for many other Department of Energy projects and was utilized as a
resource for this project.

The RTP process requires a specific feedstock size and moisture content for all types of biomass.
After multiple test runs for preparation of the softwood, switchgrass and corn stover, INL concluded
their existing equipment was unable to meet the size specification for the material. INL was limited
in the ability to complete the necessary size reduction and as a result, some of the feedstock was
larger than specified and had “little-stick” morphology. Figure 5.2 shows the hardwood biomass
composition compared to the softwood and the switchgrass. After consultation with UOP, the
decision was made to send the feedstock as processed and attempt to make changes to the feed
handling system to incorporate the processing of these materials.

Figure 5.2 Comparison of Particle Size of Biomass Feedstocks

When this material was attempted to be processed, feed system problems prevented continuous
supply of biomass to the RTP unit and resulted in interruption of unit operation. Changes were
made to the feed system, including the attachment of a vibrator to facilitate the movement of the
packed biomass, but this still was not sufficient to successfully feed the RTP unit over an extended
time period.

Three limited timeframe operations with softwood were completed as shown below in Table 5.3.
The softwood runs show limited operation between 25 and 72 continuous hours before the
feedstock would compact in the feed bin necessitating a shut of the feed handling system. A
summary of the operating data is shown below.

Page | 31
 Table 5.3 Operating Summary for Softwood Runs

Biomass Total BDMT BDMT

Run No. Hours Biomass per day

26 56.0 1.00 0.45

27 25.2 0.45 0.48

28 72.5 1.30 0.50

Total 153.8 2.75

Additional attempts were made to run the unit using switchgrass, but similar feed system problems
were encountered. After numerous attempts and realization of the limited project time remaining,
the decision was cooperatively made between UOP and the Department of Energy to shut down
the RTP unit operation portion of the IBR project.

5.3.3 Feedstock Analysis

As previously discussed, the hardwood biomass had a much smaller particle size than the
softwood and switchgrass feeds. A particle size distribution analysis was done for two samples of
hardwood feed, and the results are shown below. Particle size did not exceed 1 mm in either case.
These results are shown in Figure 5.3 below.

Page | 32
 Figure 5.3 Hardwood Biomass Particle Size Distribution

PSD analyses for softwood, switchgrass, and corn stover were provided by Idaho National
Laboratory (INL); however, these analyses may not accurately represent the size distribution of
the feedstocks that were received at the IBR site because of the long and narrow shape of the
particles. The plots below show that the size distributions showed no particles greater than a few
millimeters in size.

Figure 5.4 - Softwood Biomass Particle Size Distribution

Particle Size Distribution Results

Screens: Grinder 2", Hammermill1/2"
Sample UOP Pine, Prerun #3, Compiled Results
40.0%
35.0%
30.0%
% Retained on Screen

25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
-5.0%
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
Screen size (mm)

Page | 33
 Figure 5.5 - Switchgrass Biomass Particle Size Distribution

Particle Size Distribution Results

UOP Switchgrass Prerun #1: No Screen in Hammermill, 1" in BG480.

30.0%
25.0%
20.0%
% Retained on Screen

15.0%
10.0%
5.0%
0.0%
-5.0%
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
Screen size (mm)

From the pictures shown in Figure 5.2, it is clear that some of the sticks are much longer than 1
mm, some even up to one inch long. The method for analyzing PSD uses a screen that shakes on
a latitudinal axis and allows particles to fall through; this allows long 1” sticks to stand on end and
fall through the screen. This off-spec particle size ultimately caused operational problems with
consistent softwood and switchgrass feed to the Reactor and complete conversion of biomass to
pyrolysis oil.

Page | 34
5.4 Lessons Learned and resulting recommendations

1. Biomass specifications for small scale equipment should include L/D ratio for the particles.
The hardwood “sawdust” morphology worked well in the IBR RTP unit, but the “little stick”
morphology of softwood and switchgrass did not. Issues were encountered with
compaction of the sticks in the feed bin and feed conveyor.
2. Screens to collect “trash” in biomass feed should be installed at the inlet to feed bins.
A screen was installed on the IBR feed handling system after foreign objects inhibiting
the flow of biomass were discovered in super sacks.
3. Support for the DCS and PLC systems should be supplied locally to ensure ease of
support.
4. Demonstration plants should not be located in relatively remote locations. The Hawaiian
island location made cost, contracting, shipping, staffing and maintenance more difficult.
Plant modifications could have been done faster if support services had been more
readily available.
5. Estimates of heat loss should be based on the size and operation of the plant. The RTP
unit was non-adiabatic, requiring additional heating cables and insulation to be added in
order to reduce heat losses. Hawaii experiences relatively small temperature variation over
24 hours, but plant heat-ups still took noticeably longer at night than during the day, even
after the additions of heating cable and insulation.

5.4.1 Potential commercial relevance of technical findings

RTP is a commercialized technology, and the IBR RTP technical findings are not necessarily
indicative of expected commercial operations of full scale unit. Many of the issues identified during
the project were a direct result of accommodations made for the lower feedstock throughput and
the size of modules for shipping to Hawaii. Commercial scale designs of the RTP unit do not have
these same limitations.

Page | 35
6. Upgrader- I
The Upgrader-I process technology encompasses both solids and metals removal from pyrolysis
oil. Upgrader-I enables further upgrading of pyrolysis oil in catalytic processes for which both high
solids and metals content would be of concern. Upgrader-I consists of two processes; solids
removal in the Solids Removal Unit (SRU) and metals removal in the Metals Finishing Unit (MFU).

6.1 Solids Removal Unit

6.1.1 Operations Summary

The SRU was first operated at the IBR project site in 2012 with pyoil provided by Ensyn. The DCS
sequencing was optimized, and recommendations were made for equipment modifications that
were implemented in 2014. During 2014 and 2015 IBR operations, the SRU was run 85 times to
produce 2,776 gallons of low-solids filtrate. Hardwood-derived pyrolysis oil runs were consistent
and no major operational issues were encountered. The solids contents of the filtrates were less
than 0.1 wt-% for all 85 runs. All of the pyrolysis oil produced in the RTP unit was processed in the
SRU to remove solids. The target of filtrates containing less than 0.1 wt% solids was met in all
cases. Each filtration cycle required periodic blowdown of filter cake to reduce pressure drop.

[Link] Equipment overview

The SRU Filter is designed as a batch process capable of filtering of pyrolysis oil cycling twice an
hour for an estimated throughput of approximately 40 gallons per hour. The purpose of the SRU
filtration system is to remove the solids from the pyrolysis oil. The filter consists of two main
sections; a bottom feed portion or heel and an upper portion where the filtration occurs.

As pyrolysis oil passes through the inside of the filters, cake (solids) builds up on the filter resulting
in an increase pressure across the filter. Once the pressure drop has reached the filter design, the
cake that has built up on the filters must be removed. After completing a continuous filtration cycle,
a cake drying technique is to be employed to capture the additional filtrate (filtered pyrolysis oil)
from the wet cake. The final step in the process is discharging and collecting the cake from the
filter elements.

Page | 36
6.1.2 Process overview
The SRU Filtration System is divided into the following steps:

A. SRU Filter Preparation

Before filtering can commence, the pyrolysis oil must be preheated to lower the viscosity of the
liquid, which improves filtration capability. In order to accomplish this heating, the pyrolysis oil must
first pass through a heat exchanger on its way to the Product Filter. To avoid heat losses while in
the filter, self-regulating heat tracing was coiled around the filter’s exterior. This ensures the
temperature of the liquid does not drop dramatically as it passes through the unit.

Agitation is required so that the pyrolysis oil is well mixed before and during filtration. When
pyrolysis oil sits, there can be some degree of separation which occurs as the denser fraction
settles to the bottom of the feed vessel and the less dense fraction tends to rise.

B. Start SRU Filter Operation

A pump is used to transfer well mixed pyrolysis oil from a feed vessel, through a heat exchanger
and provide the filter with a heated supply of pyrolysis oil. Pyrolysis oil of very low solids content
emerges from a filtrate port and is collected in the Filtrate Tote. The filter consists of two main
sections; a bottom section to distribute the feed between the filter elements and an upper section
housing the filter elements.

C. Cake Drying
After completing a continuous filtration cycle, a cake drying step is to be employed to capture
additional pyrolysis oil filtrate from the wet cake. The liquid can then be returned to the feed tote
for use in subsequent cycles. When the air passes through the cake and through the elements,
it forces some of the liquid out of the cake.

It is possible to over-dry the cake, resulting in extremely dry cake which may become stuck in the
filter, so caution should be taken to avoid this possibility. It is advised that for the first few drying
stages, the top of the Product Filter be removed and each element be inspected to ensure the
cake is not too dry. Proper drying should yield a cake roughly the consistency of play-dough and
should be easily formed. If the cake is crumbly or chalk-like it has likely been dried too long and
the next drying cycle should be adjusted accordingly.

D. Cake Discharge
The final step in the process is to discharge and collect the filter cake. First, pressurize the filter
with air. Then, open the butterfly valve to the Solids Collection Drum. Air will flow through the filter
elements from the outside to the inside and out the bottom of the filter, carrying with it the cake into
the Solids Collection Drum. The gas will then exit the cake discharge vessel through the vent on
the drum

Page | 37
Figure 6.1, shown below, diagrams the SRU process steps.

Figure 6.1 - Diagram of Solids Removal Unit Process Steps

Page | 38
[Link] Run Summary

Operations in 2014 and 2015 followed the procedure established during the 2012 shakedown.
During these operations, there were a total of 85 SRU runs, producing 2,776 gallons of hardwood-
derived pyrolysis oil filtrate. Table 6.2 summarizes the SRU runs corresponding to each RTP
hardwood run.

Table 6.2. Summary of SRU Operations

RTP Filtrate Filtrations Gal/ Wt% Solids

Run
Gallons Required Filtration Feed Filtrate
No.

16 232 9 25.8 0.02

17 494 15 32.9 0.04

20 173 5 34.6 0.03

21 534 16 33.4 0.02

22 106 3 35.4 0.04

24 559 18 31.1 0.06

25 678 19 35.7 0.06

2,776 85 32.7 0.039

Page | 39
6.1.3 Accomplishments and goals

Between 2012 and 2014 operations, all major issues regarding Solids Removal Unit equipment
were addressed. The SRU was successful in its ability to reduce the solids content of pyrolysis oil
to <0.1 Wt-%. The major issue encountered during 2014 hardwood biomass runs was overly wet
cake, causing a reduction in the possible filtrate yield. While extensive testing was done in 2012 to
ensure optimum operating conditions to maximize filtrate yield, the variability between runs made it
difficult to ensure complete cake drying for every run.

The most significant issue to address was the adjustment of filtrate and regeneration times for
optimal filtrate yield and cake consistency. The slightly lower than design filtrate yield was likely
a result of excess wetness in the cake, and further drying could yield a slight increase in liquid
product.

Page | 40
6.2 Metals Finishing Unit
6.2.1 Background on Ion Exchange

Biomass-derived pyrolysis oil is a complex, highly oxygenated organic liquid containing metals and
other impurities. The metals present in the biomass feedstock, including alkali metals (sodium,
potassium etc.) and alkaline earths (calcium, magnesium), are largely retained in the as-produced
pyrolysis oil. Metals dissolved in the biomass-derived pyrolysis oil contribute to ash content of the
oil upon combustion and may be subject to environmental restrictions. In addition, when the
pyrolysis oil is used as feedstock in catalytic processes to make transportation fuel, the metals in
the oil may foul downstream equipment and are expected to inhibit or deactivate the catalysts.
The removal of metal cations from pyrolysis oil is therefore important for widespread utilization as
a biofuel.

Prior to initiation of the IBR project, UOP completed research scale work to demonstrate
demetalization of pyrolysis oil. A target of 100 weight ppm total metals (from feedstocks containing
>1000 ppm) was set as the initial technical objective.

In this work, it was demonstrated that pyrolysis oil can be substantially demetalized by ion
exchange with strong acid ion exchange resins. While ion exchange resins are well known to
remove metals from aqueous solutions, they have not typically been as effective in removing
metals from organic streams and were not previously reported for biomass-derived pyrolysis oil. In
contrast to typical organic streams, pyrolysis oil contains significant amounts of water (up to 30%
by weight) and the metal species are typically cationic – both factors that suggest ion exchange
might be possible. On the other hand, specific challenges for ion exchange of pyrolysis oil include
the relatively high oil viscosity and limited thermal stability of the feedstock.

It was found that with certain technical adjustments, particularly to the regeneration scheme,
metals could be effectively removed from hardwood and softwood -derived pyrolysis oil by ion
exchange over strong acid resins. Typically the effluent contains less than 20 ppm total metals
(excluding phosphorous, boron and aluminum which are not effectively removed by cation
exchange. Boron and aluminum are not typically present in high concentration in woody biomass.).
Over 50 adsorption/regeneration cycles were demonstrated at laboratory scale.

6.2.2 Metals Finishing Unit Design and Flow Scheme

The metals finishing process was designed for the removal of metals, primarily alkali metals and
alkaline earth metals from pyrolysis oil using the ion exchange process detailed in the previous
section. Once the capacity of the resin bed to exchange protons for metals becomes exhausted,
the resin is regenerated.

The Metals Finishing Unit (MFU) at the IBR project site was designed by and purchased from
REMCO Engineering using the ion exchange and regeneration processes developed by UOP. It is
controlled by a Programmable Logic Controller (PLC) control system that can be accessed either

Page | 41
through a remote computer or on a Human-Machine Interface (HMI) located on the unit itself. The
unit consists of two vessels, either of which could be used during metals removal, 10 tanks for the
regeneration process and three tanks for the feed and product. The latter includes a tank for the
feed that receives the product and a final storage tank where the product is stored once confirmed
that the metals content is sufficiently low. A block flow diagram for the MFU is shown in Figure 6.2.

The MFU at the IBR site was designed to run in batch mode and consists of two different size
vessels. The two vessels, each containing strong acid ion exchange resin which has the capacity
to exchange protons on the resin surface for metal ions contained within the pyrolysis oil, were
sized so that a batch was either run at design, processing approximately 2000 gal of pyrolysis oil,
or in a smaller quantity for research purposes. The target when loading the vessels was to have
the resin loading in the small vessel be 30% of that in the large vessel and thus have the capability
to process batch sizes approximately 30% of the design volume. Commercially, the MFU will run
continuously and once the capacity of the resin bed to exchange protons for metals is exhausted,
the pyrolysis oil feed will be switched to the remaining ion exchange vessel. While the pyrolysis oil
is exchanging metals and protons in the secondary ion exchange vessel, the ion exchange resin in
the primary vessel will be regenerated. Both vessels will be equally sized.

Figure 6.2: MFU Block Flow Diagram

Pyrolysis Oil Processing Regeneration

Filtered Recycle EtOH
Aqueous Effluent
Pyrolysis
Oil Fresh EtOH
Organic Effluent

Heater Air
EtOH Rinse H2O

Fresh
H2O

Recycle Regeneration
Chemical 1

MFU MFU Fresh Regeneration

Chemical 1

Recycle Regeneration
Chemical 2

Fresh Regeneration
Finished Chemical 2
Pyrolysis
Regeneration
Oil Chemical 2 Rinse H2O

Page | 42
Each batch consisted of pyrolysis oil processing and a subsequent regeneration to restore the
capacity of the resin for the next batch. Pyrolysis oil processing was completed in downflow
operation and resin regeneration in upflow operation as depicted by the arrows in the block
flow diagram.

6.2.3 Metals Finishing Unit Operation

[Link] Equipment Shakedown, Resin Loading and Chemical Inventory

The MFU skid arrived at the site during the first quarter of 2012. During 2012 operations, the
MFU was tested with water to check the PLC programming for the pyrolysis oil processing and
regeneration steps and to test the equipment. The MFU vessels were not loaded with resin and no
chemicals other than water were run through the system. Prior to shutting down the site in 2012,
all water was removed from the MFU system.

MFU shakedown resumed in the fall of 2014. The vessels were loaded with gravel and resin and
shakedown continued with water. Errors were found in the PLC code and corrected. Once the
changes were completed, the regeneration chemical tanks were inventoried and the resin was
prepared for use by completing a partial regeneration to ensure the resin was at capacity in the
acidic form.

[Link] Vessel Loading

Both the small and large vessels were loaded in early October 2014 with a strong acid resin.
Prior to loading resin, gravel was loaded into the bottom of the vessels to sufficiently cover and
protect the distributor at the bottom of the vessels. After the first few batches and subsequent
regenerations, it was noticed that resin was exiting the vessel during upflow processing steps.
Resin suspension during the upflow regeneration steps was found to be significant and new
loading volumes were determined for the two vessels and the resin bed heights were adjusted
accordingly.

[Link] Pyrolysis Oil Processing

Between batches, the resin was stored in water. Prior to starting a batch, the vessel was drained
using forced air and then rinsed with ethanol. After the ethanol was drained from the vessel,
finished pyrolysis oil that had already been processed through the MFU was used to reverse-fill the
bed with upward flow to remove any air pockets that are present. Once the vessel had filled with
the finished pyrolysis oil, a switch was made to process the unfinished pyrolysis oil (the SRU
product) in downflow operation. The batch size for pyrolysis oil was estimated using the metals
content of the pyrolysis oil and the capacity of the fresh (or freshly regenerated) resin. Batch size
varies between feedstocks as the pyrolysis oil metals content is dependent upon the starting
feedstock. For the hardwood-derived pyrolysis oil produced at Kapolei, the metals content was
slightly greater than 1000 wt-ppm. The target for the MFU was to achieve product containing less
than 100 wt-ppm total metals.

Page | 43
[Link] Resin Regeneration
Once the capacity of the resin bed to exchange protons for metals was exhausted, the resin was
regenerated. The multi-step regeneration process restores the capacity of the resin and has been
demonstrated at laboratory scale for over 50 adsorption/regeneration cycles. After completion of
the regeneration process, the resin was rinsed and stored in water until the next batch of pyrolysis
oil processing was to begin.

6.2.4 Metals Finishing Unit Results

The hardwood-derived pyrolysis oil processed in the MFU contained slightly more than 1000 wt-
ppm metals. Target metals removal was to have a finished product containing less than 100 wt-
ppm. Results are shown in Table 6.3.

Table 6.3: MFU Metals Removal Results

Run 1
110 min on Run 2 Run 3
MFU Feed* stream Product Product
Metal wt-ppm wt-ppm wt-ppm wt-ppm
Al 17 11 1 2
Ca 252 148 11 2
Cr 1 1 2 2
Co <1 <1 <1 <1
Cu <1 <1 <.1 <1
Fe 86 52 11 4
Pb <2 <1 <1 <1
Mg 96 59 3 <1
Mn 28 18 1 <1
Mo <1 <1 <1 <1
Ni 8 4 1 <1
K 436 262 10 1
Na 31 27 7 8
Sr 2 1 <1 <1
Sn <2 <1 <1 <1
Ti <1 <1 <1 <1
V <1 <1 <1 <1
Zn 21 13 17 <1
P 40 37 33 27
Total Metals 1016 633 94 47

*This feed sample was taken prior to Run 1. All hardwood-derived pyrolysis oil feed samples had similar feed
metals content. An exception is the Run 2 feed as it re-processed the Run 1 product; starting metals level for
that run was approximately 600 wt-ppm.

Page | 44
The initial batch, MFU Run 1, was sampled downstream of the MFU throughout the duration of the
run. Other than the initial material to pass through the vessel; the samples taken contained more
than 100 wt-ppm metals. Since this was the first run to be processed, there was no finished oil
available to remove air bubbles from the bed by reverse filling in upflow direction. Downflow
pyrolysis oil processing was initiated without completing the upflow step and it is likely that
channeling occurred leading to insufficient metals removal.

The resin was regenerated and the batch was reprocessed in Run 2 with successful metals
removal to less than 100 wt-ppm total metals in the product. Run 3 was the most successful run
with metals removal to less than 50 wt-ppm.

6.2.5 Metals Finishing Unit Conclusions

Ion exchange for removal of metals to acceptable levels for further upgrading to transportation fuel
blendstocks was proven at laboratory scale and validated at demonstration scale. A total of 1078
gallons of hardwood-derived pyrolysis oil were successfully processed to contain less than 100
wt-ppm total metals. One of the runs successfully reduced metals content to less than 50 wt-ppm.
Regeneration was successfully completed twice to restore the capacity of the resin between
batches.

Page | 45
7 Upgrader II Process

7.1 Introduction

This section discusses the Upgrader-II process technology, which is based on a two-stage, fixed-
bed reactor configuration to upgrade pyrolysis oil to fuels by hydroprocessing. Background on
upgrading, process concept selection, first-stage pilot plant work, catalyst development and
second-stage upgrading work are covered here.

7.2 Background

Pyrolysis of biomass to produce an oxygenated liquid product (pyrolysis oil), used for heating fuel
and as a food additive, is commercial technology 1 and has attracted significant attention as a
potential source of renewable transportation fuels 234. The essential difference between bio-oil and
petroleum is the oxygen content. The high water and organic oxygen content is derived from a
range of oxygen-containing functional types generated by the breakdown of the biomass
components. The bio-oil oxygenates are inherently polar and hydrophilic, which results in high
water solubility in the bio-oil. As-produced pyrolysis oil typically contains 20-30 Wt% water. Some
amount of water is apparently necessary for formation of a stable oil, and removal of the water (i.e.
by distillation) to very low levels without replacing with another suitable light oxygenate results in
an intractable sludge. The large water fraction dissolved in bio-oil added to the water formed by the
hydrotreating reactions, as well as the presence of organic acid functional types leads to a much
different processing environment than is found in petroleum processing.

Therefore, the ejection of oxygen from the feed molecules is the key catalytic function required for
bio-oil upgrading to hydrocarbon fuels. The two main mechanisms for oxygen ejection from bio-oil
are hydrodeoxygenation, with the oxygen removed as water, and decarboxylation, where the
oxygen is removed as carbon dioxide. Specific feed components such as carboxylic acids are
particularly susceptible to conversion by decarboxylation, while others, such as phenols are
expected to proceed mainly through hydrodeoxygenation. But the complexity of pyrolysis means
both pathways are likely to be important for a large fraction of the bio-oil.

These are not typically dominant reaction pathways for conventional refinery hydroprocessing and
so the development of catalytic technology for biomass specific feedstocks needs to consider

1
Mohan, Pittman, Jr., and Steele (2006), Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review, Energy and Fuels,
20 (3), 848 -889
2
Elliott, D.C.; Baker, E.G. (1987) Hydrotreating Biomass Liquids to Produce Hydrocarbon Fuels. Energy from
Biomass and Waste X, Chapt 42, pp 765-784; IGT Chicago.
3
Elliott, D. C.;Beckman, D.;Bridgwater, A. V.;Diebold, J. P.;Gevert, S. B.;Solantausta, Y. (1991), Energy and Fuels 5,
3, 399 – 410
4
Elliott. (2007). Historical Developments in Hydroprocessing [Link] and Fuels 21(3):1792-1815

Page | 46
catalyst formulations and process flow schemes that are differentiated from those developed for
essentially water-free petroleum processing.

7.3 UOP/PNNL/NREL CRADA

Prior to the IBR pilot scale project, UOP investigated pyrolysis oil upgrading via two stage
hydroprocessing in a Cooperative Research and Development Agreement (CRADA) Project with
Pacific Northwest National Laboratory (PNNL) and the National Renewable Energy Laboratory
(NREL). Pyrolysis oil derived from several biomass feedstocks including mixed wood residues,
poplar, and corn stover were catalytically hydroprocessed in laboratory scale equipment to
produce deoxygenated hydrocarbon liquids for analytical characterization. Basic properties of
the gasoline fraction produced from corn stover derived pyrolysis oil are shown in Figure 7.1.

Figure 7.1: Composition, Octane, and Distillation of Corn Stover-derived Gasoline 5

Typical
Gasoline
43.2
-
4.1
6.9
37.7

These analyses confirmed that the pyrolysis-derived gasoline product contained the same
hydrocarbon molecules as petroleum derived gasoline and, due to its specific composition, had
calculated octane values (RON, MON) and boiling points compatible with direct blending into
existing motor gasoline pools. Sufficient experiments were performed to confirm that petroleum-
quality gasoline, jet fuel range hydrocarbon, and diesel blend stocks could be produced from all
feedstocks tested.

Product yields and hydrogen requirements were also measured during the laboratory experiments.
Based on this information, mass and energy balances, equipment requirements and sizes, capital
investment, and cost of production were estimated for a commercial scale plant. Several different
scenarios were examined. Depending on feedstock type, pretreatment, and process operating
severity, yield potential ranged from 50-100 gallons of advanced biofuel per dry tonne of biomass.

5
The Octane Numbers are reproduced as disclosed in the CRADA report. We do not have an explanation for the
difference in the Calculated and Measured Motor Octane Number (MON).
Page | 47
Biofuel cost of production varied from $1-3/gallon depending on feedstock price, yield, and co-
product credits. Non-proprietary technical information related to conceptual plant design, scale-up,
and cost can be found in DOE Report PNNL-18284 (February 2009).

Sufficient technical information was gathered to perform preliminary life cycle analysis (LCA)
studies on biofuel derived from forest residues and short rotation forestry crops such as hybrid
poplar and willow. The LCA results indicated that the commercial production could meet the
targeted green house gas (GHG) savings required for an advanced biofuel relative to petroleum
gasoline (>60%).

A key learning from this project, however, was that pyrolysis oil upgrading by hydroconversion,
while technically feasible, suffered from poor stability due to fouling in the first catalyst bed. Run
lengths were limited to at most 1000 hours on stream (HOS) and the product quality changed over
the course of the run indicating catalyst deactivation. Upon end of run the catalyst was noticeably
fouled with carbonaceous material. Many research groups have published studies describing the
likely pyrolysis oil instability (fouling) mechanisms. 6

While coke formation from petroleum is generally either a buildup of aromatic structures on
catalyst surfaces or carbon formation on metallic sites at high temperatures, the fouling in bio-oil
processing is associated with the low-temperature range of heat-up in the processing system.
Bio-oil plugging is generally found in the front end of the catalyst bed wherein the bio-oil is passing
through heat-up range of 150-250ºC. Deposition on the catalyst of bio-oil contaminants, such as
pyrolysis char and metals (alkali, alkaline earth, transition metals, etc.) may contribute to catalyst
deactivation by blocking catalytic sites. Carboxylic acids, such as acetic acid, may play a role, but
condensation of key chemical species present in the pyrolysis oil, including aldehydic functional
groups, derived from carbohydrate structures in the biomass, with lignin-derived phenolics lead to
phenol-formaldehyde type cross-linked structures, were proposed to constitute a major fouling
mechanism.

Implementation of technology solutions to address the thermal polymerization of highly reactive

constituents of pyrolysis oil was identified as a key technical objective of the IBR project.

7.4 Lead Concept Selection

The purpose of the IBR program was conversion of pyrolysis oil to transportation quality fuels.
Stable hydroconversion operation was identified as one of the key technical challenges based on
past work both within UOP and as reported in the general literature. The main hypothesis guiding
the project was that during fixed-bed hydroconversion operations competing reactions, namely
thermal oligomerization, compete with pyrolysis oil hydroconversion. The approach to address this
relied on acceleration of the hydroconversion rate relative to thermal oligomerization rate. In

6
Hu, X.; Wang, Y.; Mourant, D.; Gunawan, R.; Lievens, C.; Chaiwat, W.; Gholizadeh, M.; Wu, L.; Li, X.;Li, C.-Z.;
AIChE Journal, 2013, 59(3), p. 888-900.
Page | 48
principle this could be accomplished either by inhibition of the thermal reaction, or promotion of the
first stage hydrogenation by improving catalyst activity or optimization of process variables.

7.4.1 Lead Concept Selection – Initial cresol dilution with full product recycle

Co-feed of a diluent stream with the pyrolysis oil was one of the more promising approaches
evaluated prior to the IBR project. Dilution of feedstock with product recycle oil is commonly
practiced in existing commercial hydroprocessing units. Dilution serves to ameliorate temperature
rise due to exothermic hydrogenation reactions and can improve hydrogen availability. In the case
of pyrolysis oil it was expected that dilution of reactive components should inhibit higher order
oligomerization reactions relative to first order hydroconversion. Therefore, product recycle was
included as a likely component of the initial process flow scheme. However, proper demonstration
was hindered by the limited availability of fully or partially hydrotreated pyrolysis oil product suitable
for use as a co-feed for fresh pyrolysis oil.

A number of surrogates for product recycle was tested including ethanol (partial deoxygenation
surrogate) and toluene (full deoxygenation surrogate). In the early testing, however, while benefits
were seen with certain co-feeds, overall process stability was either not sufficient (i.e. with
hydrocarbon co-feed) or not economical (alcohol co-feed with or without prior esterification). 7

Upon further consideration it was recognized that alkylphenols, such as cresol, would be a better
chemical match for a co-feed stream. Oxygen content, mutual solubility and vapor pressure were
more consistent with the properties of pyrolysis oil. Technical grade cresol, a mixture of C6-C9
alkyl phenols (C7 rich), was identified as a suitable surrogate to test. It was found that cresol was
miscible with pyrolysis oil at all concentrations and that the blends prepared showed substantially
improved thermal stability compared to the starting bio-oil. Furthermore, in batch autoclave testing
tars were eliminated and solids reduced compared the corresponding runs with pyrolysis oil only,
or with pyrolysis oil and hydrocarbon co-feed operation.

7.4.2 Lab Plant demonstration

With the success of batch processing of pyrolysis oil in the presence of cresol, testing was
implemented in a continuous fixed-bed laboratory plant. The laboratory plant utilized was a
simplified version of a typical pilot plant. Feed streams, catalyst loading and reactor temperature
and pressure ranges were in line with what would be expected in conventional hydrotreating pilot
plant operation. The main difference is somewhat limited automation of plant control and data
acquisition.

Pyrolysis oil first stage hydroconversion with cresol dilution (1:2 py oil/cresol pre-blended) at
startup was successfully demonstrated over sulfided base metal catalyst. Reaction conditions
were such that a two phase product mixture was obtained. The organic phase, while still rich in
oxygenated components (including co-feed cresol) was less dense than the water phase. After
80 hours on stream (HOS) the cresol/pyrolysis oil blend feed was replaced with a feed blend

7
During execution of the IBR project a patent application was published describing first stage hydrodeoxygenation of
pyrolysis oil in the presence of alcohol diluent/co-feed. US20130225887, Van Broekhoven, E. H.; et al. inventors
Page | 49
consisting of pyrolysis oil and the organic plant effluent (1:3 py oil/diluent) to simulate product
recycle. This semi-recycle operation was continued for two additional feed blends for a total run
length of over 400 HOS. This was by far the longest run achieved in internal UOP testing up to that
point in time. Overall approximately 3700 cc (4700 g) of py oil was fed to the reactor with ~1730 cc
(1800 g, 27.7% of total feed) of mixed cresols as initial co-feed.

7.5 Pilot Plant (Des Plaines)

A full sized pilot plant was made available to verify the lab plant result. The pilot plant was set
up in a configuration that included reactor effluent recycle and on-line density measurement for
conversion control. It also had a temperature profiler so the full reactor temperature profile could
be monitored. During the initial period of operation there were a number of equipment and
operational issues. Once the initial equipment issues were resolved, the reactor effluent recycle
concept was tested with variation of the temperature, pressure, and recycle gas ratio. Experiments
were done with sulfur doping and ethanol doping the feed, which did not help to mitigate the
plugging. Liquid recycle ratios as high as 16 also did not prevent plugging.

The objectives of the pilot plant test program were to determine the optimum recycle:feed ratio,
demonstrate that 100% pyoil could be fed to the plant at this ratio, confirm the product yields and
composition, identify the concentration and product stream for the feed impurities, and determine
the optimum means of separating the aqueous and organic phases in the liquid product. As the
testing was conducted, mitigation of the plant plugging problems became a major focus of the pilot
plant operations.

7.6 Pilot Plant Configuration

The pilot plant was revamped to accommodate the pyrolysis oil hydrodeoxygenation study.
The plant had a single down-flow reactor. There was a single liquid feed system with a metals
guard bed containing ion-exchange resin (this treater was removed in later runs).
Hydrogen and liquid recycle mix with fresh feed in a static mixer upstream of the reactor.
There was a three phase separator downstream of the reactor to separate the product gas
(mostly H2), organic liquid product, and aqueous product.
In general, the operating conditions were such that the organic phase was lighter than the water,
but it should be noted that the plant is capable of operating with the organic phase at the bottom of
the separator (i.e. heavier than water). Plant pressure is controlled by a regulator on the net gas
from the high pressure separator.

Page | 50
7.6.1 Pilot Plant Modifications

There were several modifications made to the pilot plant flow scheme. Most of the modifications
were in response to the extensive operational problems related to plugging in the reactor. The
modifications included replacement of the style of pressure regulation device used to minimize
fouling of the device. The feed and recycle pump systems were modified to improve flowrate
control. The feed and recycle control values were replaced due to corrosion or erosion of the
valve trims.

7.6.2 Run Summaries

The purpose of later runs in the pilot plant was to evaluate the feasibility of the proposed process
for first stage treatment of pyrolysis oil by hydrotreating at mild conditions with dilution from liquid
product recycle. Runs 2383, 2385, and 2387- 2389 did not achieve sustained operation at the
target conditions (neat pyoil feed with organic liquid product recycle) due to plugging at various
points in the plant

7.6.3 Plug characterization

Two different types of plugs were formed in the pyrolysis oil HDO process. One plug is very hard
and is insoluble in practically every solvent. The other plug is softer and is partially soluble in some
organic solvents, including methanol and chlorinated solvents. These plugs were characterized by
a number of different techniques which showed a difference in degree rather than a difference in
kind between the two types of plugs. The key information pointed to formation of a lignin-derived
phenol-formaldehyde resin.

A generic phenol-formaldehyde (Novolak) resin is shown in Figure 7.3. Phenol has three different
reactive positions, the two ortho positions (adjacent to hydroxyl group) and the para position.
Because of the multiple reactive sites, these polymers can easily branch or cross-link. Lignin
consists of many different electron-rich arenes, and these arenes have favorable reactivity in these
polymerization reactions because of their electronic properties. In fact, lignin has been used as
a cheap additive to Novolak resins in place of phenol. 8 Figure 7.4 shows some of the lignin
monomers that react with aldehydes. It is likely that dimers and other oligomers containing these
monomers also react with aldehydes.

8
Effendi, A.; Gerhauser, H.; Bridgewater, A. V. Renew. Sust. Energ. Rev. 2008, 12, 2092.
Page | 51
 Figure 7.3. Basic phenol-formaldehyde polymerization reaction

R
R R
R

HO OH
more reactive
more reactive
ortho position
ortho position
OH
OH OH OH OH OH
OH OH
CH 2O R

HO OH
OH OH
R R
Phenol:CH >1
Phenol:CH2O > :: Novolak
Novolak resin, usually H ++ catalyzed
resin, usually catalyzed
-
<1
Phenol:CH2O <
Phenol:CH :: Resol
Resol resin, usually OH
resin, usually - catalyzed
OH catalyzed HO OH

Figure 7.4. Monomers that react in phenol-formaldehyde polymerizations

OH
OH OH
OH OH
OH OH
OH
H3CO H3CO OCH3 H 3CO O
OH

O H
OH
OH OH
OH OH
OH
H3 CO
H H3 CO
H OCH3 HO
OH
HO
O
OCH3

OH OH OH OCH3
OH
OH

Page | 52
7.6.4 Lab plant demonstration with fractionated cresol range material

In light of the conclusion that the plug material was formed by the polymerization of aldehyde and
phenol components, various approaches to minimize the polymerization reaction were considered.
The approach that was considered most promising was to fractionate the first stage upgrading
product stream and recycle the material that boiled between ~180 and 210°C. This modified
process concept for pyoil upgrading requires fractionation and recycle of cresol boiling range
material from the first stage upgrading product. A lab plant was established to confirm that there
were no significant issues from the use of fractionated, compared to as received cresol. The
objective was to generate product consistent with that produced during the pilot plant operation.
That requires a lower conversion than previously targeted to maintain the cresol range material in
the product. Based on our prior experience, in practice this meant running such that a single liquid
phase product is produced with a density of between 1.02-1.05 g/mL.

Multiple feeds were tested including:

1. 3:1 Fresh (purchased) cresol/pyrolysis oil: this run essentially repeated the prior reference
experiments whereby pyrolysis oil/cresol was processed successfully for several hundred
hours in lab and pilot plants. In this run about 330 HOS were achieved with smooth operation.
2. 3:1 Fractionated cresol /pyrolysis oil: For operation with recycle of cresol only 2 fractionation
columns (or a fractionation column with a side draw) would be required. This was viewed as an
expensive option but with a high probability of success. With cresol recovered from the reaction
mixture by distillation (atmospheric pressure, bp~190-215oC) and conditions otherwise as in
the test with fresh cresol, 192 HOS (8 days) was completed without incident.
3. 3:1 Fractionated total overhead w/ cold separator product /Pyoil. Operation with a single
fractionation column with all material cresol and lighter taken overhead results in a product that
phase separates upon cooling to give a water and light oxygenate upper phase and a water
saturated cresol phase. The fractionated cresol with saturated water from the cold separator
was processed as above for 220+ HOS.

In all cases the product quality from the feeds was essentially the same.

7.7 Pilot Plant– Modified to allow recycle of cresol range material

The pilot plant was modified to allow for the recycle of cresol range material. This section
covers pilot plant Runs 2408 – 2412A that were done to demonstrate the first stage upgrading by
hydrotreating. In the first runs in this section, fractionation and recycle was done in the plant. For
the final runs, the fractionation was simulated by the addition of fresh cresol to the feed and the
products were fractionated off-line.

Page | 53
7.7.1 Pilot Plant Configuration

The plant ran in two major modes: once-through (OT) and recycle. OT mode consisted of running
the hydrogen-cresol-pyoil feed mixture through the reactor at elevated reactor temperature, and
collecting the full effluent in a product tank. A gas stream was vented from the product tank,
metered, and analyzed. OT mode also sometimes incorporated the distillation column, in which a
primarily cresol and water cut was lifted to the column overhead to separate it from the pyoil-
derived heavies fraction at the bottoms. A gas stream from the column overheads and low
pressure separator was metered and analyzed. Recycle mode was very similar to OT-mode with
the distillation column, except that the organic phase of the column overheads was recycled to the
reactor, and typically the cresol injection was dialed back by the same amount that overhead
recycle was dialed up.

7.7.2 Feedstock

The feedstock was filtered and ion-exchanged pyrolysis oil from a commercial supplier. In Runs:
2412 & 2412A the feed was spiked with sulfur from zero HOS. However, the sulfur injection only
occurred for a part of Run 2411 (475-528 HOS). In order to simulate product recycle, the pyoil
feedstock was blended with a mixture of cresol isomers boiling between ~180 and 205°C. At the
end of Run 2412A, rather than fresh cresol, cresol boiling range material from the off-line
Oldershaw fractionation was blended with the pyoil feedstock.

The hydrogen was injected once-through from the hydrogen header. There was no gas recycle
on the pilot plant.

7.7.3 Run Summaries

The main objectives of Runs 2410, 2411, 2412, and 2412A were to demonstrate 1000 hours of
stable operation, and produce products that could be used as second stage upgrader pilot plant
feed. In addition, during the later runs, a full size catalyst, as opposed to meshed catalyst particles,
was used.

7.8 Full Pyrolysis Oil Hydrodeoxygenation

During the course of the IBR project two process concepts for first stage hydroprocessing were
examined as described above,

1. Full product recycle and

2. Product fractionation with recycle of cresol-rich stream.

Page | 54
In both cases, the first stage product remains substantially oxygenated, and so a second stage of
hydroconversion is required to generate a hydrocarbon-rich final product suitable for use as
transportation fuel. During execution of the IBR project second stage upgrading was demonstrated
for both of the main process concepts. This work was completed in laboratory plants similar to
those described above for the first stage upgrading.

7.8.1 Process with Full Product Recycle

When product from the pilot plant runs in late 2011 was available, an experimental effort was
started to convert the effluent to full hydrocarbon product. For the initial work a commercially
available hydrotreating catalyst was used. The work was conducted in a laboratory plant. The run
objectives were:

1. Demonstrate that product from first stage upgrading with full recycle could be converted into
hydrocarbon suitable for fuel;
2. Demonstrate the material balance
3. Begin to address catalyst stability.

Process variables were selected based on a model study of cresol hydroconversion and included
temperature of 410oC, pressure = 800 psig, H2/HC ~ 4700 scfb (2700 for feed B), and LHSV =
0.5 hr-1.

Two feeds were employed during the run, collected from two different stages of pilot plant
operation (Table 7.2).

Table 7.2: Feed Characteristics

Oxygen Water Sulfur

Density (g/mL)
Feed (mass %) (mass %) (% by LECO)**
Feed A 9.5 3.8 1.19 0.9592
Feed B 4.6 0.72 0.71 0.8984
**Residue in product from stage 1

Feed B contained a significant amount of toluene and methylcyclohexane due to over-conversion

of cresol in first stage operation. We chose to evaluate this feed anyway to simulate 2nd stage
upgrading with low oxygen product recycle. Each feed ran for about 120 hours on stream and
included multiple weight checks.

Material balance of the plant was also evaluated. Overall material balances were fair to very good
but substantial inconsistency was observed, particularly in the liquid hydrocarbon recovery and in
the gas yield. During the post run evaluation it was determined that the wet test meter was leaking
which led to unreliable gas measurement.

Overall, 2nd stage upgrading with commercial hydrotreating catalyst does seem feasible. Catalyst
deactivation is a concern and improved material balance is needed. Note that since the feed was

Page | 55
still enriched in cresol (first stage product with limited batch recycle) some differences in 2nd stage
operation would be anticipated when true first stage product (full recycle) was run. Due to the
updated process concept for first stage operation, however, this was not fully explored.

7.8.2 Process with Fractionated Product

The ultimate process concept evaluated in the IBR program for pyoil upgrading requires
fractionation and recycle of cresol from first stage hydrodeoxygenation. The first stage upgrader is
operated at lower conversion than previously targeted to maintain the cresol content in the product.
Fractionation of the cresol, however results in a residual heavy product as well as a stream of
lighter than cresol oxygenates and water. This residual material is significantly different from the
feeds that were previously fully deoxygenated in 2nd stage processing. Therefore lab plant scoping
studies were conducted to verify that this feed material was still suitable for conversion to
hydrocarbon. A small lab plant was used due to its low feed requirement. The feed consisted of the
pilot plant product that had been fractionated in the Oldershaw distillation column at atmospheric
pressure and consisted of a mixture of the fractionated bottoms material (bp> 215oC), and light
oxygenates (BP< 195oC). The feed had a bulk composition of ~21% oxygen and 2-3% water.

Two runs were completed in the small lab plant (7 cc catalyst loading). The first run used a
commercial hydrotreating catalyst (the same as was used in the 2nd stage work reported above).
Product analysis, however, showed significant residual oxygen in the product (10-14%). Efforts to
improve deoxygenation by increasing temperature were only marginally successful indicating
either significant deactivation or catalyst bypass. The first objective, actually getting the feed
introduced to the reactor (a concern due to the tarry nature of the bottoms product) was achieved.
No issues with plant pressure were observed during the run.

Figure 7.5: GCMS of products from 2nd stage HDO of pilot plant product

oxygenates
hydrocarbons

Page | 56
GC-MS of the liquid product (Fig. 7.5) showed primarily oxygenates and hydrocarbons. The
relative amounts of each changed little over the course of the run. The fact that oxygenates are so
prevalent despite the fact that a substantial amount of water is produced suggested that perhaps
the heavy material is preferentially, or at least competitively, converted in the 2nd stage. Gel
Permeation Chromatography (GPC, Fig. 7.6) seemed to confirm this. The heavy tail present in the
feed (green line in Fig. 7.6) is substantially truncated in the reaction product, even under the
mildest conditions (blue line). A few caveats about interpreting the GPC data. Peak maximum
(or more specifically Average Molecular Weight) is likely reasonably accurate (note the cresol peak
maximum). The peak width (or edge) is not exactly correlated with the MW of heavies but it is
believed that the trend exhibited (lower MW in product than feed) is real.

Figure 7.6: GPC of 2nd stage feed, product and cresol reference

80000 GPC "Tail"

70000
P319R106T1
60000 Cresol
Feed
50000

40000
int

30000

20000

10000

0
1 10 100 1000
-10000
Molecular Weight

The gas product for all weight checks consisted of primarily methane, ethane (plus other light
hydrocarbon) and CO2.

Despite the poor catalyst activity the run was successful in two aspects: getting the feed
introduced to the reactor (a concern due to the tarry nature of the bottoms product) and no issues
with plant pressure.

Due to the higher residual oxygen level in the feed compared to what was demonstrated previously
for 2nd stage upgrading (21% vs. 10-14%) it was proposed that the alumina-based catalyst was
suffering from hydrothermal deactivation, so an experimental catalyst more stable in a high steam
environment was loaded for the second run. Over this catalyst and with the same feed, albeit at
higher reaction temperature, an additional increment of deoxygenation was achieved compared to
the earlier catalyst, but residual oxygen remained higher than desired at about 6%. Additional
process variable studies in the small reactor showed increase in hydrogen pressure caused a

Page | 57
much more significant improvement in the product than changes in other process variables. At 900
psig, the product was less viscous and lighter in color, and contained a higher concentration of
hydrocarbon components compared with 800 psig products. This was the pressure limit in the
small plant so additional second stage upgrading work was moved to a different plant with greater
temperature and pressure ranges, more extensive capability to achieve good material balance and
better data acquisition for continuous monitoring of plant operation. The main disadvantage is
a greater reactor volume requiring substantially more feed.

Therefore roughly 7 gallons of product from first stage pyoil processing were fractionated in the
Oldershaw distillation column into 3 cuts. The cuts mimic what would be done in the pilot plant,
where the middle boiling fraction is mostly cresols and is recycled into the fresh feed as a diluent,
and the heavy cut would be taken on to 2nd stage upgrading. A small sample of the fractionation
bottoms was heated in a glycol bath to determine its flow behavior at temperatures ranging up to
100oC. For this sample it was shown that heating the feed to about 70oC was sufficient to maintain
a pumpable liquid feed.

This feed was processed over 80 cc of sulfided base metal catalyst (experimental formation).
Throughout the run, pressure was held constant at 1300 psig, while temperature was varied
between 400 and 430oC to attempt to achieve full deoxygenation. Liquid feed was maintained at
LHSV = 0.25 hr-1. Hydrogen flow was ~713 sccm corresponding to H2/feed of 12,000 scft/bbl.
Three feedstocks were processed during this run over several weeks.

1. Oldershaw column bottoms plus light oxygenates (batch 1)

2. Oldershaw column bottoms plus light oxygenates (batch 2)
3. 1:1 mix of feed 2 plus product from second stage upgrading to simulate recycle operation

Excellent material balance was achieved over the course of the run.

The first product collected (~16 HOS) was clear and colorless, a stark difference from the black,
tarry feed (figure 7.7). The product was analyzed by GCMS and found to be high in hydrocarbon,
and containing very little oxygenates

Page | 58
 Figure 7.7: Feed and first product from 2nd stage upgrading

Product characterization

Gasoline

The products from pyrolysis oil upgrading to hydrocarbon fuels must eventually go into the
transportation fuel pool. The highly aromatic nature of the product obtained is most consistent with
petroleum gasoline. Based on characterization of the products produced under the IBR project we
expect the hydrocarbon product to directly (without blending) meet many of the ASTM gasoline
specifications (Table 7.3). Potential areas of concern include the benzene content, which is likely
to be greater than specification, particularly if high aromatic/naphthene ratio is maintained as
required for high octane. In addition, insufficient product was generated to determine corrosion,
gum formation of oxidative stability. It is expected that the pyrolysis oil derived hydrocarbon
material will be blended into the overall gasoline pool.

Page | 59
 Table 7.3 Gasoline from hydrodeoxygenated pyrolysis oil

Gasoline Specification Deox Pyrolysis oil

ASTM 4818 AA Pass Comment
Vapor Press. kPa (@38°C) 54 Y Expect to Pass, subject to distillation
optimization
Distillation°C
10 % vol max 70 Y Expect to Pass, subject to distillation
optimization
50 % vol min 77 Y Expect to Pass, subject to distillation
optimization
50 % vol max 121 Y Expect to Pass, subject to distillation
optimization
90 % vol max 185 Y Expect to Pass, subject to distillation
optimization
FBP (98%) 225 Y Expect to Pass, subject to distillation
optimization
Benzene vol % (fleet ave) 0.62 ? <0.5 % benzene in full boiling range product
Sulfur ppm max 80 tbd Expect to pass based on
composition/processing options
Oxygen max 2% Y Consistently achieve <1%, current target
is <0.03
Lead g/L max 0.013 Y Below limit of detection by ICP
Phosphorous g/L max 0.001 Y Below limit of detection by ICP
3
Corrosion Copper max 1 tbd
Silver 1 tbd
Gum mg/100 max 5 tbd
mL
Ox. Stability minutes min 240 tbd
Octane RON 96 Lab Plant product, based on comprehensive
GC
MON 90 Lab Plant product, based on comprehensive
GC
(RON+MON)/2 93 Lab Plant product, based on comprehensive
GC

Page | 60
Distillate

The high aromatic and cycloparaffin content of the product ensures that the pyrolysis oil derived
hydrocarbon would be a poor quality diesel fuel (cetane number probably <30). This might be
acceptable as a low value blend component, akin to petroleum light cycle oil.

An alternate opportunity for a portion of the distillate range product, could be the aviation fuel pool.
While the composition of the jet range product fails to meet the ASTM specification for density and
aromatic content (both too high, Table 7.4), some aromatics are required in jet fuel to be consistent
with current airplane engine technology. Paraffinic aviation fuel, prepared from Fischer Tropsch
liquid, or from the hydrodeoxygenation of triglyceride oil, fail to meet the density and aromatic
specification as they are too low. So the pyrolysis derived hydrocarbon material would provide a
fully renewable blendstock that would facilitate introduction of these renewable paraffinic
components into the aviation fuel pool.

Table 7.4: Aviation Fuel component from hydrodeoxygenated pyrolysis oil:

Analyzed lab plant product (fractionated, 140-280°C)

Test ASTM Result Units Comment

Spec
ACID NUMBER 0.1 <1.0 mg KOH
per g
SULFUR XRF 3000 2 MASS-PPM
C 86.3 MASS-PCT
H 10.7 MASS-PCT
N D4629 279 MASS-PPM
OXYGEN 2.8 MASS-PCT O removal to very low levels is
must have
Density D4052 0.075- 0.8902 g/mL
0.84
FLASH PT D7094 38 52 °C
(min)
Freeze point -40 -46.7 °C
D7153 (max)
VISC D445 40°C 8 1.346 CSt
Water KF 448 MASS-PPM
Halide IC <1 MASS-PPM
Aromatics (max) 25 >50% Vol. %

Page | 61
7.9 Conclusions/Recommendations

Extensive experimental work was conducted on first and second stage upgrading of pyoil using
conventional fixed bed hydroprocessing technology. While it was possible to demonstrate 1000
hours on stream of first stage upgrading, it was necessary to skim the top of the catalyst bed off
during the run to manage pressure drop build up. Recurring problems with pressure drop build up
in the first stage upgrading experiments made progress very difficult and remains an unresolved
issue that impedes the development of this technology. Characterization of the plug material that
was removed from reactors showed it to be a co-polymer from the reaction of aldehydes with
phenols. Numerous attempts were made to promote the hydroprocessing reaction while demoting
the polymerization reactions. While some progress was made, it was not enough to get to a stable
operating point that would be viable for a commercial or demonstration scale unit. For this reason,
work on first stage upgrading was not taken beyond the pilot plant scale (~1-3 kg/hr feedrate). It is
recommended that future work be conducted at a research scale until a viable process concept
can be developed.

Page | 62
8 FCC Co-processing
The fluid catalytic cracking (FCC) process converts straight-run atmospheric gas oils, vacuum
gas oils (VGO), certain atmospheric residues and heavy stocks recovered from other refinery
operations into high-octane gasoline, light fuel oils and olefin-rich light gases. The cracking
reactions take place in a vertical riser upon contact of the liquid feed with hot fluidized catalyst.
Upon contact, the feed is vaporized and reactions occur throughout the riser length with residence
time on the order of seconds. In addition to the desired reactions, coke is formed and deposits on
the catalyst. The coked catalyst and vapor phase products are separated and the products are
condensed and fractionated while the catalyst is sent to the regenerator. The coke on catalyst is
burned off in the regenerator and the regenerated catalyst is circulated back to the riser
continuously.

During co-processing, a secondary feed stream consisting of cellulosic-derived pyrolysis oil is

processed in the FCC unit along with the traditional VGO feed. The main function of co-processing
is to deoxygenate the pyrolysis oil, much of which boils in the gasoline and diesel ranges. In
addition to gasoline and diesel range material, pyrolysis oil deoxygenation also produces carbon
monoxide, carbon dioxide and water as by-products.

The pathway to commercialization for this new application to FCC technology starts with
completion of pilot plant trials with commercial feed and catalyst to simulate the performance with
co-processing in a commercial unit. Provided that the performance measured during the extended
trials is favorable, the long term solution is full commercial refinery adoption of the co-processing
solution.

A proof-of-concept trial was completed in 2013 to determine the feasibility of completing extended
FCC co-processing trials and to observe and identify any concerns for completing the short term
trials. In addition, a circulating riser pilot plant trial at UOP’s Des Plaines location was completed in
2015 to compare yields at refinery conditions with and without 3.0 wt-% pyrolysis oil addition.

8.1 Proof-of-Concept Commercial Trial

A commercial refinery, in cooperation with UOP and Ensyn, conducted a test where up to 1
vol% of pyrolysis oil was fed along with VGO to their FCC reactor. The primary objectives of
this test were to:
1. Demonstrate the feasibility of continuously injecting pyrolysis oil into an FCC reactor
riser using UOP’s engineered distributor.
2. Evaluate the impacts to the reactor and regenerator operations caused by processing
pyrolysis oil.
3. Analyze the product streams for unusual contaminants caused by the pyrolysis oil,
which could hurt the FCC product quality and/or value.

Page | 63
It was decided that a detailed material balance would not be attempted due to the short term
nature of the test and inaccurate flow measurements on some of the product flows.

8.1.1 Proof-of-Concept Commercial Trial Feed

The commercial refinery processes a straight run VGO feedstock that contains low metals and
moderately low concarbon. These characteristics along with good hydrogen content make it a
good feed for the FCC.

A skid mounted pyrolysis oil feed system was designed and provided by Ensyn. The pyrolysis oil
feed was delivered to site by a 5000 gallon tanker truck. A flexible hose was fitted to the outlet of
the tanker truck and fed to the Ensyn skid.

Pyrolysis oil was injected at a rate of 1 vol-% based on the VGO as shown in Figure 8.1.

Figure 8.1: Pyrolysis oil injection rates for the duration of the trial
___ Pyrolysis Oil Rate (gpm)
___ Pyrolysis Oil Concentration (vol-percent of feed)

8.1.2 Proof-of-Concept Commercial Trial Results

[Link] Feasibility of Continuous Pyrolysis Oil Injection to the FCC Unit

UOP’s pyrolysis oil feed distributor, developed to minimize polymerization and to allow for
uninterrupted FCC operation, worked flawlessly throughout the duration of the trial. The trial was
terminated after 16 hours due to completion of the agreed upon duration; there were no signs of
plugging at any point during the trial. Pyrolysis oil injection was limited to 1 vol-% for this trial to
minimize the potential disruption to the commercial unit operations. Extended demonstration trials
will aim to inject higher amounts of pyrolysis oil volumes in the total feed to the unit.

Page | 64
[Link] Evaluation of Operational Impacts to Reactor and Regenerator
The FCC heat balance variable, delta coke, is directly related to feed quality of the FCC. FCC units
that battle low regenerator temperatures often recycle slurry to the riser to increase delta coke and
regenerator combustion temperatures (to maintain complete combustion of CO to CO2). As
pyrolysis oil generates equal or more coke per vol-% as compared to most slurry oil (based on pilot
plant testing) heat deficient units would acquire additional benefit from co-processing pyrolysis oil.

[Link] Product Stream Analysis

Samples were taken prior to the start of, during the middle of and just before the end of the
proof-of-concept commercial trial to investigate the use of pyrolysis oil without degradation to the
catalyst, the FCC unit products and without negative impacts to downstream processes.

The straight run VGO processed at the refinery has low contaminant metals (Ni, V, and Fe).
In contrast, the pyrolysis oil injected into the FCC unit has higher levels of known catalyst
contaminant metals including calcium, potassium, magnesium, sodium, and iron. These metals are
known to deposit on the catalyst surface resulting in a shift in FCC catalyst activity and product
selectivities. Due to the short duration of this test, the buildup of contaminant metals was minimal
and thus had negligible impact on product yields. However, these elevated metals levels may have
an impact on the catalyst with long term continuous processing of pyrolysis oil in the FCC riser.
Catalyst make up rates may need to be adjusted and the catalyst formulation may need to
be altered to mitigate the metals impact on the FCC process performance.

The pyrolysis oil should be treated prior to injection in the FCC unit to reduce total metals content
below 50 wt-ppm. The alkali and alkaline earth metals, which were found to have completely
deposited on the catalyst, should not impact FCC products or downstream processes as none of
these metals were found in the product samples. The acidity of the pyrolysis oil did not appear to
impact surface area of the catalyst and coked catalyst sample analysis is ongoing to determine
whether there is a difference in the coke morphology and composition from the samples before
and during pyrolysis oil injection.

Product analyses did not show increased metals content as all metals appear to have deposited on
the catalyst. Overall, the product stream hydrocarbon characteristics remained largely unchanged.

As no metals were found in the FCC products; there should be no impact to downstream
processes due to the higher metals content of the pyrolysis oil feed in comparison to VGO.
Increased oxygenates found in the main column overhead receiver sour water stream are
essentially the same oxygenates typically found in this stream, but in higher concentrations.
This increase is manageable provided that waste water treatment is notified and can prepare
for the increased levels.

Page | 65
8.2 UOP Circulating Riser Pilot Plant Trial

Starting in late 2014, a FCC co-processing trial was completed in the circulating riser pilot plant
unit at UOP’s Des Plaines, IL location. FCC unit feed and catalyst were obtained from a refinery
and pyrolysis oil was provided by Ensyn. The purpose of this study was to determine and compare
the yield slates of vacuum gas oil (VGO) baseline tests to those with 3.0 wt-% addition of
pyrolysis oil.

8.2.1 Pilot Plant

UOP’s circulating riser pilot plant supports FCC unit research and development efforts. The final
data set consisted of 26 tests, including replicates, for 8 separate operating conditions. The test
duration was 1 hour and does not include the time necessary to line out the plant prior to starting
a test.

8.2.2 Circulating Riser Pilot Plant Trial Feed

VGO and equilibrium catalyst (E-Cat) samples were obtained from a commercial refinery for use
in the testing. Properties of the feed are shown in Table 8.1. Pyrolysis oil was provided by Ensyn.
Metals concentration in the pyrolysis oil was reduced to approximately 50 wt-ppm total metals
(or less) prior to the trial.

Table 8.1: Pilot Plant Trial VGO Feed Properties

METHOD Component Value Units

BASIC NITROGEN LIQ U269 Basic Nitrogen 0.047 Mass-Pct

MICRO CARBON RESIDUE 5 GM R206 D4530 Carbon Residue 0.33 Mass-Pct

DENSITY RELATIVE 2ML R206 D4052 Degree API 24.77 Degree API

DENSITY RELATIVE 2ML R206 D4052 Density 4052 (15°C) 0.9047 g/ml

GC D2887 DISTILL 538C BP at 50.0 Mass-Pct 445.4 Degree C

HEPTANE INSOLUBLE 30GM R206 U614 Heptane Insoluble 0.014 Mass-Pct

METALS TRACE IN OILS 50G R215 U389 Total 1.4 Mass-PPM

SULFUR HYDROCARB XRF 10G R235B Sulfur XRF 3609 Mass-PPM

IN PPM 5ML R210 D4629 CHEMOLUMINESCENCE Nitrogen (D4629) 1368 Mass-PPM

H2 BY NMR 8 DRAM VIAL 21E CPT-H-1 Hydrogen 12.65 Mass-Pct

Page | 66
8.2.3 Circulating Riser Pilot Plant Trial Results

One goal of the trial was to determine a range of catalyst-to-oil ratios that would bracket targeted
conversion, yields, and operating conditions for VGO baseline tests (those without pyrolysis oil
injection) that compare reasonably to the commercial unit performance. As the trial progressed and
tests were completed, the range of targeted catalyst-to-oil ratios was finalized. The response of
conversion to changes in catalyst-to-oil ratio was as expected with increasing conversion realized
for increasing catalyst-to-oil ratio at the targeted operating conditions. The data obtained brackets
the targeted commercial conversion.

Pyrolysis oil conversion to gasoline and light cycle oil (LCO) range material was observed,
with VGO baseline and pyoil test yields for gasoline are shown in Figure 8.2. Dry gas, liquefied
petroleum gas (LPG) and hydrogen yields were reduced. The decrease in hydrogen yield occurs
as hydrogenation reactions take place with the hydrogen deficient components present in the
pyrolysis oil.

An increase was observed for coke yield in the pyrolysis oil experiments. Additionally, carbon
monoxide, carbon dioxide and water are by-products of FCC co-processing. Carbon dioxide and
carbon monoxide, present in the product gas, increase through removal of carboxyl or carbonyl
groups of the compounds present in pyrolysis oil through decarboxylation and decarbonylation
reactions, respectively. Water is present both from the pyrolysis oil feed and also from
deoxygenation of the pyrolysis oil. The observed yields were consistent with previous FCC
co-processing studies completed by UOP.

Figure 8.2: Gasoline yields from circulating riser pilot plant trial

Circulating Riser Pilot Plant Trial, 3.0 wt-% Pyrolysis Oil

53
VGO
VGO + 3 wt-% Pyrolysis Oil
51
Gasoline, wt-%

49

47

45

43
57 59 61 63 65 67 69 71 73 75 77
Conversion, wt-%

Page | 67
A total of 26 tests were completed with replicates completed at each targeted catalyst-to-oil ratio.
The completion of replicates increased confidence in the yield trends observed. At the end of the
study, additional tests were completed to determine the volume of additional pyrolysis oil that could
be accommodated in the UOP circulating pilot plant. Increased pyrolysis oil volumes would help to
minimize sample error and improve confidence in the measured yield effects.

8.3 FCC Co-processing Conclusions

The proof-of-concept commercial trial demonstrated the feasibility of continuously injecting

pyrolysis oil into a FCC reactor riser using UOP’s engineered distributor. The injector worked
flawlessly for the duration of the trial. The product hydrocarbon character remained largely
unchanged and no elements were observed that could prevent completing a longer term, four
week commercial trial. Reactor pressure, regenerator temperature and delta coke responses
were noticed upon introduction of pyrolysis oil injection to the system.

The trial completed in UOP’s circulating riser pilot plant allowed for comparison of yields
obtained from VGO only injection with 3.0 wt-% pyrolysis oil injection. A total of 26 tests were
completed to span eight test conditions (four catalyst-to-oil ratio targets both with and without
pyrolysis oil). Replicate experiments increased confidence in the observed trends which were
consistent with previous pyrolysis oil co-processing experiments completed by UOP.

Once deoxygenated, the pyrolysis oil converted to gasoline and diesel range material.
Reductions were seen in LPG, dry gas and hydrogen yields. Carbon dioxide, carbon monoxide
and water were added to the product slate as a result of pyrolysis oil deoxygenation and
pyrolysis oil feed composition. The measured coke yield increased at constant conversion as
expected, and would benefit heat deficient FCC units.

Pilot plant trials in UOP’s circulating riser are the first step in the pathway towards
commercialization. Completing trials similar to this first pilot plant trial with customer feed and
will likely be necessary to secure extended commercial trials.

Page | 68
9. Commercialization and Regulatory pathway
Commercializing the technology for converting pyrolysis oil to transportation fuels is dependent on
which upgrading pathway is chosen. While the commercialization pathways for converting pyoil to
transport fuels by either upgrading or co-processing in an FCC unit are similar, they are not
identical. This discussion will concentrate on the commercialization of the co-processing
technology, but we will consider aspects that are related to upgrading.

Pyrolysis oil upgrading to transportation fuels requires hydroprocessing infrastructure similar to a

commercial petroleum refinery. The operating conditions required to hydroprocess pyrolysis oil are
similar to what is used for heavy petroleum fractions. The commercialization plan for upgrading
assumes some level of integration within an existing petroleum refinery to improve the overall
energy efficiency and production outlook for the unit. This plan would also allow the pyrolysis oil
upgrading facility to capitalize on the utilization of all refinery operations staff. This would bring a
level of processing experience to the operating facility that would not be possible otherwise and
would likely result in successful operation of the unit.

Infrastructure currently exists at petroleum refineries to facilitate the incorporation of pyrolysis oil
directly into in existing product distribution system for both gasoline and diesel fuel. This integration
with downstream processing equipment would allow the refinery to deliver high quality products to
the marketplace. Using existing refinery infrastructure or reusing retired assets would maximize the
total investment return of an upgrading project while minimizing total project capital expenditures.

UOP has been in direct contact with petroleum refiners around the world to gauge their interest
in renewable transportation fuels produced from pyrolysis oil. While most refiners have been
interested in this type of technology, their level and type of interest has been split geographically.

Petroleum refiners in the United States desire a partially renewable fuel solution to meet the
Renewable Fuel Standard (RFS), but want to primarily continue to produce petroleum fuels using
their own infrastructure. Current mandate levels influence how refiners go about meeting biofuel
requirements and refiners today prefer a renewable fuel solution that can be met using existing
infrastructure with very little capital expenditure.

The United States refining industry is split between pure refining companies and fully integrated oil
companies. The pure, non-integrated refining companies do not own upstream oil assets and are
often willing to explore other fuel processing opportunities. Integrated oil companies prefer to
process their upstream oil assets and biofuel production can be seen as outside their traditional
scope. The refining industry’s desire for a partially renewable fuel solution will likely result in the
adaptation of co-processing of pyrolysis oil strategy in the future

Refiners within Europe have expressed a preference for 100% neat renewable fuel solutions.
They value 100% renewable fuels higher than blended partially renewable fuels, as their
marketplace places a higher value on greenhouse gas reduction and the sustainability of

Page | 69
transportation fuels. While certain member states of the European Union do have biofuel
mandates, most are not binding and do not contain penalty clauses. Therefore, their preference for
renewable fuels is more from a desire for sustainability than from transportation fuel mandates.

The geographic discrepancy between northern European countries and southern European
countries contribute to their preference for renewable fuel technology. Many northern European
countries utilize a lot of renewable energy currently for power generation and therefore do not have
as many petroleum refineries, while southern countries without access to hydropower or other
cheap renewable energy options tend to have a more developed refining industry. Northern
countries are more willing to build greenfield renewable fuel projects that produce 100% neat
renewable fuels given their lack of existing refinery assets. It is in these countries that full
upgrading of pyrolysis oil would likely be the preferred renewable fuel solution.

The Asian fuel market is a combination of desire for either a neat renewable fuel or a partially
renewable transportation fuel. Their preference often comes from whether or not they have a fully
developed petroleum refining industry already in place. Interest from project developers in
locations around Southeast Asia are to produce 100% neat renewable fuels, while petroleum
refiners and project developers in mainland Asia have expressed more interest in co-processing
pyrolysis oil as the preferred solution.

9.1 Implementation

Refiners view the utilization of their existing FCC equipment as a logical solution for the production
of renewable transportation fuels. Co-processing pyrolysis oil is an integrated solution that offers
refiners an opportunity to produce partially renewable fuels for meeting the RFS program mandate.
Besides utilizing existing refinery infrastructure, co-processing is a low capital cost solution that
allows a refiner to meet its RFS cellulosic mandate.

Co-processing of pyrolysis oil is a simple solution that fits today’s refinery outlook in many ways.
Renewable fuel mandates are often the main driver behind refiners’ investigation into renewable
fuel solutions. Refiners would prefer a low-cost low capital expenditure solution and co-processing
fits this desire by utilizing existing refinery infrastructure. With only the need for pyrolysis oil
storage tanks and feed line piping to the FCC unit, the required cost for site improvements can
be kept low.

Another driver for interest in the co-processing solution is the increased use of shale oil (tight oil)
by refiners within the United States. Shale tight oils generally result in lower delta coke in an FCC
unit and therefore can cause problems managing the regenerator temperature required to remain
in heat balance. Co-processing of pyrolysis oil can be a solution for this problem. When pyrolysis
oil is injected into an FCC unit, an increase in delta coke is seen resulting in an increase in
regenerator temperature. With the increased penetration of shale oil usage across the United
States, co-processing can be seen more as a solution to an operational problem rather than just as
part of a mandate for renewable fuels.

Page | 70
Co-processing by a petroleum refinery can also help them meet the marketplace desire for cleaner
burning fuels. As their customer base has an increasing need for cleaner fuels to meet emission
requirements, being able to offer a partially renewable fuel with low sulfur becomes much more
desirable. In addition, there is an increased focus by customers on the sustainability of all fuels.
A refiner that can demonstrate that they are incorporating second generation renewable fuels into
their supply can be viewed more favorably by the marketplace.

Figure 9.1 -Summary of Advantages of Co-processing Pyoil

• Co-processing RFO in an FCCU produces renewable

gasoline & diesel blendstocks
Compliance solution
controlled by the refiner • Generate biofuel credits directly at the refinery

• Reduce cost of compliance

• Produce pipeline-compatible biofuels with existing

FCC equipment
Utilize existing refinery
infrastructure • Eliminate dedicated blending infrastructure

• No blendwall limitations

• Minor capital costs for storage & feed injection

equipment
Easy to implement
• UOP provides injection equipment & operations
assistance

The pathway for commercialization of co-processing is likely to be a decentralized energy plan

where the RTP unit is situated within a biomass basket, not likely next to a petroleum refinery.
This allows for one of the primary advantages of the RTP unit to densify and liquefy the biomass to
minimize the overall cost of transportation. Placing the RTP unit close to the biomass source limits
the transportation costs to the RTP unit but maximizes the transportation savings from the RTP
unit to the refinery.

Page | 71
 Figure 9.2 - Co-processing of pyrolysis oil pathway

Within the United States there are 95 operating petroleum refineries with approximately 5.6 million
barrels per stream day (BPSD) of FCC unit capacity. Assuming a 3% blend of pyrolysis oil for co-
processing, there are potentially 2.47 billion cellulosic gallons (ethanol equivalent) of production.
Many of these refineries are located nearby to multiple sources to biomass materials for potential
conversion into pyrolysis oil. This amount of cellulosic fuel will advance efforts by the EPA to meet
the primary goals of the RFS program.

Page | 72
9.2 Regulatory Pathway – Part 79 and Part 80

In addition to the business aspects of commercialization, there are the regulatory steps that need
to be followed as well. All transportation fuel facilities in the United States must register their fuels
with the Environmental Protection Agency and the Internal Revenue Service. This is true for both
traditional petroleum fuels and renewable fuels. For transportation fuels made from upgraded
pyrolysis oil or co-processed with pyrolysis oil, the pathways to registering the fuels would be
similar but not identical.

There are two primary approvals required to register the co-processed fuel for sale. The
first approval is primarily concerned with the air emissions from the combustion of the fuel in
gasoline and diesel vehicles. This approval is known as “Part 79”. The EPA requirements for the
registration of designated fuels and fuel additives (F/FA) are stipulated by sections 211(b) and
211(e) of the Clean Air Act (CAA). Part 79, Title 40 Code of Federal Regulations (CFR) addresses
requirements for the “Registration of Fuels and Fuel additives”. Under the Tier 1 requirements of
the CFR, manufacturers of transportation fuels (F/FAs) are required to supply EPA with:

• The identity and concentration of individual emission products from the F/FA
• An analysis of potential emissions exposures
• Any available information regarding the health and welfare effects of the whole and
speciated emissions.

As a result, emission generation, collection, and analysis of the regulated emissions and of
selected volatile- and particulate-phase unregulated emissions, and a speciation of volatile phase
hydrocarbon compounds are performed.

Part 79 requires standardized tests be conducted to analyze the emissions generated during the
combustion of the fuel in vehicles and to determine whether those emissions meet EPA guidelines.
The approval process is primarily concerned with the Health Effects of fuel combustion. Part 79
testing involves a comparison of the emissions profile of an EPA-approved “base” transportation
fuel against that of a “candidate” transportation fuel.

The full fuel application consists of three parts. Tier 1 covers emissions’ testing that determines
whether the combustion emissions meet current fuel standards, plus a full analysis of the
emissions speciation and a literature review of the species contained within the emissions.
Tier 2 is concerned with toxicological testing that may be required if the emissions are substantially
different than other previously tested fuels. For co-processed fuels, this is not likely to be
applicable. Tier 3 covers other questions identified by the EPA during the review of
the application.

The entire testing process can take anywhere from 6 to 18 months depending on how quickly the
fuel can be produced and delivered to the testing facility and the entire analysis completed. The
final application can be a very large document and is therefore prepared by experienced
consultants within this transportation fuel area.

Page | 73
Co-processing pyrolysis oil will produce fuel blendstocks for both the gasoline and diesel pools.
Since both types of fuel will be entering the transportation fuel market, two completely independent
Part 79 applications must be submitted for the co-processed gasoline and co-processed diesel
fuel. In July 2015, IBR project partner Ensyn Corporation received Part 79 approval for co-
processed diesel, known as RFDiesel, from the EPA. Similar approval was received for the co-
processed gasoline, known as RFGasoline, in November 2015.

The second primary registration for renewable fuel producers is called Part 80. This registration
is specifically related to registering the fuel producing facility under the EPA Renewable Fuel
Standard. This registration is only for facilities that produce renewable fuels. The particular
requirements can be found under the Clean Air Act, 40 CFR Subpart K- Renewable Fuel Standard.

This process properly identifies and qualifies individual fuel producing facilities within the RFS2
renewable fuel program. It is a qualification program to verify that the particular fuel facility is
capable of producing the type of renewable fuel pathway applied for under the RFS2 program.
For example, a corn ethanol facility that is applying for Part 80 approval must verify that is capable
of producing that type of biofuel from a specific set of feedstocks.

To verify the facility is capable of producing renewable fuels, an independent engineer is hired to
visit the facility and visually inspect the equipment and production process. This engineer must
have prior fuel processing experience or specific experience within that type of fuel production.
Once the engineer’s inspection visit is complete, a formal report is completed and sent to the EPA.

For co-processed fuels produced from pyrolysis oil, two separate facilities will need to be
registered under Part 80. First, the RTP plant that produces pyrolysis oil will need to be inspected
and verified it is using the verifiable cellulosic feedstocks. Second, the petroleum refinery that co-
processes the pyrolysis oil will need to be observed and verified. The petroleum refinery’s FCC unit
and the associated delivery equipment for the pyrolysis oil would be viewed by an independent
engineer to confirm proper operating application. The independent engineer would also verify the
downstream processing pathway at the refinery to ensure the renewable transportation fuels are
being produced.

This review process can take anywhere from 6 to 9 months depending on the speed of the
independent engineer’s review and the final report. Once this written report is delivered to the EPA,
the EPA has 90 days for a formal review and approval. Currently under this program, there are
over 700 renewable fuel facilities and importers registered with the EPA to deliver transportation
fuels within the United States.

Table 9.1 contains a summary of application steps for renewable fuel certification under both
Part 79 and Part 80.

Page | 74
 Table 9.1 - Registration Summary Table

(Adapted from EPA website at [Link]

Step Registration Requirements Comment

See Table 1 of Title §80.1426 to

Determine whether your fuel may qualify
determine whether your fuel falls
1 under an existing Renewable Fuel pathway
under an approved RFS fuel
under RFS.
pathway.

Register under 40 CFR Part 79, Fuels and

Importers, not foreign producers,
2 Fuels Additives Registration (FFARs) if the
need to register under FFARs
renewable fuel is meant for on-road use

After registering under Part 79, create a New

[Link]/otaq/fuels/reporting
3 Company Request to register under 40 CFR
/[Link]
Part 80.

Obtain an engineering review conducted by

4 a third-party independent professional Review required by Part 80.
engineer per §80.1450(b)(2).

Provide permits or production records to

5
support the facility’s production capacity

Include the type of process heat

Provide a process heat fuel supply plan
6 fuel supplied, the name and
pursuant to §80.1450(b)(1)(iv).
address of the supplier

Companies registering to use Biogenic

Include the type of each food
Waste Oils/Fats/Greases or Non-cellulosic
waste supplied, the name of each
Portions of Separated Food Waste
7 supplier, the address of each
feedstocks must provide a separated food
supplier, how the waste is being
waste plan pursuant
collected
to §80.1450(b)(1)(vii)(B)

All applicants should review the entirety of

§80.1450 to determine if the fuel pathways
8
in their registration warrant any additional
documentation requirements

Foreign producers who opt to generate their Please review §80.1466 and
own RINs upon importation of fuel to the contact the EPA Fuels Programs
9
United States, must submit additional Support line for further
documentation instructions

Summarized at [Link]

Page | 75
10. Feedstock Characterization and Life Cycle Analysis

10.1 Feedstock Characterization

Any integrated biorefinery must utilize a wide range of 2nd Generation Feedstocks, including
forestry residues and major agricultural wastes as well as high potential energy crops. Flexibility is
important for satisfying the need to produce fuel in locations with varying biomass types. For this
project, many different types of biomass feedstocks were evaluated for their potential for producing
pyrolysis oil. Eighty five production tests were completed on twenty four different types of biomass
feedstocks. These feedstocks were evaluated for their ability to maximize liquid pyrolysis oil yield
and overall energy yield, while being of suitable quality for upgrading to transportation fuels.

All feedstock testing was completed at Ensyn’s Renfrew facility utilizing their Bench Scale Unit
(BSU). The BSU is a pilot scale plant using the Rapid Thermal Processing (RTPTM) technology to
convert biomass into pyrolysis oil.

The feedstock characterization tests attempted to analyze a full slate of different types of biomass,
including potential feedstocks that were not currently anticipated to be utilized for pyrolysis oil
production. The bulk of the feedstocks tested were woody biomass, residuals or wood blends.
This was due to the likelihood that the majority of future pyrolysis oil facilities would be centered in
wood baskets because woody biomass has an established business infrastructure surrounding the
collection and use of residual materials. Of those woody materials, multiple combinations of clean
wood and bark were tested to optimize the best combination of yield and supply economics. The
different kinds of woody biomass included Albizia, Eucalyptus, Finnish softwood, Noble Fir, Oak,
Pine and Spruce. The agricultural biomass materials tested included corn stover, guinea grass,
palm kernel shells, switchgrass and wheat straw. As a potential future feedstock, algae was
tested.

Table 10.1 below shows the various characteristics of some of the feedstocks tested. Important
characteristics highlighted include:

• For “dirty” corn stover, the extremely high ash content. This material contained a lot of dirt,
rocks and other inorganic material that did not get converted into pyrolysis oil upon
conversion. This verified the need to ensure that agricultural materials from the field are
as clean as possible.

• For the wild algae, the same high ash content as corn stover will likely cause problems with
maximizing liquid yield. In addition, the high nitrogen content would likely be of concern for
upgrading of pyoil made from algae.

Page | 76
 Table 10.1 – Feedstock Properties Tested at Bench Scale

Bark, Sugar "Dirty"

Corn Stover Guinea "Wild"
Feedstock Properties shavings, Cane Corn Switchgrass
(Washed/ Grass Algae
sawdust Bagasse Stover De-ashed)

Moisture Content** Wt-% 6.0 5.3 – 6.5 5.4 – 6.2 4.0 5.0 – 6.0 4.9 – 6.4 5.7 – 7.0

Ash Content Wt-%, dry 2.12 2.81 27.98 0.56 9.45 26.72 3.58

Bulk Density @ 20°C, kg/dm3 0.2240 0.1249 0.5061 0.1461 0.1479 0.7806 0.1655

Carbon Content Wt-%,, dry 50.09 48.20 36.58 47.9 43.58 41.36 47.63

Hydrogen Content Wt-%,dry 5.87 5.82 4.46 5.7 5.48 5.58 5.79

Nitrogen Content Wt-%, dry 0.27 0.30 1.06 0.45 0.72 6.11 0.45

Sulfur Content Wt-%, dry <0.05 0.09 0.12 0.05 0.09 0.80 0.06

Oxygen Content Wt-%,, dry 41.66 42.78 29.81 45.31 40.68 19.44 42.5

HHV MJ/Kg, MAF 20.38 19.39 20.32 18.65 18.71 25.82 19.53

Cellulose Wt-%, MAF 31.51 43.82 39.05 46.27 39.47 11.08 34.60

Hemicellulose Wt-%, MAF 18.58 24.82 24.61 31.26 24.80 10.70 33.69

42.67 (non
Lignin Wt-%, MAF 37.20 26.97 24.61 19.05 18.78 22.03
polysacc.)

Acetone Extractives Wt-%, MAF 3.98 1.53 2.28 0.87 4.07 7.12 1.77

Water Extractives Wt-%, MAF 8.73 2.86 9.54 2.56 12.85 28.43 7.93

Page | 77
Table 10.2 below shows the various metals content of some of the feedstocks tested. This is
important because any metals that are in the biomass feedstock end up in the pyrolysis oil. Then,
for upgrading to transportation fuel, these metals contaminate the catalyst in either an upgrading
technology or in a co-processing application in a refinery FCC unit. Therefore, it is vitally important
that the metals are minimized either in the feedstock or the pyoil after conversion.

The four metals that are of primary concern are magnesium, calcium, potassium and sodium. These
large quantities, especially in algae, need to be addressed through post-conversion filtration. For this
project, a Metals Finishing Unit was incorporated to lower the total metals count to a number
acceptable for fuel upgrading.

Page | 78
 Table 10.2 – Feedstock Metals Content (average of 3 runs in most cases)

Metal (ppm) Wood Scrap Bagasse Dirty Stover Washed Stover Guinea Grass Algae Switchgrass

Aluminum (Al) 441.4 1779 6407 5157 55.2 10453 78

Barium (Ba) 85.7 8.96 148 98.7 5.42 212.5 8.8

Boron (B) 7.5 2.33 17.36 ND 9.93 22.5 2.74

Calcium (Ca) 5712 593 5578 4715 5982 10195 2322

Chromium (Cr) 0.65 6.12 29.6 23.6 ND 14.4 0.1

Cobalt (Co) 0.2 0.93 ND 0.74 0.4 3.3 ND

Copper (Cu) 1.8 6.46 6.27 7.66 3.32 44.7 3

Iron (Fe) 337.5 2755 1673 1400 73.9 8434 81.6

Lead (Pb) ND 0.35 3.1 2.5 ND 11.6 ND

Magnesium (Mg) 389 284 1302 1194 2078 2445 1909

Manganese (Mn) 114 50 79.3 66.3 48.4 1678 43

Nickel (Ni) 0.4 1.44 ND 2.5 ND 12.6 0.4

Potassium (K) 1454 2016 13168 7073 8939 8850 3439

Silver (Ag) ND ND ND ND ND 1.18 ND

Sodium (Na) 54.5 154 2369 1621 1430 3626 543

Strontium (Sr) 22.3 5.6 58.2 44.2 26.3 95.3 8.7

Tin (Sn) ND ND ND ND ND 3 ND

Titanium (Ti) 20.8 328 130.6 97.3 12.7 861 3.9

Vanadium (V) 1.1 10.2 ND ND ND 21.3 ND

Zinc (Zn) 77.9 14.6 68.2 72.2 15.9 164 15.33

Page | 79
After all the different feedstocks were tested, the results were aggregated to produce an average
liquid yield as depicted below in Figure 10.2

Figure 10.2 – Feedstock Average Liquid Yield

From a liquid yield perspective, that the results are primarily as expected given that all liquid
yield is primarily a function of the amount of cellulose contained within each type of biomass.
Feedstocks that have a high ratio of cellulose to lignin generally produce more liquid yield and
less char product compared to other feedstocks.

From the liquid yield testing, interesting things to note include:

• Clean white wood feedstocks almost always had the highest liquid yield. With cleaner
feedstock, there is less bark that can be turned into char.

• Albizia’s lower liquid yield in comparison to other woody biomass feedstocks may possibly
be offset by its lower cost, given that it can be an invasive species.

• The discrepancy between corn stover liquid yield and de-ashed corn stover is substantial.
Ash content will always have a direct effect on liquid yield, but it is of vital importance for
agricultural residues because of the tendency to pick up other non-organic materials during
feedstock harvesting.

• The pyrolysis oil from algae yielded a very low amount of liquid, but the primary problem
was the unusual viscosity. The pyrolysis oil was more like tar than liquid, and would likely
be difficult for upgrading.
Page | 80
From all the feedstocks tested, the high char yield for the dirty corn stover and algae were
identified as likely problems. For the corn stover, some of this can be alleviated through cleaning of
the feedstock before processing, but de-ashing is likely required to get the maximum liquid yield
from both the corn stover and the algae.

As previously mentioned, the algae produced pyrolysis oil that was similar to tar in consistency and
very difficult to handle. Upon further testing, the algae pyoil quickly polymerized when exposed to
heat, and therefore would likely be very difficult to upgrade to transportation fuels through either
hydroprocessing or co-processing.

The bench scale unit testing found no “show stoppers” for most of the biomass feedstocks with the
exception of the algae. While the preference for processing dense feedstocks with high cellulose
content will remain, other feedstocks will be suitable for pyrolysis conversion depending on
geographic region and cost.

10.2 Integrated Biorefinery Feedstock Requirements

Biomass feedstock requirements for the development of an RTP facility can be substantial,
depending on its proximity to a geographical feedstock basket. The amount of biomass required
for an integrated biorefinery depends on what type of further upgrading to transportation fuels
is desired.

For upgrading of pyrolysis oil through hydroprocessing to neat renewable fuel blendstocks,
economies of scale become a factor due to the large capital expenditures for the hydroprocessing
unit. Even though the technology pathway is not fully developed, it is expected that a
hydroprocessing facility would require 10,000 barrels per day (420,000 gallons per day) of
pyrolysis oil to make the economics feasible. This volume of pyrolysis oil would require six
RTP 400 units to supply the hydroprocessing upgrading unit.

An RTP 400 unit processes 132,000 bone dry metric tonnes (BDMT) of biomass per year.
Therefore, the six RTP units for an upgrading facility will require 792,000 BDMT of biomass per
year. While this quantity seems large, there are multiple potential biomass baskets around the
United States capable of supplying in excess of this quantity.

In Figure 10.3, areas with large quantities of forest residuals are depicted in the Southeast,
Northeast, upper Midwest and Pacific Coast that could easily supply this amount of biomass.
From these areas alone, there is more than 18,000,000 BDMT of forest residues available
annually. Nationwide, forestry residue has the potential to supply about 100 million tons per
year of biomass feedstock for advanced biofuels production.

Page | 81
 Figure 10.3 – Major sources of forest residues in the US (per NREL)

For agricultural crop residues, the potential supply is even larger. The Midwestern farming states
can supply more than 45,000,000 BDMT per year from crop residues available for biomass
conversion. See Figure 10.4 below. Corn stover has been estimated to potentially supply
of 250 million tons per year of biomass feedstock produced across a wide cross-section of the
continental US. But note; there is still uncertainty regarding the quantity of agricultural residues
that can be removed from farmland without potentially affecting the soil quality.

Figure 10.4 – Major sources of crop residues in the US (per NREL)

Page | 82
Dedicated 2nd generation energy crops also have the potential to provide greater than 100 million
tons per year of feedstock for biofuels production. Two of the leading energy crops with high
potential for US cultivation are switchgrass and algae.

The biomass requirements for co-processing are not as large and therefore can expand the
number of potential sites that are able to deliver pyrolysis oil to a petroleum refinery. A single
RTP 400 unit that processes 132,000 BDMT per year produces a nominal 20,000,000 gallons of
pyrolysis oil. This quantity is ideal for a 50,000 BPSD FCC unit to blend 3% pyrolysis oil along
with their vacuum gas oil (VGO) feed to produce partially renewable fuel blendstocks.

Figure 10.5 below depicts the location of petroleum refineries across the United States. Given the
dispersion of refineries nationwide, availability of biomass should not be a limiting factor for the
commercialization of pyrolysis oil upgrading or co-processing.

Figure 10.5 – US Petroleum Refinery Locations

The refining industry currently uses a number of methods to meet their renewable fuel mandates
and often the decision on which path to take is dependent on location and availability of biofuel
options. Being able to site an RTP 400 unit within close logistical proximity is an important factor.
Based on analysis of the forest and agricultural residues, an RTP project can be sited in almost
any location nationally to service refining customers.

Page | 83
10.3 Life Cycle Analysis of Pyrolysis Oil Upgrading

A large determinant in the market viability of upgrading pyrolysis oil is the amount of greenhouse
gas (GHG) reduction provided by the finished fuel. The EPA’s Renewable Fuel Standard (RFS)
requires minimum GHG reductions for a fuel to qualify under each category in the program.

Cellulosic biofuels under the RFS program must reduce GHGs by a minimum of 60% compared to
petroleum transportation fuels. At the time the RFS regulations were approved, transportation fuels
(gasoline and diesel) contained a blended average of 90.2 gCO2-equivalent per MJ. Therefore,
for a fuel to qualify as cellulosic, it must contain less than 36.2 gCO2e per MJ.

As part of the integrated biorefinery project, Michigan Technological University (MTU) completed
a Life Cycle Analysis (LCA) of pyrolysis oil upgrading to transportation fuels for multiple biomass
feedstocks based on laboratory scale data provided by UOP.

The LCA project boundary covered the entire “well to wheel” or “field to wheel” pathway from
biomass feedstocks to the combustion of finished transportation fuel. This LCA boundary is
shown below in Figure 10.6.

Figure 10.6 – LCA Project Boundary (from MTU study)

Ethanol Electricity

Fertilizer Biomass Biomass Biomass

Fuel
Production Transport Pretreatment

Electricity Pyrolysis Stabilization & PyOil Natural Gas

Natural Gas Transport Upgrading
Steam RTP Process Steam

Biofuel Biofuel
Transport Combustion

Filter Cake

Page | 84
The results of this LCA analysis are depicted below in Figure 10.7 for nine types of biomass
feedstocks. Note, all feedstocks with the exception of wild algae reduced GHG emissions by 60%
to less than 36.2 gCO2e per MJ and qualifies as cellulosic biofuels under the RFS program.

Figure 10.7 – Quantity of GHGs by Feedstock Type (from MTU study)

60% reduction level

UOP is confident that fuels produced from hydroprocessing of pyrolysis oil from second generation
feedstocks should qualify as cellulosic biofuels based on this study and others investigating similar
technologies. The Michigan Tech LCA Executive Summary is included as an Appendix to this
report.

Independent of this project, a life cycle analysis of co-processing pyrolysis oil in an FCC unit at a
refinery has shown renewable fuels with less than 25 gCO2e per MJ. This result gives assurance
that co-processed fuels will also qualify as cellulosic biofuels.

Page | 85
11. Project Lessons Learned
Although the project faced multiple challenges, many lessons were learned that can be of value
to future projects in the area of biomass conversion and upgrading to transportation fuels. This
section presents a summary of these lessons.

General Project

• The location of the one ton per day RTP unit in Hawaii presented significant logistical
challenges and added cost to the project. Functions such as contracting, purchasing, shipping,
staffing and maintenance were made more difficult and costly by the remote location. Costs of
airfare to the project site and housing for project staff were also higher. Future projects should
consider using sites in more accessible, lower cost locations.

• Weekly update conference calls and annual Comprehensive Project Review meetings are
an important communication link between the DOE and industrial partners. Frequent
communications about project status and issues are needed to help keep the project on
track and ensure an effective collaboration of the industrial partners with the DOE.

RTP Operations

• Ensure that biomass feedstock specifications for the RTP operations are clearly detailed and
are adhered to during pretreatment. During biomass preparation, INL was unable to meet the
size specification for softwood, switchgrass, and corn stover, and the biomass had “little-stick”
morphology. The off-spec morphology of the softwood caused many operational difficulties and
product quality issues in the RTP unit.

• A “trash” screen should be added to the biomass feed system to catch oversize pieces of
biomass and other debris before they get into the RTP unit. This would help to avoid
operational problems.
• Heat losses need to be factored into the RTP design for smaller scale pilot units, based on
the needs of the project. Heat losses at the one ton per day scale had an effect on the energy
balance of the unit and required the usage of torch oil to keep the unit running. Based on
previous experience with larger units, this was not expected. Smaller units may need to be
designed differently than larger units to minimize heat losses; simply scaling down a larger
design will not necessarily work.

MFU Operations

• Industry standard DCS and PLC systems should be used to ensure uniformity and ease of
technical support and maintenance. Using a custom-designed system may potentially lead to
integration issues or problems from sole-source of supply.

Page | 86
• The MFU design should be customized for pyrolysis oil service. Ensure that all materials of
construction, including piping and seals in pumps and valves, have been verified for
compatibility with pyrolysis oil. Ion exchange resin beds and liquid distributors should also
be designed specifically for pyrolysis oil.

Pyrolysis Oil Upgrading

• The primary shortcoming of the project was the inability to successfully upgrade pyrolysis oil to
transportation fuel blendstocks by hydroprocessing at the pilot plant scale. Hydroprocessing
upgrading was the step-change technology at the center of the integrated biorefinery project.
The difficulties encountered in pyrolysis oil upgrading were the primary obstacle that resulted in
project delays, budget overruns and project scope changes. Future projects should carefully
consider the potential failure modes in the early stages and clearly identify where technology
breakthroughs will be needed to achieve the project objectives.
• Upgrading of pyrolysis oil by standard fixed-bed hydroprocessing was more difficult than
expected due to reactor plugging caused by polymerization of the feed at normal operating
temperatures. An alternative type of reactor design (e.g., slurry) may be needed.

• More fundamental research is needed to understand how pyrolysis oil polymerization (and
reactor plugging) in the first stage of upgrading can be prevented. The project team was not
able to find a solution to prevent polymerization of pyrolysis oil in the first stage, fixed bed
reactor system. Although some successful results were obtained in laboratory testing,
challenges remain in demonstrating the solution at a larger scale.
• Due to the project timeline, design of the demonstration scale UG-II process for Kapolei began
before the fixed-bed hydroprocessing concept was proven in the pilot plant at Des Plaines.
Considerable cost and effort went into designing the UG-II demonstration unit that ultimately
was not built due to the inability to demonstrate the technology in the pilot plant. Future projects
should ensure that technology is thoroughly proven at the pilot scale before embarking on the
design of a larger scale unit.

• Complex processes like UG-II should be broken down into smaller sub-processes
(e.g., reaction and distillation) and each sub-process tested individually before attempting to
test the entire process in an integrated manner. This would facilitate troubleshooting of each
sub-process without having to consider how the sub-processes interact with each other.
Significant progress was made in the pilot plant after this approach was taken in late 2014
(i.e., operation of the first-stage reactor in once-though mode without recycle and offline
product fractionation).

Page | 87
 Appendix
UOP, LLC

Pilot-Scale Biorefinery: Sustainable Transport Fuels from Biomass and

Algal Residues via Integrated Pyrolysis, Catalytic Hydroconversion and
Co-processing with Vacuum Gas Oil

Award No. EE0002879

 Appendix A

Pacific Northwest National Laboratory

Note:

This appendix is the final report provided by Pacific Northwest National Laboratory on
Wastewater Cleanup by Catalytic Hydrothermal Gasification. This work was part the IBR
project scope.
 PNNL-24386

Prepared for the U.S. Department of Energy

under Contract DE-AC05-76RL01830

Pilot-Scale Biorefinery: Sustainable

Transport Fuels from Biomass via
Integrated Pyrolysis and Catalytic
Hydroconversion – Wastewater
Cleanup by Catalytic Hydrothermal
Gasification
DC Elliott
MV Olarte
TR Hart

June 2015
 PNNL-24386

Pilot-Scale Biorefinery: Sustainable Transport

Fuels from Biomass via Integrated Pyrolysis
and Catalytic Hydroconversion – Wastewater
Cleanup by Catalytic Hydrothermal
Gasification

DC Elliott
MV Olarte
TR Hart

June 19, 2015

Prepared for
the U.S. Department of Energy
under Contract DE-AC05-76RL01830

Pacific Northwest National Laboratory

Richland, Washington 99352
 Executive Summary

DOE-EERE Bioenergy Technologies Office has set several goals to increase the use of
bioenergy and bioproducts derived from renewable resources. One of these goals is to facilitate
the implementation of the biorefinery. The biorefinery will include the production of liquid
fuels, power and, in some cases, products. The integrated biorefinery should stand alone from an
economic perspective, with fuels and power driving the economy of scale while the
economics/profitability of the facility will be dependent on existing market conditions.

UOP LLC proposed to demonstrate a fast pyrolysis-based integrated biorefinery. Pacific

Northwest National Laboratory (PNNL) has expertise in an important technology area of interest
to UOP for use in their pyrolysis-based biorefinery. This Cooperative Research and
Development Agreement (CRADA) project provides the supporting technology development
and demonstration to allow incorporation of this technology into the biorefinery. PNNL
developed catalytic hydrothermal gasification (CHG) for use with aqueous streams within the
pyrolysis biorefinery. These aqueous streams included the aqueous phase separated from the fast
pyrolysis bio-oil and the aqueous byproduct streams formed in the hydroprocessing of the bio-oil
to finished products.

The purpose of this project was to demonstrate a technically and economically viable
technology for converting renewable biomass feedstocks to sustainable and fungible
transportation fuels. To demonstrate the technology, UOP constructed and operated a pilot-scale
biorefinery that processed one dry ton per day of biomass using fast pyrolysis. Specific
objectives of the project were to:

• Identify acceptable renewable feedstocks that are sustainable, currently available and/or
whose availability can be increased in the future.
• Demonstrate that these feedstocks can be pyrolyzed to an oil suitable for further
processing.
• Demonstrate that pyrolysis oil can be successfully upgraded to transportation fuels.
• Define the economics of the process.
• Perform a Life Cycle Analysis (LCA) of the process.

The anticipated outcomes of the project were a validated process technology, a range of
validated feedstocks, product property and life cycle data, and technical and operating data upon
which to base the design of a full-scale biorefinery. The anticipated long-term outcomes from
successful commercialization of the technology were: (1) the replacement of a significant
fraction of petroleum based fuels with advanced biofuels, leading to increased energy security
and decreased carbon footprint; and, (2) establishment of a new biofuel industry segment,
leading to the creation of U.S. engineering, manufacturing, construction, operations and
agricultural jobs.

PNNL development of CHG progressed at two levels. Initial tests were made in the
laboratory in both mini-scale and bench-scale continuous flow reactor systems. Following

i
positive results, the next level of evaluation was in the scaled-up engineering development
system, which was operated at PNNL.

The results of tests with bio-oil aqueous fraction suggested that the organic contaminants,
even at high concentration, can be effectively converted to a fuel gas product by CHG.
However, corrosion of the stainless steel reactor system by the acidic components appeared to be
significant, based on the metal deposits on the catalyst, but no noticeable change was seen in the
reactor itself. Sulfur stripping may also be necessary in order to maintain long-term catalyst
activity. Carbon fouling of the catalyst was also suggested by the loss of porosity and surface
area. Handling of the aqueous stream was also identified as an issue as the acidic components
were corrosive to steel cans and could not be stored for any period of time in such without
resulting in a significant iron contamination and even failure of the container.

ii
 Acronyms and Abbreviations

BET Brunauer-Emmett-Teller
BJH Barrett-Joyner-Halenda
BPR back-pressure regulator
Btu British thermal unit
CHG catalytic hydrothermal gasification
COD chemical oxygen demand
CRADA cooperative research and development agreement
CRS continuous-flow reactor system
CSTR continuously stirred tank reactor
DOE U.S. Department of Energy
EERE Energy Efficiency and Renewable Energy Office
GC gas chromatography
HHV higher heating value [MJ/kg]
HPLC high performance liquid chromatography
IBR integrated biorefinery
IC ion chromatography
ICP inductively coupled plasma, atomic emission spectroscopy
ID internal diameter
LC liquid chromatography
LCA life cycle analysis
LHSV liquid hourly space velocity
MAWP maximum allowable working pressure
MJ mega-Joule
MSRS mobile scaled-up reactor system
NA not analyzed
ND below detection limit
ppm parts per million
P&ID process and instrumentation diagram
SCF standard cubic foot
SS stainless steel

iii
 Contents
Executive Summary ............................................................................................................................. i
1.0 Introduction ................................................................................................................................ 1
2.0 Methods and Equipment ............................................................................................................. 3
2.1 Lab-scale continuous-flow reactor system ......................................................................... 3
2.2 Mini-reactor system............................................................................................................ 6
2.3 Scaled-up reactor system.................................................................................................... 7
3.0 Bench-Scale Processing Tests .................................................................................................. 10
3.1 Lab-scale tests with fast pyrolysis product derivatives .................................................... 10
3.1.1 CT32-33 (11/5/2010)............................................................................................. 10
3.1.2 CT45-1 (3/30-31/2011) ......................................................................................... 10
3.1.3 CS6-06 (4/26-29/2011) ......................................................................................... 11
3.1.4 CT45-2 and CT45-3 (5/23-24 & 6/7-8/2011)........................................................ 11
3.2 Mini-reactor tests with UOP-derived hydrotreating byproduct feedstocks ...................... 13
3.2.1 UOP1 aqueous bio-oil upgrading (9/21-22/2011) ................................................. 15
3.2.2 UOP2 fixed bed HT byproduct aqueous (10/13/2011).......................................... 15
3.2.3 UOP3 mixed HT byproduct aqueous (12/14-15/2011 & 1/3-4/2012)................... 15
3.2.4 Catalyst bed analysis following aqueous product CHG ........................................ 16
3.2.5 UOP4 HT (3/14-15/2012) ..................................................................................... 17
3.2.6 UOP 4B HT (8/7/2012) ......................................................................................... 18
3.2.7 UOP 4C HT (9/6-7/2012) ...................................................................................... 18
3.2.8 UOP5 HT (11/30-12/1/2012) ................................................................................ 19
3.2.9 UOP6 HT (2/21/2013 – 2/23/2013)....................................................................... 22
3.2.10 UOP7 HT (3/13/13 – 3/15/13)............................................................................... 25
3.2.11 UOP8 HT (3/28/2013 – 3/30/2013)....................................................................... 28
3.2.12 UOP9 HT (12/17/2013 – 12/19/2013)................................................................ 31
3.3 Mini-reactor tests with model compound mixtures .......................................................... 33
3.3.1 L1 (8/27/2013 – 9/6/2013) .................................................................................... 33
3.3.2 L2 (1/6/2014 – 1/15/2014) .................................................................................... 37
3.4 Mini-reactor tests with PNNL hydrotreater aqueous products ......................................... 39
3.4.1 HT-AQ-CHG-1 (8/20/2014 – 8/29/2014) ............................................................ 40
3.4.2 HT-AQ-CHG-2 (10/3-5/2014, 11/12-14/2014, and 11/22-23/2014).................... 41
3.4.3 HT-AQ-CHG-3 (2/3-4 and 7-8/2015) .................................................................. 43
4.0 Scaled-up reactor tests with simulated hydrotreater byproduct water ...................................... 45
5.0 Conclusions and Recommendations ......................................................................................... 45
6.0 References ................................................................................................................................ 50

iv
 Figures
Figure 1. Schematic of Catalytic Hydrothermal Gasification ............................................................. 3
Figure 2. Process Flow Schematic of the Bench-Scale Continuous-Flow Reactor System ............... 4
Figure 3. Mini-reactor Assembly ....................................................................................................... 7
Figure 4. Process and Instrumentation Diagram of the mobile CHG reactor system ......................... 8
Figure 5. Product gas composition ................................................................................................... 10
Figure 6. Feed showing segregation of oil and aqueous phase. Supernatant became clearer after
days of standing in the refrigerator ........................................................................................... 22
Figure 7. Reactor assembly ............................................................................................................... 23
Figure 8. COD values of the experiment .......................................................................................... 24
Figure 9. CO2 and CH4 distribution during the run........................................................................... 24
Figure 10. Feed showing segregation of oil and aqueous phase. ...................................................... 25
Figure 11. COD values of the liquid products .................................................................................. 26
Figure 12. CO2 and CH4 distribution during the run ......................................................................... 27
Figure 13. IBR-UOP8 Feed. ............................................................................................................. 28
Figure 14. COD values of the product water .................................................................................... 30
Figure 15. CO2 and CH4 distribution of the gas product................................................................... 30
Figure 16. Oil scum on container wall for IBR-UOP9 feed. ............................................................ 31
Figure 17. COD values of the product water and gas flow rate in UOP9 HT................................... 32
Figure 18. Gas composition and flow rate in UOP9 HT ................................................................... 33
Figure 19. COD values of the product water and gas flow rate in model compound test L1 ........... 34
Figure 20. Relationship of gas flow rate and LHSV in model compound test L1 ............................ 35
Figure 21. Carbon conversion to gas, COD conversion and mass balance in model compound test
L1 .............................................................................................................................................. 35
Figure 22. Gas product composition in model compound test L1 .................................................... 36
Figure 23. Comparison of model compound feed and aqueous product........................................... 36
Figure 24. COD values of the product water and gas flow rate in model compound test L2 ........... 38
Figure 25. Carbon conversion to gas, COD conversion and mass balance in model compound test
L2 .............................................................................................................................................. 38
Figure 26. Gas product composition in model compound test L2 .................................................... 39

v
 1.0 Introduction

DOE-EERE Bioenergy Technologies Office has set forth several goals to increase the use
of bioenergy and bioproducts derived from renewable resources. One of these goals is to
facilitate the implementation of the biorefinery. The biorefinery will include the production of
liquid fuels, power and, in some cases, products. The integrated biorefinery (IBR) should stand-
alone from an economic perspective with fuels and power driving the economy of scale while the
economics/profitability of the facility will be dependent on existing market conditions.

UOP LLC proposed to demonstrate a fast pyrolysis-based integrated biorefinery. PNNL

has expertise in an important technology area of interest to UOP for use in their pyrolysis-based
biorefinery. This CRADA project provided the supporting technology development and
demonstration to allow incorporation of the technology into the biorefinery. PNNL developed
catalytic hydrothermal gasification (CHG) for use with aqueous streams within the pyrolysis
biorefinery. These aqueous streams included the aqueous phase separated from the fast pyrolysis
bio-oil but, more importantly, the aqueous byproduct streams formed in the hydroprocessing of
the bio-oil to finished products.

The purpose of this project was to demonstrate a technically and economically viable
technology for converting renewable biomass feedstocks to sustainable and fungible
transportation fuels. UOP was tasked with demonstrating the technology by constructing and
operating a pilot-scale integrated biorefinery that could process one dry ton per day of biomass
using fast pyrolysis. Subsequent upgrading to transportation fuels was not attempted at the IBR.
Specific objectives of the project were to:

• Identify acceptable renewable feedstocks that are sustainable, currently available and/or
whose availability can be increased in the future.

• Demonstrate that these feedstocks can be pyrolyzed to an oil suitable for further
processing.

• Demonstrate that pyrolysis oil can be successfully upgraded to transportation fuels.

• Define the economics of the process.

• Perform a Life Cycle Analysis of the process.

The anticipated outcomes of the project were a validated process technology, a range of
validated feedstocks, product property and life cycle data, and technical and operating data upon
which to base the design of a full-scale biorefinery. The anticipated long-term outcomes from
successful commercialization of the technology are: (1) the replacement of a significant fraction
of petroleum based fuels with advanced biofuels, leading to increased energy security and
decreased carbon footprint; and (2) establishment of a new biofuel industry segment, leading to
the creation of U.S. engineering, manufacturing, construction, operations and agricultural jobs.

1
 PNNL was tasked with developing CHG for use with aqueous streams within the
pyrolysis biorefinery. These aqueous streams included the aqueous phase separated from the fast
pyrolysis bio-oil and the aqueous byproduct streams formed in the hydroprocessing of the bio-oil
to finished products. As other aqueous streams were identified as of interest within the
biorefinery, these might also have been tested in CHG, but none were so identified. The
development work progressed at two levels. Initial tests were made in the laboratory in mini-
reactor scale and bench-scale continuous-flow reactor systems. Following positive results, the
next level of evaluation was in the scaled-up engineering development system.

The Statement of Work for this CRADA project included two tasks for PNNL.
Task 1.
PNNL completed development tests of CHG for use with aqueous streams within the pyrolysis
biorefinery. These aqueous streams included the aqueous phase separated from the fast pyrolysis
bio-oil and the aqueous byproduct streams formed in the hydroprocessing of the bio-oil to
finished products. The initial development work was done in the bench-scale continuous-flow
reactor systems. Subsequent long-term tests were performed in the micro-scale reactor system.

Task 2.
Following initial tests in Task 1, in the laboratory, the next level of evaluation were in the scaled-
up engineering development system. The demonstration tests were performed at Richland,
Washington.

The Statement of Work included the following milestones for PNNL.

Milestones (months from start)

MS-1 Demonstrate effectiveness of CHG to recover energy from 42

aqueous waste streams and meet NPDES site discharge requirements

MS-2 Verification of CHG treatment of wastewater streams produced during 59

scaled-up bio-oil hydroprocessing tests

MS-3 Completion of final report 60

2
 2.0 Methods and Equipment

CHG of organic contaminated wastewaters provides a highly efficient pathway to

medium-Btu fuel gas. This gas product can be used directly in heat and power applications or
has potential to be cleaned to pipeline quality gas or feedstock for hydrogen production. As
compressed or liquefied natural gas, it has potential to displace imported petroleum used in
transportation applications. Hydrothermal processing utilizes water-based solutions at moderate
temperatures (350 °C) and sufficient pressure (3,000 psi) to maintain the water in the liquid
phase as depicted in Figure 1. Previously, papers have been published addressing the processing
environment, i catalyst systems for this environment, ii,iii continuous-flow reactor tests with fixed
beds of catalyst in a tubular reactor, iv and process development tests with wet biomass
feedstocks. v

Figure 1. Schematic of Catalytic Hydrothermal Gasification

The use of hydrothermal processing (high-pressure, high-temperature liquid water) has

received relatively limited study. vi One application of this processing environment has been
demonstrated in the catalytic gasification of [Link] In this application, heterogeneous metal
catalysts accelerate the reaction of organics with water and produce methane and carbon dioxide
as the product gases. It has been reported both as a means of recovering useful energy from
organic-in-water streams and as a water treatment system for wet organic contaminants.

Developing catalysts for this processing environment has also been an important factor in
making this processing technology [Link],iii More recently, we have demonstrated more stable
catalyst formulations for wet gasification as described in patent claims. vii,viii

The equipment and procedures described below were used for the testing.

2.1 Lab-scale continuous-flow reactor system

The bench-scale Continuous-flow Reactor System (CRS) was designed for obtaining
engineering data for the continuous-flow CHG process. The system consists of the high-pressure

3
pump feeding system, product recovery system, data acquisition and control system, furnaces,
and other equipment required to utilize the 1-liter tubular catalytic reactor. The feed line,
operated at ambient temperature, was ½-inch 316 stainless steel (SS) tubing with 0.049 inch
wall. All process lines at temperature of 200 ºC or above were ¼-inch 316 SS tubing with 0.065-
inch wall. The system was based on a throughput of 1.5 liter of solution per hour and was
typically operated over a test period of 6-10 h. The process flow diagram is shown in Figure 2.
The solids separator and sulfur stripper were bypassed in these tests as the mineral content and
sulfur content were considered too low to be of concern.

The pumping subsystem was designed to operate at feed pressures up to 3700 psig and
flow rates between 0.2 and 4.0 L/h. This system consists of an Isco 500D pump. The valves and
tubing on the high-pressure metering syringe pumps were configured to fill and empty the pumps
based on controller commands. System piping included 0.5-inch (0.065-wall) 304 tubing SS on
the outlet of the pump. Pump inlet piping was 0.5-inch (0.035 wall) 304 SS tubing. All valves
and valve trim (except the pressure-control valve) were also made of SS. Using the Isco pump,
the feeding rates were measured directly by the screw drive of the positive displacement syringe
pump.

Figure 2. Process Flow Schematic of the Bench-Scale Continuous-Flow Reactor System

4
 The gasification reactor was a 1-inch ID X 72-inch-long 304 SS tube. The vessel
(MAWP 10,000 psi @ 72 ºF, or approximately7500 psi @ 400 ºC) had bolted-closure endcaps
with metal o-rings on each end. The reactor furnace was a 6-kWe resistance heater split into
three separately controllable zones. The pressure was controlled with a dome-loaded diaphragm
back-pressure regulator (BPR).

After exiting the reactor, the product gases and water were cooled and vented through a
BPR to an online phase separator. After separation the offgas was cooled by another chilled heat
exchanger to further remove any entrained water, the liquid was accumulated in a weighed tank.
The offgas was measured by a wet test meter and analyzed by a gas chromatograph. Pressure
transducers on each vessel recorded pressures and noted pressure drops due to restriction and
plugging. Each vessel and most transfer lines were also monitored for temperature. Three
rupture discs protected the system.

The data acquisition and control system used in the CRS was a hybrid computer-based
system employing discrete data acquisition devices and single-loop process controllers
communicating to a central computer. The computer was used during experiments to monitor
the process, calibrate instruments, and record data for later analysis. The data acquisition/control
system regulated the furnaces and recorded the process parameters and offered off-normal
warnings and auto-shut down. LabVIEW was used to coordinate these activities. Non-control
sensors such as thermocouples and pressure transducers were monitored via the data acquisition
unit.

Actual startup of the experiment usually required 2 to 4 h to bring operating conditions to

the desired levels. Operating data were recorded, and data windows were defined based on
steady-state (or near steady-state) operating conditions.

Gas samples could be withdrawn manually and analyzed every 30 to 60 min. The
gaseous stream was mainly composed of CO2, CH4, H2, and C2+ hydrocarbons, as well as water
vapor. Gas analysis was performed by gas chromatography (GC) as described [Link] The gas
samples were withdrawn after cooling and depressurizing the product effluent to near ambient
conditions. In the process, the gas product is effectively scrubbed by the liquid aqueous
byproduct. As a result, the liquid byproduct contains some dissolved product gases. Particularly
in the case of carbon dioxide, it is retained in the water by participating in acid/base chemistry
with alkali cations derived from the feedstocks.

Once the gas samples from the experiments were analyzed, calculations were made to
determine the conversion of the organic feedstock to gases. Carbon conversion to gas was then
calculated on a mass basis for the carbon in the product gases as a percent of the carbon in the
feedstock. The carbon balance is the key elemental balance for this process.

The liquid effluent was analyzed for chemical oxygen demand (COD) and pH. Percent
COD reduction from the feed to product was a major process indicator, which was monitored to
determine catalyst effectiveness. Anions, including chloride, were measured by ion
chromatography (IC) using a Dionex DX 500 IC consisting of a GP40 Pump, EG40 Elluent
Generator, ED40 Electrochemical Detector, with an AS3500 autosampler. An ASRS-Ultra 4

5
mm suppressor was used (at 100 mW) to minimize baseline drift. The chromatography was
performed using an AG-11 guard column and an AS-11HC column running at 30 °C, with an
hydroxide gradient from 0.5 mM to 41 mM and a flow rate of 1.2 ml/min. Certified standards
were used to calibrate the IC.

Elemental analysis of liquid and solid samples was performed by inductively coupled
plasma - optical emission spectrometry (ICP). The ICP was a Perkin-Elmer 3000DV with an
AS90 Autosampler, which has an instrument detection limit of about 1 ppb (for most elements)
with a linear calibration up to 100 ppm (for most elements). Solid samples were prepared via
microwave digestion in concentrated nitric and hydrochloric acids, then diluted to volume. The
ICP was calibrated and verified with two independent certified standard sets. Spikes and
dilutions were down for each batch of samples to check for and/or mitigate any matrix effects.
The ICP process ran a constant pump rate of 1.5 ml/min for all samples and standards during
analysis. A 3 ml/min rinse and initial sample flush was used to switch between each sample and
standard. The plasma was run at 1450 W with argon flow. Trace metal grade (sub-ppb) acids
and two independently NIST-certified calibration standard sets were used for calibration and
method verification.

2.2 Mini-reactor system

Essentially the same methods were followed in the experiments using the mini-reactor
system with the exception that the mini-reactor was
designed for unattended, around-the-clock operation
to validate long term processing.

The mini-reactor process flow was similar to

the earlier bench-scale tests. A separate solids
settler portion of the reactor system was connected
into the flow line downstream of the preheater. In
this way, mineral precipitate could be settled and
collected without plugging the catalyst bed and,
subsequently, the reduced sulfur could be scrubbed
from the stream by a bed of nickel particles, to
reduce the poisoning of the catalyst bed. In the
initial tests there was no sulfur scrubbing component
used. After it was determined that there was a need
for sulfur scrubbing was the nickel metal bed placed
into the flow. The mineral precipitation capability
was no needed in any of the tests due to the low
mineral content of the feedstocks.

The main reactor has a length of ½” tube

(0.049” wall) with a “t” fitting on the end, as shown
Figure 3. Mini-reactor Assembly in Figure 3 — overall it is about 14.5 inches long

6
with a 1/8” thermocouple down the center line. A calculation of the volume would be somewhat
higher than the 22 cc of the Ru/C catalyst used in the experiment. The volume of catalyst was
determined from the measured mass and calculated based on an apparent bulk density that was
also measured (0.42 g/ml). Therefore, the catalyst added to the reactor must have not completely
filled the annular space due to some hang-up around the thermocouple. The catalyst usage
(liquid hourly space velocity) was based on the actual amount of catalyst added for the test. The
Ru/C catalyst is a BASF formulation, which contains 7.8 wt % Ru on a proprietary graphitized
carbon extrudate.

2.3 Scaled-up reactor system

The mobile scaled-up reactor system (MSRS) was designed based on the bench-scale
unit. It was used previously for on-site demonstrations of CHG at industrial plants. ix For the
iv

tests reported here, it was updated with new electronic controls and data logging system, based
on the Honeywell system used at the Hawaiian IBR site. In addition to the 500D ISCO pump, a
high capacity version HPX was added to the mobile unit. With the two pumps, the capacity was
increased to 20 L/h to utilize the four tubular reactors at full capacity of LHSV at up to 5 L of
feedstock solution, L-1 of catalyst h-1. The P&ID of the mobile system is shown in Figure 4.

Another modification completed on the mobile unit was the addition of a recirculating
cooler to eliminate the once-through cooling in the original system. For this application the
solids separation was not needed so the settling tank and blow-down system was bypassed.

The MSRS is housed in a 20-ft x 8-ft trailer that is divided into two sections. The smaller
of the two sections, located at the front of the trailer, serves to house the data acquisition and
control equipment, the analytical equipment, all relevant documentation, and operator controls.
The larger of the two sections located at the back of the trailer houses the reactor system.

The reactor system is completely self-contained within the back portion of the trailer.
There is an operator access corridor located at the front of the system which provides access to
the feed pumps, gas/liquid separator, and the back pressure regulators, as well as the switch panel
to activate all of the system’s components. Access to the back portion of the trailer is limited to
the access corridor, once the system is under pressure and fluid temperatures exceed 100 °C.

This system is designed to operate at relatively high temperatures and medium pressures
at a MAWP of 426 °C and 3400 psig. The conversion reaction typically employs a catalyst to
accelerate steam reforming and methane synthesis reactions to produce a useable, medium-BTU
fuel gas.

7
 Figure 4. Process and Instrumentation Diagram of the mobile CHG reactor system

The objectives for testing are to determine the potentials for waste volume reductions, energy
conversions, and catalyst performance for the organics-laden, aqueous feedstocks being produced by bio-
oil processing.

The MSRS is comprised of heaters and pumps and tubing. The process pumps organics in water
up to 3000 psig and 350 °C. The solution is passed through a filter to remove any inorganics, then across
a catalyst bed to gasify the organics to mostly carbon dioxide and methane. The heat is recovered via
tube-in-tube heat exchangers. The product is further cooled with a closed loop chiller to ambient
temperature. The product is relatively clean process water and gas. The pressure is let down and the
water/gas is separated and quantified. A gas chromatograph is used to determine the composition of the
product gas. Pressure is maintained via two parallel, two-stage TESCOM back-pressure regulator
combinations.

8
MAWP 3400 psig 426 °C
Nominal operation 3000 psig 350 °C
Liquid Flow rates 4-20 L/h
Gas production rates 10 – 2000 L/h Organic loading dependent
Gas composition 58% CH4, 40%CO2,
2% H2, trace other
Component ratings:
Tubing after cooling, ½” 0.049” 3700 psig Ambient to 100 °C
Tube in tube HX (Shell side) ¾” 4582 psig 426 °C
0.109” tubing
Tube in Tube HX (tube side) 3/8” 5135 psig 426 °C
0.065” tubing
M1 heater tubing 3/8” 0.065” 5135 psig or (4940 psig 426 °C or (538 °C heater max)
H2 max)
Reactors HIP (TOC11-40) 5950 psig 426 °C
Solids separator 1.8L (4680) 4200 psig 600 °C
BPR’s two stage letdown TESCOM’s 6000 psig 74 °C
Gas liquid separator After BPR’s (low 150 psig ambient
pressure side) PRV set at 40 psig

9
 3.0 Bench-Scale Processing Tests

3.1 Lab-scale tests with fast pyrolysis product derivatives

The initial tests in this project focused on an aqueous fraction of bio-oil in anticipation that the
bio-oil feedstock for hydrotreating could be improved by separation of the water and some of the light
oxygenate components. These low molecular weight components were viewed as not useful for
production of liquid hydrocarbon fuels and the hydrotreating of these materials would be a potential waste
of hydrogen.

3.1.1 CT32-33 (11/5/2010)

The first sample of aqueous bio-oil received from UOP (CT32-33) was shipped in a steel 5-gallon
can. Corrosion of the can was evident. ICP analysis of the sample showed 1970 ppm of Fe in the liquid.
Liquid chromatography analysis identified the primary components as: 8.33% acetic acid, 2.71%
hydroxyacetone, 1.36% methanol, 0.86% formic acid, and 0.32% ethanol

This sample was processed in the bench-scale CRS using

the standard conditions of 350°C, 3000 psig, and a LHSV of
1.7 L feed - L-1 of catalyst - h-1. In the test the COD was
reduced from the starting level of 165,300 ppm by 99.96%,
producing a process effluent at 135 ppm COD. The gas
product yield was 0.91 L / g organic of a medium-Btu gas
with 627 Btu/SCF and the composition depicted in Figure 5.
The test was performed with pH adjustment by addition of
NaOH and Na2CO3, to increase the feed pH of 3.4 to >5,
Figure 5. Product gas composition which led to salt precipitation in the catalyst bed, plugging
from CHG of aqueous bio-oil. the reactor and terminating the test. pH adjustment was not
deemed necessary for this work and was not used again.

3.1.2 CT45-1 (3/30–31/2011)

The second sample of aqueous bio-oil (CT45) was received after shipment in a plastic bucket.
Liquid chromatography analysis identified essentially the same primary components: 4.89% acetic acid,
1.64% hydroxyacetone, 0.55% methanol, 1.62% formic acid, and 0.55% ethanol, but also 1.58% 1,2,3-
butanetriol and 1.09% 1-butanol as well as an astounding 10.33% succinic acid with 1.18% of a peak
identified as (but unlikely) sorbitol.

The first test with this feedstock was performed in the mini-reactor system using the aqueous
fraction diluted 50/50 with deionized water. This attempt to reduce the COD of the feedstock was
unsuccessful as some of the organic components phase-separated upon addition of more water. The
mixed phases were fed to the reactor but organic phase was susceptible to thermal reaction in the heat-up
zone of the reactor forming a polymeric solid that plugged the reactor system and terminated the test.

10
 The CT45-1 test was followed by three successful tests performed with aqueous bio-oil fractions.
A single bed of Ru/C catalyst was used throughout these three tests. Time on stream amounted to 137.5 h
total for the three tests. Up to 99.98% conversion of organic (measured by COD) was achieved. The
feedstocks were all about ~15% organic content (COD 160,200-178,250 ppm). LHSV from 2-4 L feed -
L-1 of catalyst - h-1 were evaluated. The test temperature was typically 350 °C, but 200 °C was also
tested. The operating pressure was within the range of 190-200 atm. The typical gas composition of
55%/40% CH4/CO2 was produced with 1-2% H2, 0-5% of higher hydrocarbon gases.

3.1.3 CS6-06 (4/26–29/2011)

For this test, the feedstock was produced at PNNL using a wiped-film distillation system. An
aqueous fraction was recovered from the lower density bio-oil fraction of the bio-oil feedstock (CS6-06)
produced at the National Renewable Energy Laboratory as part of CRADA #259x using corn stover
feedstock. The lower density fraction phase separated from the bio-oil after condensation and cooling.
The wiped-film distillation set point was 40ºC (pot temperature) with 70ºC film temperature and -1ºC in
the primary trap followed by an isopropanol/dry ice cold finger. The system was operated at 5 torr (0.1
psi) pressure. As this was a continuous-flow still, a 300 ml/h feed rate was used after the first hour of
devolatilizing. During the distillation, 32.5% was non-volatile, 60.4% of light distillate was recovered,
85% of which is water, and there was a loss of 7.1%. The analysis of the light distillate showed that it
had a very similar composition to that determined for the CT32-33 product described above. LC analysis
identified the primary components as: 7.10% acetic acid, 3.19% hydroxyacetone, 1.51% methanol, 0.38%
formic acid, and 0.61% ethanol with 0.39% ethylene glycol and 0.24% succinic acid. Lesser amounts of
other C3 to C5 alcohols and aldehydes were also noted. Chloride analysis detected 5 ppm Cl. ICP
analysis showed 11 ppm of S and 1 ppm Na. The labor requirement was high for producing the CHG
feedstock; 2/3 day distillation provided only 1/3 day of feed for the bench-scale reactor.

The aqueous fraction was tested for CHG in the mini reactor to allow a longer test. The test
extended over 72 hours including a period of operation at lower temperature, 200 ºC. High conversion
(99.93% of 177,600 ppm COD at 2.01LHSV 1) was attained at the typical operating temperature of
350 ºC. At the lower temperature, the conversion was greatly reduced with the COD being reduced only
5% at LHSV of 4. By increasing the temperature back to 350 ºC, the conversion returned to a high level
of 99.98% of a diluted feedstock (89,000 ppm COD) at the higher LHSV of 4.00.

3.1.4 CT45-2 and CT45-3 (5/23–24 & 6/7–8/2011)

The same catalyst bed was used in a subsequent test using the UOP CT45 feedstock. As shown
in Table 1 the activity of the catalyst remained high for over 110 hours, but some deactivation can be
seen in the reduced conversion measured over the last 20 plus hours on stream.

During the period of high conversion, the gas composition was consistently 54-62% methane with
36 - 45% CO2. The hydrogen and ethane concentrations each remained below 2%. As the activity level
fell toward the end of the experiment, the gas composition shifted to less methane (48% CO2 and 43%
CH4 at the end) and more hydrogen, ethane and higher hydrocarbons.

1
The unit for LHSV values mentioned in this report is L feed - L-1 of catalyst - h-1.

11
 Table 1. Results of CHG with aqueous bio-oil distillate

Test Time on stream Feed COD LHSV Product COD Conversion % g COD/hr
CT45-2 77-89.75 160,200 2.22 32 99.98 11.6
92-98.67 160,200 2.50 37 99.98 13.0
100-105.5 160,200 3.40 155 99.92 17.6
CT45-3 107.5-111.5 178,250 1.97 56 99.98 11.4
119.5-121.5 178,250 3.07 1,830 99.27 17.6
133.3-138.2 178,250 2.01 8,913 96.47 11.2

The catalyst was analyzed after recovery from the reactor. As seen in Table 2, the analysis of the
fresh catalyst showed, in addition to the Ru, some trace elements, such as K, Ca, Si, Na and some S with
Mn and Al. The feedstock analysis suggested that only minor amounts of Na, Al, Si, and perhaps S
would be introduced. However, after use the recovered catalyst contained a sizeable amount of Si and S,
as well as Cu and Na. There was actually a lower level of K and Ca than found in the fresh catalyst.
Most disturbingly, there was a large amount of Fe, Ni, Cr and even Mo—all elements found in the
stainless steel composing the reactor wall. The presence of Cu is curious and unexplainable; it was found
at such concentration on some of the catalyst extrudates (0.63wt% of total recovered catalyst) that they
had a reddish tinge. The S is a key catalyst poison of concern when using Ru. The other elements could
act as catalyst site fouling agents, causing deactivation as well, especially in combination with suspected
carbon deposition. Nitrogen physisorption tests of the fresh and spent catalysts showed significant loss in
pore volume and surface area (complete loss of the micropores) over the period of use, as presented in
Table 3.

Table 2. ICP analysis of the components in the CHG tests with aqueous bio-oil distillate

element feed fresh catalyst used catalyst

Ru <2 91465 65460
K <2 1571 275
Ca <2 523 47
Si 5 259 2796
S 1 143 1993
Na 9 96 150
Al 9 42 <50
Mn <2 31 <50
Cu <2 <25 613
Fe <2 <25 2754
Ni <2 <25 2308
Cr <2 <25 335
Mo <2 <25 84

12
 Surface area, porosity, and pore size analyses were determined using nitrogen
adsorption/desorption collected with a QUANTACHROME AUTOSORB 6-B gas sorption system. The
samples were degassed at 150 °C for 8-16 h under vacuum. The degassed samples were analyzed by
nitrogen adsorption/desorption at a constant temperature 77.4K. The volume amount of N2 gas that
adsorbed/desorbed on/from the surface of sample vs. relative pressure was measured. The surface area
was determined from the isotherm using the 5 points BET method. The BJH method was used for the
porosity and pore size analyses.

Table 3. Physisorption analysis of the catalysts in the CHG tests with aqueous bio-oil distillate

measurement fresh catalyst used catalyst

2
BET surface area, m /g 767 226
2
t- method micro-pore surface area, m /g 166 0
BJH pore volume, cc/g 0.54 0.26

The results of these tests with bio-oil aqueous fraction suggest that the organic contaminants,
even at high concentration, can be effectively converted to a fuel gas product by CHG. However,
corrosion of the stainless steel reactor system by the acidic components appears to be significant, based
on the metal deposits on the catalyst. Sulfur stripping may also be necessary in order to maintain long-
term catalyst activity. Carbon fouling of the catalyst was also suggested by the loss of porosity and
surface area.

3.2 Mini-reactor tests with UOP-derived hydrotreating byproduct

feedstocks

Subsequent tests in the mini-reactor used as feedstock various aqueous products from the UOP
bio-oil hydrotreating tests. Details of the production of the aqueous byproducts were not provided by
UOP except for titles as used below. The biomass feedstock was not identified either. The first three
products expressed a range of product compositions and concentrations, as seen in Table 4. Based on
these results, it appears that a range of hydrotreating was undertaken such that the pyrolysis bio-oil
product was more or less hydrogenated. Ammonium analysis showed a low level (4.5ppm) in the
aqueous feed but a higher level (468ppm) in the fixed bed product.

Trace elements were quite low in the hydrotreating aqueous byproducts (shown in Table 5).
Traces of biomass minerals are present. Sulfur content ranged over an order of magnitude from 2 to 20
ppm. Further analysis of the “mixed” feedstock showed a density of 0.9806 with a viscosity of 10.69 at
40 °C. Ion chromatography of the feedstocks for anion analysis (chloride, sulfate, etc.) was unsuccessful
because of baseline displacement and interference apparently due to organic anions; however, sulfate was
not detectable.

13
Table 4. Chromatography Analysis of UOP-Provided Feedstocks
aqueous fixed bed Mixed
COD 166,467 ppm 55,500 ppm 87,048 ppm
NH4+ 6.5 ppm 468 ppm 651 ppm
HPLC
formic acid 0.04% 0.005% 0.153%
acetic acid 6.57% 0.098% 0.438%
propionic acid 0.33% 0.004% ND
glycolic acid 0.21% NA 1.111%
methanol 0.45% 0.012% ND
ethanol 0.03% 0.383% 0.051%
propanol ND 0.055% ND
isopropanol 0.06% 0.127% 0.011%
butanol 0.25% 0.015% ND
pentanol 0.16% 0.072% 0.382%
sorbitol 0.07% NA 0.285%
xylose ND NA 0.122%
glucose ND NA 0.088%
methyl-acetate 0.10 ND ND

GCMS
phenol present ND NA
o-cresol present present NA
p-cresol present present NA

Table 5. Trace elements by ICP-OES

element aqueous fixed bed mixed
sulfur 2.2 5.8 19.3
silicon 81.9 52.7 14.9
iron 26.9 <0.1 <0.8
sodium 6.9 6.0 6.8
aluminum 5.9 <0.1 1.3
molybdenum <0.1 5.6 <0.8
calcium 1.1 0.9 1.5
potassium 3.9 <0.5 3.8
tungsten 1.3 <0.1 <0.8
nickel 1.1 <0.1 <0.8
phosphorus <0.1 1.0 8.8
zinc 0.3 <0.1 <0.8
magnesium 0.1 <0.1 <0.8

14
3.2.1 UOP1 aqueous bio-oil upgrading (9/21–22/2011)
The first hydrotreater aqueous byproduct provided by UOP was a higher concentration organic in
water liquid. This product was subsequently identified as produced in a continuously stirred tank reactor
(CSTR) hydrotreater system. The feedstock was prepared by filtration of a trace amount of floating scum
(0.4 g from 2200 g filtrate). There remained a floating layer of oil after the filtration. The float oil was
not sent to the gasifier, but would be returned to the hydrotreated oil product stream in the actual process.

The mini-reactor system was used for the test at typical processing conditions of 350 ºC, 3000
psig with a 29-ml catalyst bed and was kept on stream for 31 h until all the liquid was processed.

In the test, the COD was reduced from 160,750 ppm by 99.9% producing a 265 ppm COD
effluent at a LHSV rate of 2.2. The gas yield was 0.77 L/g organic (estimated from HPLC results) of a
medium-Btu gas, 611 Btu/SCF. There was no difficulty in the operation of the test. Flow and product
composition were consistent throughout. The chloride level in the effluent varied throughout the test
from 66 ppm to 656 ppm and back to 21 ppm. No explanation for this variation is evident. Analysis by
HPLC suggested that the remaining organic in the effluent was glucose and xylitol, but these
identifications were not confirmed. Although the total sulfur in the feed was measured at 2.7 ppm, the
last product out of the test measured at 9.5 ppm. Sulfate was so low that is was not quantifiable.
Similarly, Ru was below the level of detection <0.8 ppm.

3.2.2 UOP2 fixed bed HT byproduct aqueous (10/13/2011)

The second test using the fixed bed HT aqueous as the feedstock followed after the first using the
same catalyst bed. After three weeks in cold idle mode, the catalyst bed was reduced under hydrogen
flow while heating through the temperature range from 50 to 350 °C over 5 h, before idling at temperature
for 17 h (overnight). For the test, typical conditions were used – 350 ºC, 3000 psig with the 29-ml catalyst
bed. The test extended for 14.2 h, during which the COD was reduced from 59,500 ppm by 99.6% to
produce a 287 ppm COD effluent. At a LHSV of 4.3, 1.18 L/g organic of a medium-Btu gas (691
Btu/SCF) was produced. There was no difficulty in the operation of the test. Flow and product
composition were consistent throughout. The chloride level in the feed was determined to be 138 ppm
with no detectable sulfate. The effluent chloride level varied throughout the test from 249 ppm to 436
ppm. No explanation for this high level or variation is evident. The sulfate level remained <1 ppm
throughout. Analysis by HPLC suggested that the remaining organic in the effluent was glucose and
xylitol with some isopropanol, but these identifications were not confirmed. Although the total sulfur in
the feed was measured at 5.8 ppm, the last product out of this test measured only 0.5 ppm. Sulfate was
again so low that is was not quantifiable. Similarly, Ru was below the level of detection <0.8 ppm.

3.2.3 UOP3 mixed HT byproduct aqueous (12/14–15/2011 &1/3–4/2012)

The “mixed” hydrotreater aqueous byproduct provided by UOP was an intermediate
concentration organic in water liquid. The feedstock was prepared by separating a floating oil (160.23 g
from 5962 g aqueous feed). The float oil was not sent to the gasifier, but would be returned to the
hydrotreated oil product stream in the actual process.

The third test using the mixed HT aqueous as the feedstock followed after the first two using the
same catalyst bed. After 2 months in cold idle mode, the catalyst bed was reduced under hydrogen flow

15
while heating through the temperature range from 50 to 350 °C over 2 h, before idling at temperature for
12 h (overnight). For the test, typical conditions were used 350 ºC, 3000 psig with the 29 ml of catalyst.
The test extended for 13.8 h of steady conditions. During the steady state, the COD was reduced from
87,050 ppm by 99.9% to a 15 ppm COD effluent. Using a higher LHSV of 3.5, 0.58 L/g organic of a
medium-Btu gas (557 Btu/SCF) was produced. Afterward (during the night) a process upset occurred and
the system shutdown with an over-pressure alarm. The system was flushed with water for 5 days to clear
blockage and clear feed contamination from the product collection system. The catalyst bed was then
reduced under hydrogen flow at 280 °C for 4 h before returning to water flush mode at reaction
conditions for 2 weeks. Upon reintroduction of the feed to the reactor system, the conversion was found
to be less than 75%, so the test was terminated and the catalyst removed for analysis. Although the total
sulfur in the feed was measured at 19.3 ppm, the three products tested from the steady state portion at the
beginning and again at the end of this test measured <0.8 ppm. Sulfate was not discernable. The test
effluent had non-detectable (<0.8ppm) Ru.

3.2.4 Catalyst bed analysis following aqueous product CHG

Table 6 summarizes the elemental analysis of the feeds and the catalyst samples from these three tests.
Although the sulfur content of the feedstocks is quite low, the catalyst is significantly sulfurized by the
exposure. The apparent strong attraction of the Ru metal to the S, apparently to form the sulfide is
evident considering the high concentration of the S at the front end (bottom) of the catalyst bed. Ni, Mo,
and Cr are likely derived from construction material of the reactor system while the Fe could suggest a
similar mechanism, but also could result from the aqueous bio-oil upgrading (CSTR) feedstock, which
carried 26 ppm Fe, probably from corrosion during storage in steel cans. Al and P deposits apparently
come from the feedstocks, while the Cu source is still unknown. The elements Na, K, Si, and Ca, which
are trace components in the carbon support, appear to be stripped from the catalyst by the hydrothermal
process. Although the catalyst bed samples display a reduced level of Ru, this result is believed to be due
to carbon deposits in the pores of the catalyst, thus diluting the catalyst metal and potentially blocking
catalyst sites. This conclusion is supported by the finding of no Ru leaving the system in the aqueous
products.

Table 6. Analytical results from ICP-OES

Ru P S Fe Cu K Na Si Al Ca Ni Mo Cr
%
Ppm
Feeds 0 0-9 3-19 27-0 0 0-4 6-9 82-15 6-1 1 0-0.6 0-6 0

Fresh 9.1 <25 143 NA <25 1571 96 259 42 523 <25 <25 <25
catalyst
Spent, 5.0 166 4500 348 710 82 <40 178 175 117 2597 119 405
bottom
Spent, 6.3 449 3051 1205 180 82 <40 168 268 227 1372 362 256
middle
Spent, 7.7 148 108 2137 112 83 <40 147 163 45 1061 109 38
top

16
 At the end of the “Mixed” feed test, deposits were recovered from several places in the feed
system. The deposit in the preheater line was predominately Fe, Ni, and Cr with Al, P, S and Si as the
other major components. These elements account for only 1% of the mass, suggesting that the bulk of the
plug is carbonaceous polymer. Deposits further in the system contained some amounts of Ru, suggesting
that catalyst fines may have made their way into the plugs. Higher levels of S, Ni and even Cu were also
found in these deposits.

3.2.5 UOP4 HT (3/14–15/2012)

The results with the initial feedstocks containing only low levels of sulfur showed that system
modification to accomplish sulfur scrubbing from the feed would be required to protect the CHG catalyst.
The feedstock used in this test was from the same source as the previous test, “mixed” aqueous (Table 4
and 5). However, it phase separated after refrigerated storage for a month and a half. Also, the aqueous
was filtered as visible solids were noted. The filtered solids amounted to only 0.0036 wt% of the
aqueous. IC analysis of the feed (6 months after UOP3 test and 3 months after the UOP4 test) suggested
it contained 40 ppm chloride, but 29 ppm sulfate. This measurement of high sulfate in the stored
aqueous byproduct suggests that the initial sulfide in the feed has oxidized to sulfate over time. ICP
analysis shows 25 ppm sulfur, therefore about 40% of the sulfur had been oxidized to sulfate. In the case
of P, the ICP analysis showed 13 ppm. These ICP numbers are slightly higher compared to 19 and 9 ppm
of S and P, respectively, measured earlier (see Table 5). Other elements found in this analysis were 28
ppm Na, 22 ppm Si, 3.5 ppm K, 1.9 ppm Al, and 1.8 ppm Ca. Clearly there was not enough Ca to
precipitate all the sulfate in the solids settler, allowing some sulfates to affect the catalyst activity
over time. GCMS analysis of the feed showed the typical pyrolysis products of acetic acid,
hydroxyacetone, and levoglucosan, as well as a collection of phenol and alkyl-phenols.

For this test, a sulfur scrubbing bed of a nickel containing formulation (BASF G1-80 with
0.1%Cu) was placed in the system upstream of the Ru/C catalyst bed. Start-up of the system included a
28 h period of water flow during which a noticeable, amount of catalyst powder was flushed from the
system. This powder represented an insignificant mass loss from the catalyst bed, but suggested that a
water wash of the catalyst before filling the reactor could eliminate some downstream plugging problems.

The feed was introduced into the system and the catalyst initially exhibited high conversion at
LHSV of 3, which within 12 h began to fail and within 17 h had dropped from nearly 100% conversion to
less than 65%, with a similar drop in gas production rate. The feed COD was reduced from 76,500 ppm
to around 10 ppm initially, and then to 26,350 ppm later. GCMS was used to identify some residual
organic material, primarily phenol and alkyl-phenolics, but also some light carboxylic acids and cyclic
ketones, as well as levoglucosan. In the gas product, GCMS was used to identify some residual organic
material, primarily benzene and toluene with lesser amounts of alkyl-benzenes and small amounts of
phenol and alkyl-phenolics, with some light alkanes.

IC analysis of the aqueous products from the CHG showed they contained the expected level of
chloride, from 32 to 44 ppm. The sulfate had been reduced in the process to levels varying from 1 to 7
ppm. However, after filtering the aqueous products, ICP analysis showed <1 ppm S in the clear water
sample. There was no evidence of catalyst metal leaching as the Ru measurement was below the
detection level, <0.18 ppm. Analysis of the filtered solids showed that they were composed of catalyst

17
dust (Ru) as well as sulfur scrubber particulates (Ni, Si, Mg), corrosion products from the reactor system
(Fe, Cr, Mn), and Ca, Na, K, Al, and P from the feedstock. Some of this material may also be
precipitated sulfate and phosphate compounds.

Analysis of the catalyst and scrubber beds after the test confirmed that a significant amount of
sulfur was captured in the scrubber (about ¼) but over 10% ended up on the catalyst while another quarter
passed through in the aqueous stream. The sulfur on the catalyst and in the aqueous apparently is the
sulfate that was not captured by the Ni scrubber. Some of the balance of the sulfur precipitated at
hydrothermal conditions but was not collected in the separator vessel, but was mobilized by the liquid
water flow and passed on through the system to the condensate product collector.

The detailed analysis showed that the Ni in the scrubber was only slightly (<1%) reacted to nickel
sulfide. It did not serve as an effective sulfur scrubber. However, the fact that a large portion of the S
was in the form of sulfate, and therefore less reactive, is at least a partial explanation. The sulfur was
evenly distributed through the Ni bed suggesting that it was reacting slowly. However, the Ru was a
much more effective trap for the S, the S being over five times higher in concentration at the front of the
bed (862 ppm) versus the back of the bed (157 ppm), with the middle portion of the bed having an
intermediate level of S, while the fresh catalyst had <50 ppm. The high level of sulfiding of the Ru/C
catalyst at the front end of the bed, about 5% of stoichiometric, compares to a level of 40% sulfidation
reported in the literature as the maximum obtained in a hydrothermal environment. Therefore, the Ru
catalyst had been sulfided to a large degree and certainly enough to have a large effect on its gasification
and methanation activity. EDS analysis showed that the sulfur was highly associated with both the Ni
and the Ru. There was also evidence of Ca and P (probably as calcium phosphate) coating the surface of
the Ni scrubber. The Ru concentration in the spent catalyst had been reduced to 3/4th of the starting level,
presumably by carbon deposition in the pores, as there was no evidence of Ru loss to the aqueous stream.
In the case of the Ni scrubber, the pore volume was essentially unchanged but the surface area was
reduced to 193 m2/g compared to a literature value of 272 m2/g.

3.2.6 UOP 4B HT (8/7/2012)

A second attempt with this feedstock was made with a modified preheater/separator design to
allow for better capture of precipitated minerals especially as Ca was added to the feedstock as Ca(OH)2
at 0.3 g/L. The same sulfur scrubber material and Ru/C catalyst were used in the test at the same
processing conditions: 350 °C, 3000 psig, and 3 LHSV. The system became plugged in less than 6 h of
operation. Analysis of the recovered solids showed that the Ca/S ratio of 1.1 was more than sufficient to
suggest calcium sulfate precipitation. With the inclusion of P in the calculation, assuming a 1.5 ratio of
Ca/P, then the Ca/S ratio drops to 1, providing further confirmation of calcium phosphate and calcium
sulfate precipitation. However, both Ca and S were also found in the sulfur scrubber bed while Ca, S and
P were all found in the Ru/C bed, suggesting that the precipitator system was not fully functional in the
mini-reactor system.

3.2.7 UOP 4C HT (9/6–7/2012)

A third attempt with this feedstock was made with the modified preheater/separator design to
allow for better capture of precipitated minerals especially as Ca was added to the feedstock as Ca(OH)2
at 0.3 g/L. The same S scrubber material and Ru/C catalyst were used in the test at the same processing

18
conditions: 350 °C, 2950 psig, and 2.5 LHSV. The system remained unplugged during the total time on
stream of 19.8 h of operation until the feedstock was consumed. The conversion remained high
throughout with no evidence of catalyst deactivation. COD conversion was 99.99% producing a product
gas of 70% methane, 28.5% carbon dioxide, and less than 1% hydrogen. The feedstock COD was 79,500
ppm COD and the effluent was about 13 ppm.

After the test, a mineral plug was found in the feed line as it passed through the preheating zone.
The mineral composition suggested calcium sulfate and phosphate but the ratios were not quite the perfect
match as there was only enough calcium to account for ¾ of phosphate after subtracting the requirement
for sulfate. Analysis of the recovered solids showed that the Ca/S ratio of 1.8 was not quite sufficient to
suggest total calcium sulfate precipitation because with the inclusion of P in the calculation, the Ca/P was
only 1.1. Still, both Ca and S were also found in the sulfur scrubber bed and in the Ru/C bed, suggesting
that the precipitator system was not fully functional for sulfate capture while capturing all the P.

3.2.8 UOP5 HT (11/30–12/1/2012)

The results with the initial feedstocks containing only low levels of sulfur showed that system
modification to accomplish sulfur scrubbing from the feed would be required to protect the CHG catalyst.
The feedstock used in this test was a new shipment from UOP labeled, “Aqueous product from plant 27.”
This feed labeled as IBR-UOP5 was received on July 25, 2012. It was a mixture of oil and aqueous phase
products. A settling experiment was done overnight showing that most of the oil settled at the bottom of
the container.

Separation of the oil phase from the aqueous phase was done by (1) centrifugation and (2)
filtration. Centrifugation settled the oil much faster. Oil layered the surface of the centrifuge tubes.
However, since the oil seemed to have similar viscosity as that of water, decanting the top layer became
difficult once the oil was disturbed. The oil amount in the IBR-UOP5 sample seemed to be between 6-11
wt%.

Vacuum filtration through regular filter paper or glass fiber worked for previous UOP aqueous
samples in removing the more viscous oil, but not for this one. Oil breakthroughs occurred. Gravity
filtration with regular filters allowed better separation, but took a long time. Kaowool (fibrous insulation)
was then used as a filter medium and seemed to work well in separating the bulk oil. The thickness and
the large surface area of the kaowool allowed water to flow freely while trapping almost all of the oil. A
few small droplets got through and breakthrough still occurred after over a liter of sample processed and
so the kaowool did need to be replaced. The oil could be potentially recovered by washing the kaowool
with the proper solvent.

Leaching of materials from the kaowool was determined by running ICP on the product recovered
from filtering regular DI water. The result is shown in Table 7 suggested that there were minor amounts
of Na and Si, but a surprising amount of S.

19
 Table 7. Analytical results from ICP-OES analysis of UOP5 aqueous streams
Ca S Si K Na Fe Mg Al
Feed, as-received, 5.2 24 12 10 9.4 1.5 1.5 <0.9
settled
DI water through <0.8 14 2 <1.5 3 <0.8 <0.8 <0.8
Kaowool filter
settled, filtered feed to 46 26 11 11 8.0 2.0 1.2 <0.8
reactor UOP5A
UOP5A early sample <0.8 <0.8 26 14 17 <0.8 <0.8 <0.8

UOP5A end sample <0.8 <0.8 22 11 7.9 <0.8 <0.8 <0.8

settled, filtered feed to 49 21 10 10 10 2.6 1.4 0.8

reactor UOP5B
UOP5B early sample <0.8 <0.8 6.7 2.0 4.7 <0.8 <0.8 <0.8

UOP5B end sample <0.8 <0.8 6.1 2.4 4.9 <0.8 <0.8 <0.8

LC analysis of the feed showed the typical pyrolysis product, acetic acid, with indications of
ethanol, succinic acid, phenol, acetaldehyde, propylene glycol, and sorbitol with several unidentified
components at higher concentration. The TOC of the filtered feedstock was measured as 1.54 % for 5A
and 1.35 % for 5B.

IC analysis of the as-received, settled feed suggested it contained 28 ppm chloride, but 17 ppm
sulfate. This measurement of significant sulfate in the accumulated aqueous byproduct suggests
that the initial sulfide in the feed may have been oxidized to sulfate over time. ICP analysis showed
24 ppm sulfur, therefore about 35% of the sulfur was in the oxidized form of sulfate. In the case of P, the
ICP analysis showed <0.9 ppm. Other elements found in this analysis were 9.4 ppm Na, 12 ppm Si, 10
ppm K, 1.5 ppm Fe, with 5.2 ppm Ca and 1.5 ppm Mg. Again, there was not enough of the alkaline
earths to effectively precipitate all the sulfate in the solids settler, which would impact on the
catalyst activity over time. Before feeding into the CHG system, about 0.084 g/L of Ca(OH)2 was added
into the filtered IBR-UOP5 with the objective of precipitating out SO4-2 before it could poison the
catalyst.

For test UOP5, as in UOP4, a sulfur scrubbing bed of a nickel-containing formulation (BASF
G1-80 with 0.1% Cu) was placed in the system upstream of the Ru/C catalyst bed. Start-up of the system
for UOP5A included a 4 h reduction of the catalyst at 350 °C under hydrogen flow followed by cooldown
over the weekend. A 10 h period of water flow for heat-up of the reactor system preceded actual
feedstock introduction. After 14.6 of operation the system was determined to be at steady state. The test
continued for an additional 23 h until the feed sample was exhausted. The catalyst exhibited high
conversion at LHSV of 2.5 throughout, with COD conversion at four 9s (99.99%). The feed COD was
reduced from 53,500 ppm to around 10. The gas product was consistently around 60% CH4 and 40% CO2
with only 0.4% H2 and undetectable hydrocarbon gas products.

20
 As shown in Table 7, ICP analysis showed <1 ppm of either S, Ca, Mg, or Fe in the aqueous
product samples. Soluble elements included K, Na, and Si. There was no evidence of catalyst metal
leaching as the Ru measurement was below the detection level, <0.8 ppm.

The test was restarted nine days later as UOP 5B using the same scrubber and catalyst beds with a
second aliquot of the “Aqueous product from plant 27” which had been settled, filtered and doped with
Ca, similar to UOP 5. The system was heated up over 3.3 h and then feed was introduced when the bed
was at temperature. The catalyst continued to show high activity with COD conversion at LHSV of 3 at
99.98% over the relatively short 6.8 h test, which used up the balance of the feedstock. The feed COD
was reduced from 44,000 ppm to around 20. The gas product was consistently around 70% CH4 and 30%
CO2 with only 0.3% H2 and undetectable hydrocarbon gas products.

As shown in Table 7, ICP analysis again showed <1 ppm of either S, Ca, Mg, or Fe in the
aqueous product samples as was seen in UOP5A. Soluble elements included K, Na, and Si, although at
lower levels than seen in UOP5A. There was no evidence of catalyst metal leaching as the Ru
measurement was below the detection level, <0.8 ppm.

ICP analysis was performed on the solid residue collected from the settler of the UOP5A/B run.
The residue was recovered by scraping off the inner surface of the ¾” tubing. The major components are
Ca (33130 ppm), Fe (28940 ppm), S (28990 ppm), Ni (20580 ppm) and Ru (15460 ppm). The identified
metals were about 13.5wt % of the sample analyzed, suggesting that most of it was coke/organic. Ca and
S were not strictly stoichiometrically present for the formation of CaSO4 (Ca/S = 0.91, when it should
have been 1). The lesser components were Al (2548 ppm), P (1887 ppm), Cr (1650 ppm), Si (1347 ppm),
Mg (593 ppm) Na (321 ppm), and Mo (244 ppm). The presence of Ni and Ru, Si and Al suggests that
there were catalyst bits present in the settler. The extra sulfur may have been associated with the
catalysts. Since the settler comes before the reactor, the presence of the catalyst particles was interesting.
Two options could explain what happened: (1) catalyst particles were blown back during the ex situ leak
check in the hood prior to the test, or (2) backflow in the instance that the feed pressure became less than
the reactor pressure (in the presence of a stopped pump). The first explanation was avoided in subsequent
tests by pressurizing in the same direction of flow as in the actual operation. Preliminary H2 leak check
was done in the hood for better access to the myriad of fittings of the modified micro unit. The counter
flow of the test gas was done because of convenience and limitation based on the configuration of the
reactor with respect to the hood. P was also present in the residue, suggesting that the Ca was able to
capture some as Ca3(PO4)2. However, this result would further lower the Ca/S ratio. One explanation
would be precipitation of some of the sulfate by iron, perhaps through a reactor wall corrosion related
mechanism.
Analysis was also performed by XRD on the S scrubber material and the gasification catalyst
following the test. The composition detected for the spent material in the S scrubber bed did not include
sulfide as would be expected if the scrubber were functioning to capture the 21-26 ppm S in the
feedstock. The Ru catalyst bed included graphite and Ru metal phases with the Ru crystal size calculated
at 21 nm, but again no sulfide was detected. Although the S was reduced to an undetectable level by
passing through the reactor system, the fate could not be determined based on these analyses. CaCO3 was
detected in both the scrubber bed and the gasification catalyst. This suggests that there was excess
calcium, which did not precipitate as sulfate or phosphate, and which precipitated when the carbon
dioxide formation became sufficient.

21
3.2.9 UOP6 HT (2/21/2013 – 2/23/2013)

Feed Pilot plant aqueous phase water received from UOP (1/17/2013), vacuum
filtered through GF-B, COD = 185,000 ppm, pH = 3.60
Reactor configuration sulfur scrubber bed followed by gasification catalyst bed
Total run time 34.54 h
Reason for EOR feed was all gone
Catalysts Main: C3610 – 24.7 ml; 10.87 g
Sulfur scrubber: Raney Ni (Raney 5886 3x8 catalyst, Nickel Cat 19400-72) –
22.45 g (wet); ~ 8ml
Flowrates 1st LHSV = 2/h (25 h)
2nd LHSV = 2.5/h (9.5 h) (from 14:32 2/22/13 to EOR)
Product COD ranges Port: 0-20 ppm
Pot: 0-10 ppm
Gas products CO2 range (time): 24.4% – 38.8% (5th to 34.5th h)
CH4 range (time): 65.5% - 76.5% (5th to 34.5th h)
H2 range (time): 0.5% - 0.7% (5th to 34.5th h)
Ethane range (time): 0.1% to 0.5% (at least last 8 h)

The feedstock sample was received from UOP on 1/18/13 labeled as aqueous phase pyrolysis oil.
The sample out of the jar showed an oil phase settling at the bottom (Figure 6a) and a supernatant that
became clearer after standing for a few days inside the refrigerator (Figure 6b). After vacuum filtering
through a Whatman GF-B filter, the liquid that came out was quite clear. However, after a few days of
standing on the lab bench, the filtrate turned greenish with some oil precipitating at the bottom (Figure
6c). COD of the as-received filtrate was 187,500 ppm while the COD of the filtered sample was 185,000
ppm. The difference may be due to filtering out of some organics or simply variation in measurement. As
received, the pH was 3.6 and the TOC was 4.35 %.

a b c

Figure 6. Feed showing segregation of oil and aqueous phase. Supernatant became clearer after
days of standing in the refrigerator

22
 ICP of the liquid showed the presence of the following elements: 18 ppm Na, 6.5 ppm K, 3.2 ppm
Fe, 1.7 ppm Si, and 10 ppm S. Based on the small amount of S, no Ca(OH)2 addition was done, as was
done in UOP4 and 5 to precipitate out anions such SO4-2 and PO4-3.
From the filtration procedure, the
recovered oil was about 11.5 wt% of the
filtered sample.

The reactor assembly used in this

experiment is shown in Figure 7. This
was same as the reactor used in IBR-
UOP5b except for the replacement of the
bulkhead union fitting (encircled portion
in Figure 4) that allows for easier
dismantling and leak-proofing of the ¾”
fittings, as compared to a regular union.

During reduction, the top catalyst

bed was reduced at 350 °C for 4 h. The
bottom reactor though was only reduced
at 320 °C due to overheating of the
reactor wall temperature (>400 °C) at
higher set bottom temperature. This
problem was alleviated by improved
insulation during the actual run,
attributing the need for the heater to
produce more heat to compensate for the
heat loss to attain the target temperature.
Figure 7. Reactor assembly A slower and graded approach (~5 °C
higher than current temperature reading)
to keying in the set point to the bottom heating tape was also seen to be advantageous. During the actual
gasification experiment, a highest of 40 °C differential between the bottom and the reactor wall
thermocouple temperatures was seen.

Exotherms in both the top and bottom beds seemed to be present at the early stages of the
experiment. Temperatures at the top thermocouple (about 4-5” into the bed) peaked at 2 h after start of
feed at 362 °C, while the reactor wall temperature (outside the reactor but about 2” into the bottom reactor
bed) peaked after 3 h at 391 °C. However, the top temperature evened out at around 350 °C after about 18
h on stream. At the same time, the reactor wall temperature was at 380°C. The bottom thermocouple
readings were between 349.6 °C to 350.3 °C.

Figure 8 summarizes the COD readings from the sample port (“port”) and the liquid coming
out of the spout going to the receiving vessel (“pot”). COD reduction was always 99.9% throughout the
course of the run.

23
 IBR-UOP6 COD
25
LHSV = 2/h LHSV =
20
2.5/h

15
COD, ppm

COD, port
10 COD, pot

0
2/21/13 12:00 2/22/13 0:00 2/22/13 12:00 2/23/13 0:00

Figure 8. COD values of the experiment

IBR-UOP6 CO2 and CH4 gas data

100
90 LHSV = 2/h LHSV =
2.5/h
80
70
60
vol %

50 CO2
40 CH4
30
20
10
0
2/21/13 12:00 2/22/13 0:00 2/22/13 12:00 2/23/13 0:00

Figure 9. CO2 and CH4 distribution during the run

The major gas products were CO2 and CH4. Their concentrations are plotted as a function of time
in Figure 9. Steady-state seemed to have been attained after about 5 h of time-on-stream with an average
ratio of 30:70 of CO2 and CH4 respectively. During LHSV = 2, no ethane was found. After the switch to a
higher flowrate of 2.5, ethane was seen to be 0.1 to 0.5% of the gas stream. However, the COD did not
show much accompanying change.

24
3.2.10 UOP7 HT (3/13/13 – 3/15/13)
Feed Aqueous phase from pyrolysis oil hydrotreating (bench scale) from UOP
(1/11/13), filtered through regular filter paper using positive pressure filter,
COD = 48,600 ppm, pH = 7.13 as received
Reactor configuration sulfur scrubber bed followed by gasification catalyst bed
Total run time 33.5 h
Reason for EOR Lack of feed, product quality went down (much higher COD)
Catalysts Same as IBR-UOP6. (follow-on experiment with re-reduction in between)
Main: C3610 – 24.7 ml; 10.87 g
Sulfur scrubber: Raney Ni (Raney 5886 3x8 catalyst, Nickel Cat 19400-72) –
22.45 g (wet); ~ 8ml
Flowrates 1st LHSV = 5 (22.8 h)
2nd LHSV = 1 (10.7 h) (13:20 3/14/13 to EOR)
Product COD ranges Port: 14 – 0 ppm (first 20 h); up to 18050 ppm
Pot: 353 – 44 ppm (first 20 h); up to > 17600 ppm
Gas products CO2 range (time): 25.2-27.6% (9th to 20th h); 54.2–39.4% (20th h to EOR)
CH4 range (time): 68.2%-70.5% (9th to 20th h); 43.7-55.6% (20th h to EOR)
Ethane range (time): 0.1% to 1.66% (5th h to EOR)

The sample used for feedstock in this test

was received from UOP on 1/11/13 labeled as
“aqueous phase from pyrolysis oil.”
Communication with Steve Lupton at UOP
clarified that this sample came from their bench
scale plant. The sample out of the jar showed an
oil phase that segregated and floated on top of an
opaque orange liquid with a small amount of
heavy oils settling at the bottom (see Figure 10).
Most of the floating oil was filtered out through a
positive pressure filter apparatus, though some
small particles still went through with the filtrate.
COD from the as-received filtrate was 45,833
ppm. As received, the pH was 7.13. Residual Figure
Figure 10.
10. Feed
Feed showing
showing segregation
segregation of
of oil
oil and
and
feed in the feed container at the end of the run aqueous phase.
was at 48,000 ppm while a small amount
recovered from the pump was at 140,000 ppm. This very much higher concentration at the end of the run
suggests the possibility of residual oils precipitating out from the filtrate while the feedstock was held in
the pump.

ICP of the liquid showed the presence of only small amounts of trace elements: 29 ppm Si, 14
ppm Na, 7.3 ppm K, and 2.2 ppm Al. The critical element sulfur was found at 80 ppm. By IC analysis it
was determined that 54 ppm of sulfate were present, representing 22% of the sulfur. Chloride was found
at 39 ppm and the amounts of other trace anions were low: 2.4 ppm PO4, 1.4 ppm NO3, with peaks
tentatively identified as Br at 2.4 ppm and F at 3.3 ppm.

25
 Liquid chromatography of the feed was also done. The more polar components were resolved,
identified and quantified. Major components were acetic acid (0.198 wt%), ethanol (0.176 wt%) and
phenol (0.101%). Some unidentified high eluting components were seen and are expected to be alkyl or
methoxyphenols. Minor components (i.e. less than 0.1 wt%) consist of glycolic acid, formic acid,
ethylene glycol, propylene glycol, propanoic acid, methanol, ethanol, acetone, 1-propanol, 2-methyl-1-
propanol and 1-butanol. TOC of this feedstock was 1.30 %.

The same reactor and catalysts bed as UOP6 was used for this experiment.

During reduction, the top catalyst bed was reduced at 350 °C for 4 h. The bottom reactor though
was only reduced at 305 °C. The corresponding bottom reactor wall temperature at this bottom reactor
temperature was 400 °C. During actual gasification run conditions, the top and bottom temperatures were
consistently around 350 °C with the bottom reactor wall temperature at an average of 30 °C hotter. The
pump stopped around 5 h into the run because of an erroneous high pressure reading due to a failing
pressure transducer. The top reactor temperature read 352 °C about 25 min after the feed flow was
restarted.

Figure 11 summarizes the COD readings from the sample port (“port”) and the liquid coming out
of the spout going to the receiving vessel (“pot”). After 20 h into the run, the COD conversion went
down from 99.9% to 76.9%, from which the COD of the product progressively went higher in the last 10
h of the experiment despite decreasing the flowrate to an equivalent LHSV of 1 from 5.

IBR-UOP7 COD
20000
18000 sample port
16000 pot
14000
COD, ppm

12000
10000
8000 LHSV = 5/h LHSV = 1/h
6000
4000
2000
0
0 10 20 30 40
Time-on-stream, h

Figure 11. COD values of the liquid products

26
 IBR-UOP7 CO2 and CH4 gas data
120
CO2 CH4
100
LHSV = 5/h LHSV = 1/h
vol % 80

60

40

20

0
0 10 20 30 40
Time-on-stream, h

Figure 12. CO2 and CH4 distribution during the run

The major gas products were CO2 and CH4. As seen in Figure 12, steady-state seemed to have
been attained after about 9 h of time-on-stream with an average ratio of 26:69 of CO2 and CH4,
respectively. However, corresponding to the increase in the product COD, the ratio became closer to
50:50 CO2:CH4. As the LHSV was adjusted to attempt to recover the higher level of COD conversion,
CO2 seemed to trend back downward while the CH4 amount inched higher. Due to lack of feed, this trend
could not be fully validated through longer time-on-stream.

One possible explanation for this apparent loss of catalyst activity is the poisoning effect of S on
the Ru/C catalyst. Though the Raney nickel acts as a sulfur trap, S breakthrough to the Ru catalyst still
occurred. As shown in Table 8, S was distributed throughout the scrubber and catalyst beds.

XRD analysis of the spent scrubber bed and catalyst bed did not identify any sulfur phase in
either bed. There was little noticeable change in either bed. There was an indication of aluminum
hydroxide (boehmite) in the sulfur scrubber, which appears to be a hydrothermal product of the residual
aluminum metal in the Raney nickel. Boehmite is the stable phase at hydrothermal conditions. There
was a tentative identification of a very minor amount of aluminum nitride throughout the catalyst bed and
a hint of calcium carbonate in the bottom (entry point) of the catalyst bed.

The likely cause of the sulfur poisoning of the catalyst bed despite the presence of a sulfur
scrubbing bed was the sulfate content in the feedstock. IC analysis of the feed showed 54 ppm sulfate
(effectively 23% of the total sulfur in the feedstock). In addition, small amounts of other anions were
identified, 39 ppm chloride and 2 ppm phosphate. The sulfate form of sulfur is not effectively captured in
the Raney nickel sulfur scrubbing bed. The reason for the presence of sulfate in the product of strongly
reducing conditions was not clear.

27
Table 8. Elemental analysis of used bed components, ppm

Elements Sulfur Scrubber Ru/C Catalyst

bottom top bottom middle top
Nickel 460150 436650 5145 4422 2635
Aluminum 394900 404750 2089 1795 1139
Ruthenium ND ND 43100 51665 48560
Sulfur 4679 4126 4260 4837 3076
Silicon 3245 2925 262 406 296
Iron 1119 953 147 379 463
Cobalt 370 419 350 210 143
Calcium 60 49 113 134 105
Potassium ND ND 120 111 107
Magnesium 218 220 ND ND ND
Sodium ND ND 91 91 100
Titanium ND ND 93 107 97
Manganese 34 40 ND ND ND
Molybdenum ND ND 66 ND ND

3.2.11 UOP8 HT (3/28/2013 – 3/30/2013)

Feed Pilot Plant 581 aqueous phase (1/11/13) from UOP, decanted, unfiltered,
COD = 371,333 ppm as received, pH = 3.70 as received, 1400 ppm Fe
Reactor configuration sulfur scrubber bed followed by gasification catalyst bed
Total run time 39 h
Reason for EOR Over-pressure, plug formation
Catalysts Main: C3610 – 26.25 ml; 11.55 g
Sulfur scrubber: Raney Ni (5886 3x8 catalyst, 19400-72) 25.52 g (wet) ~
8ml
Flowrates LHSV = 2.5/h
Product COD ranges Port: 0 – 35 ppm
Pot: 0 – 9 ppm
Gas products CO2:CH4 30:70

Sample was received from UOP on

1/11/13 labeled as Pilot Plant 581.
Communication with Steve Lupton at UOP
detailed that the pilot plant was equipped with an
on-line hot separator that was very good at
separating the oil phase from the aqueous phase.
The catalyst used to produce this aqueous product
was determined by UOP to be less active than
expected. The liquid, though colored, seemed to
be free of oil (Figure 13). It was not filtered
before being processed. As received, the COD

Figure 13. IBR-UOP8 Feed.

28 No observable amount of oil found.
was recorded at 371,333 ppm while the pH was at 3.70. Among the last three experimental feeds (i.e.
IBR-UOP 6-8), this sample had the highest COD value. ICP of the liquid showed a high concentration of
Fe (1408 ppm) aside from the usual biomass elements of K (3 ppm), Mn (6 ppm), and Na (9 ppm). Fe is
presumed to be a contaminant from corrosion of the metallic container in which this liquid was collected.

Though S was not seen in the ICP (<0.8 ppm), a small amount of SO4-2 was reported at about 8
ppm (see Table 9). Considering the product quality degradation that occurred in the previous experiment
(IBR-UOP7), a small amount of Ca(OH)2 was added to the feed in an effort to capture the S as sulfate
precipitate.

Table 9. Anions identified in the feed through ion chromatography of the feed, as received
Anion Concentration, ppm
-
Cl 95
NO3- 18
SO42- 8
PO43- 0

Liquid chromatography of the feed was also done. The more polar components were resolved,
identified and quantified. Major components were summarized in Table 10. Identified minor components
(< 0.1 wt%) include propylene glycol, valeraldehyde and phenol. Unlike the aqueous feed produced from
the “bench” process (IBR-UOP7), the majority of the organic in this aqueous phase were alcohols rather
than the more recalcitrant phenolics. Based on this LC analysis a total organic carbon (TOC) can be
calculated as 9.94 %, while the measured TOC for this aqueous product was 8.06 %.

The same reactor configuration was used as with IBR-UOP6 and IBR-UOP7. However, a fresh
sulfur scrubber bed and catalyst bed were used. Ca(OH)2 was added at 20.5 ppm.

Table 10. Main compounds identified and quantified through liquid chromatography of the feed (as
received)
Compound Amount, wt%
Acetic Acid 2.3
Propanoic Acid 0.4
Methanol 0.9
1,2-butanediol 0.2
Ethanol 11.0
acetone 0.8
methyl acetate 0.4
1-propanol 1.4
Ethyl Acetate 0.9
2-Butanol 0.4
2-methyl-1-propanol 0.3
1-butanol 0.3

During reduction, the catalyst bed was reduced at 350 °C for 4 h. The bottom portion of the
reactor containing the sulfur scrubber was only reduced at 315 °C. The corresponding bottom reactor
wall temperature at this bottom reactor temperature was less than 400°C. During actual gasification run

29
conditions, the top reactor temperatures were higher, reaching as high as 387 °C and the temperature was
less stable.

Figure 14 summarizes the COD readings from the sample port (“port”) and the liquid coming out
of the spout going to the receiving vessel (“pot”). After 33.5 hours at LHSV of 2.5, COD conversion was
still at 99.9%. However, an overpressure occurred and caused the end of the run. A plug was probably
formed between the settling tank and the lower reactor bed. The solids recovered from the settler tank
were 68% iron with minimal amount of other elements, which suggests an iron oxide or hydroxide
precipitate. Spent catalysts were analyzed by ICP and XRD. Sulfur poisoning was found throughout the
beds at low level. The fact that it was not concentrated at the front of the nickel bed suggests that sulfate
was the form as it entered the bed and was reduced and reacted to metal sulfide as it passed through the
beds. Iron deposits were also found throughout the beds. Calcium manganese and sodium were also
found concentrated at the front end of the catalyst bed.

IBR-UOP8 COD
100
80
LHSV = 2.5/h
60
COD, ppm

40
20
0
-20 0 10 20 30 40
Time-on-stream, h

Figure 14. COD values of the product water

The major gas products were CO2 and CH4 (Figure 15). Steady-state seemed to have been
attained within 4 hours of time-on-stream with an average ratio of 30:70 of CO2 and CH4 respectively.

IBR-UOP8 CO2 and CH4 gas data

80 CO2
60 CH4
vol %

40

20

0
0 10 20 30 40
Time-on-stream, h

Figure 15. CO2 and CH4 distribution of the gas product

30
3.2.12 UOP9 HT (12/17/2013 – 12/19/2013)
Feed HT Upgrading aqueous phase (9/17/13) from UOP, decanted, unfiltered,
COD = 76,000 ppm as received, pH = 5.04 as received, 1750 ppm NH4
Reactor configuration sulfur scrubber bed followed by gasification catalyst bed
Total run time 46.5 h
Reason for EOR out of feed
Catalysts Main: C3610 – 11.72 g→27.9 mL
Sulfur scrubber: Raney Ni (Raney 5886 3x8 catalyst, Nickel Cat 19400-72) –
17.4 g (wet)
Flowrates LHSV = 2/h
Product COD ranges Port: 97 – 24 ppm
Gas products CO2:CH4 32:67

The feedstock sample was received from UOP on 9/19/13, labeled as Upgrader water. The liquid
had a small amount of precipitated oil on the bottom of the transfer vessel (Figure 16). It was not filtered
before being processed. As received, the COD was recorded at 76,000 ppm while the pH was at 5.04.
ICP of the liquid showed the usual biomass elements of Si (33 ppm), Na (10 ppm), K (7 ppm), and Fe
(2 ppm).

S was also present in the ICP (11 ppm)

and SO4-2 was reported at 56 ppm (see Table 11)
equivalent to 19 ppm S. The other anion
contaminants were present in low concentration.

Table 11. Anions identified in the feed through

ion chromatography of the feed, as received
Anion Concentration, ppm
SO42- 56
Cl- 44
NO3- 12
PO43- 3
Br- 2
Figure 16. Oil scum on container wall for IBR-UOP9
feed.

Gas chromatography of the feed was also done. Major components were summarized in Table
12. The majority of the organic in this aqueous phase were the more recalcitrant phenolics with alcohols
also present in significant amounts. The amounts shown in Table 11 are only semi-quantitative relative
amounts as a proportion of the volatile components.

The same reactor configuration was used as with IBR-UOP6, 7, and 8; however, a fresh S
scrubber bed and catalyst bed were used.

31
Table 12. Main compounds identified through gas chromatography-mass spectrometry of the feed
(as received)
Compound Amount, wt%
Ethanol 22.2
1-propanol 3.7
2-Butanol 1.0
Phenol 1.1
methyl phenol 48.3
dimethyl phenol 7.5
ethyl phenol 3.5
C3-phenol 0.4

Figure 17 summarizes the COD readings from the sample port and the gas flow rate over the 46
hours of test. The COD in the effluent dropped continuously through the test suggesting some initial
contamination in the sample line. After 46.5 hours at LHSV of 2, COD conversion was still at >99.9%.
The gas product flow increased quickly over the first 10 hours to the steady flow throughout the most of
the test. It remained high throughout with no indication of catalyst deactivation.

Figure 17. COD values of the product water and gas flow rate in UOP9 HT

The major gas products, as is typically seen, were CO2 and CH4 (Figure 18). Steady-state
seemed to have been attained within 6 h time-on-stream with an average ratio of 32:67 of CO2 and CH4
respectively. GC-MS analysis of the product water showed no evidence of organic contaminants and the
graph was essentially the same as that from a deionized water blank.

32
 Figure 18. Gas composition and flow rate in UOP9 HT

3.3 Mini-reactor tests with model compound mixtures

In order to perform long-term catalyst activity tests in the mini-reactor, a chemical model of the
hydrotreating aqueous product was produced in quantity from model compounds, because the aqueous
feedstock was not available from UOP in sufficient quantity. The model compounds were purchased
from Sigma-Aldrich at 99% purity or as high a purity as was available. The specific components were
selected based on the list of components found by UOP in their byproduct waters from hydrotreating bio-
oil in their first 35 runs at Des Plaines. The composition of three separate batches made for the tests is
shown in Table 13. Although sulfide was expected in the aqueous byproduct, it was not measured by
UOP and was not included in the simulant. Alkali metals and alkaline earths were believed by UOP to be
below 10 ppm and were not added to the simulant.

3.3.1 L1 (8/27/2013 – 9/6/2013)

Feed model compounds in DI water simulant
Reactor configuration no sulfur scrubber bed before gasification catalyst bed
Total run time 148 h (time on stream)
Reason for EOR out of feed
Catalyst C3610 – 21.14 g→50.3 mL (126g/300cc density measurement)
Flowrates LHSV = 2/h → 4/h → 5/h
Product COD ranges Port: 227 ppm – 16 ppm → 78 ppm – 6 ppm → 20 ppm – 7 ppm
Gas products CO2:CH4 30:68 → 31:67 → 30:67

33
Table 13. Composition of three batches of simulated bio-oil hydrotreating water product
Weight, g
organics Batch 8/27/13 Batch 9/3/13 Batch 9/5/13
Acetone 26.1 26.3 26.2
Ethanol 887.0 887.5 887.0
p-cresol 132.3 132.3 132.3
methanol 115.0 115.0 115.0
1-propanol 113.0 113.0 113.1
isopropanol 29.0 29.0 29.0
1-butanol 26.0 26.2 26.0
2-butanol 22.0 26.3 22.1
acetic acid 243.0 243.1 243.1
ethyl acetate 28.0 28.1 28.1
phenol 4.1 4.2 4.2
SUM 1625.5 1631.0 1626.1
water 10000.0 9397.5 10000.0
Factor (wt.O/[Link]) 0.14 0.15 0.14
COD, ppm measured 319,500 410,000 313,500
C 52.7% 52.8% 52.7%
H 11.5% 11.5% 11.5%
O 35.8% 35.7% 35.8%

The first simulant run proceeded over two weeks including a shutdown over the long holiday
weekend. In Figure 19 the data for COD in the effluent and the gas product flow rate are plotted over the
time of the experiment. The large blank area in the middle is the weekend break wherein the reactor was
shutdown to idle at room temperature with no feedstock flow. After the weekend the reactor was heated

Figure 19. COD values of the product water and gas flow rate in model compound test L1

back up to temperature with feedstock flow increased to a higher LHSV of 4 and then a final operating
period at an LHSV of 5. The initial high COD in the effluent is attributed to contamination in the sample

34
port from insufficient clean out. The COD of the effluent appears to stabilize at a very low level with
COD conversion at >99%.

There is a direct linear relationship between increasing gas flow rate and increasing LHSV as
shown in Figure 20.

LHSV, h-1

Figure 20. Relationship of gas flow rate and LHSV in model compound test L1

Based on the data in Figure 19, conversion data can be calculated. Carbon conversion to gas,
COD conversion and overall material balance are plotted in Figure 21. The data calculations are based
on windows of operation extending from about 3 to 5 hours. The consistency of the data is demonstrated
in the graph.

Figure 21. Carbon conversion to gas, COD conversion and mass balance in model compound test
L1

35
 Gas composition was also quite consistent throughout the test, as seen in Figure 22.

Figure 22. Gas product composition in model compound test L1

Analysis of the liquid product by GCMS can be compared to the feedstock, as shown in Figure
23. While the model compounds used in the test are readily observable, there is no detectable organic
material in the aqueous product from the test following the catalytic processing.

Figure 23. Comparison of model compound feed and aqueous product

These results verify that the major organic components in the byproduct aqueous stream from
hydrotreating bio-oil can be effectively converted to gas product by CHG at LHSV of up to 5 even with a
high concentration of organic loading of 14 to 15 wt%. These tests were performed without potential
interference from trace components such as alkali or sulfide.

36
3.3.2 L2 (1/6/2014 – 1/15/2014)
Feed model compounds in DI water simulant with glucose and furfural added
Reactor configuration no sulfur scrubber bed before gasification catalyst bed
Total run time 160 h (time on stream)
Reason for EOR out of feed
Catalyst same catalyst bed from test L1
Flowrates LHSV = 2/h
Product COD ranges Port: 51 ppm – 7 ppm
Gas products CO2:CH4 31:68

The second simulant run proceeded over two weeks including a shutdown over the weekend. The
test included four operating periods: 1) the model mixture was processed to confirm that the catalyst bed
was still active, 2) glucose was added to the model mixture, 3) furfural was added to the model mixture,
and 4) both glucose and furfural were added to the model mixture. The actual components added to make
up the feedstocks are detailed in Table 14. The overall composition remains fairly consistent in all cases.

Table 14. Composition of feed streams used in test L2

Batch 1/6/14 Batch 1/8/14 Batch 1/9/14 Batch 1/13/14
IBR-UOP-L1 last Additional Additional Model compound mix
feed glucose furfural + glucose + furfural
acetone 26.16 7.85 10.44 15.25
ethanol 887.02 266.11 354.51 161.95
p-cresol 132.29 39.69 52.96 77.27
methanol 115.04 34.51 46.09 67.07
1-propanol 113.08 33.92 45.39 65.91
isopropanol 29.03 8.71 11.67 16.95
1-butanol 26.05 7.82 10.47 15.14
2-butanol 22.13 6.64 9.26 13.03
acetic acid 243.11 72.93 97.24 141.64
ethyl acetate 28.06 8.42 11.26 16.38
phenol 4.19 1.26 1.61 2.34
water 8373.84 2512.15 3318.51 3512.30
glucose 30 47.25
furfural 34.59 49.53
Sum of feed 10000.00 3030.00 4004.00 4202.00
Sum of organic 1626.16 517.85 685.49 689.70
Factor (wt. org/wt. feed) 0.16 0.17 0.17 0.16
C 52.7% 52.0% 53.2% 53.2%
H 11.5% 11.2% 11.1% 9.9%
O 35.8% 36.8% 35.6% 37.5%

37
 Figure 24. COD values of the product water and gas flow rate in model compound test L2

In Figure 24, the data for COD in the effluent and the gas product flow rate are plotted over the
time of the experiment. The large blank area in the middle is the weekend break wherein the reactor was
shutdown to idle at room temperature with no feedstock flow. After the weekend the reactor was heated
back up to temperature with feedstock flow. The initial high COD in the effluent is attributed to
contamination in the sample port from insufficient clean out. The COD of the effluent appears to stabilize
at a very low level with COD conversion at >99%.

Based on the data in Figure 24, conversion data can be calculated. Carbon conversion to gas,
COD conversion and overall material balance are plotted in Figure 25. The data calculations are based
on windows of operation extending for varying lengths of time. The consistency of the data is
demonstrated in the graph.

Figure 25. Carbon conversion to gas, COD conversion and mass balance in model compound test
L2

38
 Gas composition was also quite consistent throughout the test, as seen in Figure 26 with little
variation with the minor changes in feed composition.

Figure 26. Gas product composition in model compound test L2

Similar to test L1, analysis of the liquid product by GCMS showed there was no detectable
organic material in the aqueous product from the test following the catalytic processing. Additional
analysis by LC showed effectively the same thing, no observable peaks.

These results verify that the major organic components in the byproduct aqueous stream from
hydrotreating bio-oil can be effectively converted to gas product by catalytic hydrothermal gasification
even with a high concentration of organic loading of 16 wt%. Further, the reaction is not interfered by the
inclusion of either glucose or furfural or both. These two components represent the highly reactive bio-
oil components, which are acknowledged to be the prime candidates for catalyst fouling in the
hydrotreating process. This test showed that if even a significant portion of these components somehow
bypass the hydrotreater reaction zone and were found in the aqueous byproduct, they would not cause
problems in the subsequent catalytic gasification for fuel value recovery. However, these tests were
performed without potential interference from trace components such as alkali or sulfide.

3.4 Mini-reactor tests with PNNL hydrotreater aqueous products

In order to prepare for scaled-up operations to demonstrate the CHG technology on bio-oil
hydrotreating aqueous byproduct, the aqueous products from some of the bio-oil hydrotreating tests at
PNNL were used as feedstock. The scaled-up operation was planned to follow these tests when the
scaled-up hydrotreater at PNNL became operational and sufficient quantities of aqueous byproduct were

39
available. In addition, a new formulation of the ruthenium on carbon catalyst produced by BASF had
become available for use in our technology and validation of the new formulation was also needed.

3.4.1 HT-AQ-CHG-1 (8/20/2014 – 8/29/2014)

An initial test was made with a combined aqueous product from a number of hydrotreating tests.
This material had been in storage for anywhere from weeks to months after the various hydrotreating
tests. Although the specific details for its production were not determined, it was assumed that this would
be useful test of a product representative of the range of products as produced by UOP, as described
above. The test was made with a new formulation of the Ru on carbon extrudate catalyst from BASF.

Feed mixed aqueous byproducts from PNNL bench-scale bio-oil hydrotreating

Reactor configuration sulfur scrubber bed before gasification catalyst bed
Total run time 185 h (time on stream)
Reason for EOR deactivated catalyst
Catalyst 15.22 g Raney Ni, 504020656 Ru/C 11.93 g→23.39 mL (0.51g/cc density
measurement)
Flowrates LHSV = 2/h → 1/h
Product COD ranges 52 ppm – 6000 ppm
Gas products CO2:CH4:H2:C2 [Link] → [Link]

The test was on stream for 6 days before catalyst deactivation became evident. The effluent COD
was never as low as with the model compound tests using the C3610 catalyst. Also, the hydrogen level in
the product gas remained higher than is typically seen previously. The COD in the effluent continued to
increase in subsequent days even after the feed rate was reduced to 1 LHSV. Simultaneously, the gas
composition exhibited the recognized transition from that of a highly active catalyst (strong gas flow with
high methane to carbon dioxide ratio with minimal hydrogen and higher hydrocarbon gases) to that of a
deactivated catalyst (minimal gas flow with high hydrogen, increased higher hydrocarbon gases and
greatly reduced methane and carbon dioxide). Calculations based on gas composition and gas flow show
that the methane and carbon dioxide track together and fall to 1/10th of the initial production while the
hydrogen production about doubles.

Analysis of the aqueous streams by ICP showed a significant amount of S in the feedstock, 43
ppm while the other elements were below detection (<1 ppm) except 23 ppm Si, 11 ppm Na, and 3 ppm
K. IC analysis suggested that most of the S was present as sulfate (93 ppm SO4 = 31 ppm S). The
product effluent showed an initial content of 3 ppm sulfur but none subsequently, suggesting that the
sulfur scrubber (or catalyst bed) was acting as an effective sulfur capture point. The IC analysis
essentially confirmed these readings. Si was nearly constant throughout. Na and K were much elevated
at the beginning of the test, 115 ppm and 75 ppm respectively, and fell back over time, suggesting some
leaching of these elements from the carbon support. There was a slightly noticeable Ni content (1-3 ppm)
in the effluent sporadically through the test. Ru was below the level of detection throughout the test.

Analysis of the scrubber bed and catalyst bed following the shutdown of the test suggested that
the deactivation of the catalyst was a direct result of S poisoning. Although the Ni bed analyses ranged
from 3038 to 16370 ppm S, the Ru catalyst analysis was consistently 11640 to 11260 ppm S. Such a high

40
loading of S on the Ru catalyst, equivalent to a 30% transition to RuS2, in a hydrothermal environment is
expected to effectively deactivate the gasification activity. It further suggests that the sulfur scrubber bed
was not effective in the current configuration. However, since most of the S was found as sulfate in the
feedstock it is not surprising that the Ni did not effectively capture the sulfur as it has been recognized
that the Ru can reduce and capture sulfate much more effectively while the Ni is effective with reduced
sulfur forms.

Subsequent to the CHG test, more detailed analysis of the HT byproduct aqueous phases
produced at PNNL was undertaken. Four product samples were tracked for S by ICP and SO4 by IC as
shown in Table 15. The results suggest that the initially collected products contain high levels of sulfur,
which falls dramatically over a week. The conclusion was that H2S dissolved in the water was volatilized
over time. At the same time, the products as produced contained no detectable sulfate. It was found that
sulfate had formed over 3 weeks, apparently by oxidation of the remaining dissolved sulfur compounds.

Table 15. HT Aqueous sample analyses

58-5 58-5 58-9 58-9 59-5 59-5 59-9 59-9

S, ppm SO4, ppm S, ppm SO4, ppm S, ppm SO4, ppm S, ppm SO4, ppm

as produced 763 ND NA ND 566 ND NA ND

1 days NA ND NA ND NA ND NA ND

5 days NA NA 21 NA 49 NA 42 NA

3 weeks 46 0.5 22 0.3 62 4.8 46 29

ND = below level of detection NA = not analyzed

3.4.2 HT-AQ-CHG-2 (10/3-5/2014, 11/12-14/2014, and 11/22-23/2014)

A second test was undertaken with the PNNL aqueous byproduct from bench-scale tests of bio-oil
hydrotreating. The feedstock was better controlled and was taken from the hydrotreater to the gasification
rig after only a short interlude and with care to limit oxidation. The test was performed in three parts as
dictated by the availability of the aqueous stream from the HT tests.
Feed aqueous byproduct from PNNL bio-oil hydrotreating HT221, HT224, HT225
Reactor configuration sulfur scrubber bed before gasification catalyst bed
Total run time 98.9 h (time on stream)
Reason for EOR deactivated catalyst
Catalyst 9.84 g Raney Ni, 504020656 Ru/C 11.81 g→22.71 mL (0.52g/cc density
measurement)
Flowrates LHSV = 2/h
Product COD ranges 14-44 ppm, 100-2100 ppm, 120-10000 ppm
Gas products CO2:CH4:H2:C2 [Link] → [Link] → [Link]

The composition details of the three feedstocks are provided in Table 16. The other metals not
listed were below the level of detection, <0.8 ppm or <1 ppm. The products actually represent very

41
different hydrotreater operations. HT221 was catalyzed by sulfided catalysts while HT224 and HT225
were catalyzed by precious metal catalysts, which are expected to act as sulfur stripping beds. This effect
is most obvious in HT225 while less so for HT224. Further tests were restricted to the use of aqueous
from hydrotreating with sulfided catalysts only, in order to test the capability of the system for operation
with the more typical catalyst system.

Table 16. Bench-scale HT aqueous byproducts used in CHG minireactor tests

Feedstock HT 221 HT 224 HT 225

S
59 ppm 34 ppm <5 ppm
Na
10 ppm 43 ppm 30 ppm
K
<2 ppm 249 ppm 162 ppm
Ca
<0.8 ppm 544 ppm 372 ppm
Fe
<0.8 ppm 348 ppm 205 ppm
Mg
<0.8 ppm 283 ppm 177 ppm
Mn
<0.8 ppm 26 ppm 16 ppm
Si
14 ppm 1 ppm 1 ppm
COD
16,100 ppm 266,200 ppm 259,000 ppm
N-NH4+ 29 ppm NA NA
pH NA 3.61 4.02
SO4= NA 8 ppm 0.34 ppm

Cl- NA 35 ppm

The tests were operated over the time period described above and the test was terminated during
rd
the 3 portion of the experiment because of blockages of flow. Examination of the BPR discovered black
dust suggesting catalyst disintegration. Examination of the catalyst bed after the experiment discovered
that it was solidified with a white precipitate such that the catalyst bed pellets required drilling to allow
their removal from the tubular reactor. Based on the feedstock analyses in Table 16 showing significant
levels of Fe, Ca, and Mg, it is not surprising that the analysis of the sulfur scrubber bed and the catalyst
bed also showed major deposits of these elements, as presented in Table 17. The sulfur scrubber bed
showed evidence of functioning as designed, but a significant amount of sulfur made it way to the Ru
gasification catalyst and probably contributed to the deactivation of the catalyst.

42
 Table 17. Recovered solids analysis following the three periods of the CHG 2 test

CHG2 Raney Ni CHG2 Ru/C

elements
S scrubber bed catalyst bed
Ni 426550 1218
Al 274200 542
S 8158 1673
Na 46 74
K 92 199
Ca 161 57730
Fe 1940 22645
Mg 422 22645
Mn <35 2056
Si 137 158
Ru <35 51805
Ba <35 64
Sr <35 175
Ag 38 97
Co 288 <35
Cr <35 39
Pt <35 356
Zr <35 53

3.4.3 HT-AQ-CHG-3 (2/3-4 and 7-8/2015)

Feed aqueous byproduct from PNNL bio-oil hydrotreating HT230

Reactor configuration sulfur scrubber bed before gasification catalyst bed
Total run time 78.9 h (time on stream) minus 8 hours of low pressure operation
Reason for EOR deactivated catalyst
Catalyst 8.12 g Raney Ni, 504020656 Ru/C 11.07 g→21.29 mL (0.52g/cc density
measurement)
Flowrates LHSV = 2/h
Product COD ranges 105-443 ppm → 4100-6300 ppm

The mini-reactor for CHG was restarted, after replacing the sulfur scrubber bed and the
gasification catalyst bed following CHG2, when feedstock became available from the bench-scale
hydrotreater operating with a sulfided catalyst. The test operated over two days until the BPR failed and
the reactor pressure dropped to 10 atmospheres over night while the feed continued to flow. The feed
flow was stopped, the BPR was repaired and the test restarted. The catalyst activity had already given
indication of deactivation, based on slightly increased COD in the effluent. The effluent COD level
suggests a COD conversion of 99% falling to 97%. The normal corroborative data from the gas
composition was unavailable due to operator error with the gas chromatograph. After restart, the catalyst
activity, based on COD conversion, was much reduced and the test was terminated shortly thereafter. The

43
COD conversion was only 65% at the restart and it fell to 47% within 12 h of operation. No attempt was
made to regenerate the catalyst.

Sulfate analysis was performed on the feedstock and the aqueous product by ion chromatography
(IC). The feedstock contained 34 ppm sulfate while the products ranged from 1 to 3 ppm sulfate,
suggesting that there was significant sulfur retention in the reactor. The sulfate analysis was confounded
by an underlying broad peak that was not identified. The predominant peak in the IC analysis of the
feedstock was tentatively identified as thiosulfate, which was not detected in the CHG products; therefore
we conclude that it was reacted at the hydrothermal conditions. Chloride levels were also significant,
about 25 ppm in the feedstock and 30 ppm in the CHG aqueous product.

44
 4.0 Scaled-up reactor tests with simulated hydrotreater
byproduct water
As the hydrotreater planned for the Hawaiian IBR was never built and the scaled-up hydrotreater
at PNNL was never operational in the time period of this project, a simulant test was made with the
mobile scaled-up unit. The test was performed as a means to demonstrate the refurbished system and
validate its operability for future needed processing. It is envisioned that HTL aqueous byproduct
processing or hydrotreater aqueous processing by CHG might be a useful option and would be wanted
when the scaled-up processing units for HT and HTL at PNNL are brought on-line later in FY2015 or
FY2016.

In order to perform a test in the scaled-up reactor, a chemical model of the hydrotreating aqueous
product was produced in quantity from model compounds, similar to that tested earlier. The model
compounds were acquired at 99% purity or as high a purity as was available. The specific components
were selected based on the list of components found by UOP in their byproduct waters from hydrotreating
bio-oil in their first 35 runs at Des Plaines. The composition of the batch made for the test is shown in
Table 18. Although sulfide was expected in the aqueous byproduct, it was not measured by UOP and
was not included in the simulant. Alkali metals and alkaline earths were believed by UOP to be below 10
ppm and were not added to the simulant.

Table 18. Composition of the batch of simulated bio-oil hydrotreating water product

organics Weight, kg
ethanol 12.00 (95%) + 3.75 (100%)
methanol 1.92
1-propanol 1.76
2-propanol 1.12
1-butanol 0.45
2-butanol 0.33
acetic acid 3.21
ethyl acetate 0.16
phenol 2.46
SUM 27.16
water 295
Factor (wt. organic/wt. feed) 0.082
COD, ppm measured 171,500
C 53.1%
H 11.6%
O 35.2%

45
 A single test was made in the scaled-up reactor on May 12, 2015.

Feed aqueous byproduct simulant

Reactor configuration no sulfur scrubber bed before gasification catalyst bed
Total run time 16 h (time on stream) including 7 hours of heat-up and cool-down with DI
water flow
Reason for EOR poor pressure control resulted in inconsistent operation
Catalyst 504020656 Ru/C 2371g → 4.4 L (0.54g/cc density measurement)
Flowrates LHSV = 2/h
Product COD ranges 50 ppm initially, but essentially 0 ppm through most of the test

This short demonstration test validated the operating systems in the mobile scaled-up reactor
system with the exception of the 2-stage BPR pressure let-down operation. During the heat-up portion of
the test using only DI water as the feedstock, there was steady pressure control. However, within 20
minutes of introduction of the organic laden feedstock followed by the initiation of gas production, the
BPR system began to react erratically. Within the first hour of feedstock introduction, the initial water
samples showed evidence of catalyst particulate carryover, despite extensive washing of the catalyst bed
previously. After 2 h with feedstock on line the water samples had cleared of particulate and remained so
for the balance of the test. Yet the pressure instability continued throughout the test and even increased in
severity toward the end of the test. After shutdown of the simulant feed and return to DI water flush the
pressure instability subsided. These results suggest a difficulty for proper pressure letdown over the
Tescom BPRs with the mixed phase product containing both liquid water and fuel gas.

During the 5 h window of relatively steady-state operation, gasification of the organics in the
feedstock was essentially complete with no measureable COD in the effluent. The product gas
composition was essentially 67% methane, 31 % carbon dioxide and 2% hydrogen. While feeding the
171,500 ppm COD feedstock at a LHSV of 1.94 L/L/h, a gas product of 97 L of gas per L of feedstock
was produced. There was no evidence of catalyst deactivation over the period of this short run as the gas
composition remained constant as did the COD conversion.

46
 5.0 Conclusions and Recommendations

PNNL was tasked with developing catalytic hydrothermal gasification (CHG) for use
with aqueous streams within the pyrolysis biorefinery. These aqueous streams included the
aqueous phase separated from the fast pyrolysis bio-oil and the aqueous byproduct streams
formed in the hydroprocessing of the bio-oil to finished products. As other aqueous streams
were identified of interest within the biorefinery, these might also have been tested in CHG, but
none were identified. The development work progressed at two levels. Initial tests were made in
the laboratory in mini-reactor scale and bench-scale continuous-flow reactor systems. Following
positive results, the next level of evaluation was in the scaled-up engineering development
system.

The results of tests with bio-oil aqueous fraction suggested that the organic contaminants,
even at high concentration, can be effectively converted to a fuel gas product by CHG. However,
corrosion of the stainless steel reactor system by the acidic components appeared to be significant, based
on the metal deposits on the catalyst, but no noticeable change was seen in the reactor itself. Sulfur
stripping may also be necessary in order to maintain long-term catalyst activity. Carbon fouling of the
catalyst was also suggested by the loss of porosity and surface area. Handling of the aqueous stream was
also identified as an issue as the acidic components were corrosive to steel cans and could not be stored
for any period of time in such without resulting in a significant iron contamination and even failure of the
container.

Tests were performed in the mini-reactor with various aqueous products from the UOP bio-oil
hydrotreating tests used as feedstock. The products expressed a range of component composition and
concentration. Based on these results it appeared that a range of hydrotreating of bio-oil was undertaken
such that the product was more or less hydrogenated and the resulting aqueous byproduct had more or
less soluble oxygenated components. COD of up to 178,250 ppm was effectively processed at 2 LHSV in
fixed-bed, continuous-flow catalytic reactors achieving >99% conversion in both the bench-scale system
and the mini-reactor. The typical product containing methane and carbon dioxide was produced.

Trace elements were quite low in the hydrotreating aqueous byproducts. Traces of biomass
minerals were present. Sulfur content ranged over an order of magnitude from 2 to 20 ppm. Ion
chromatography of the feedstocks suggested that sulfate was not detectable.

The results with these feedstocks containing only low levels of sulfur showed that system
modification to accomplish sulfur scrubbing from the feed would be required to protect the CHG catalyst.
After refrigerated storage for a month and a half, analysis by IC suggested the aqueous byproduct
contained 29 ppm sulfate. This measurement of high sulfate in the stored aqueous byproduct suggested
that the initial sulfide in the feed was oxidized to sulfate over time. ICP analysis showed 25 ppm sulfur,
therefore about 40% of the sulfur had been oxidized to sulfate. Other elements found in this analysis
were 28 ppm Na, 22 ppm Si, 3.5 ppm K, 1.9 ppm Al, and 1.8 ppm Ca. Clearly there was not enough Ca
to precipitate all the sulfate in the solids settler, which would allow the soluble sulfate to impact on the
catalyst activity over time.

47
 Detailed analysis of the HT byproduct aqueous phases produced at PNNL was undertaken for S
by ICP and SO4 by IC. The results suggested that the initially collected products contained high levels of
sulfur, which fell dramatically over a week. The conclusion was that H2S dissolved in the water was
volatilized over time. At the same time, the products as produced contained no detectable sulfate. It was
found that sulfate had formed over three weeks, apparently by oxidation of the remaining dissolved sulfur
compounds.

Analysis of the catalyst and scrubber beds after one test confirmed that a significant amount of
sulfur was captured in the scrubber (about ¼) but over 10% ended up on the catalyst while another quarter
passed through in the aqueous stream. The sulfur on the catalyst and in the aqueous apparently was the
sulfate that was not captured by the Ni scrubber. Some of the balance of the sulfur precipitated at
hydrothermal conditions but was not collected in the separator vessel, but was mobilized by the liquid
water flow and passed on through the system to the condensate product collector.

The detailed analysis showed that the Ni in the scrubber was only slightly (<1%) reacted to nickel
sulfide. It did not serve as an effective sulfur scrubber. However, the fact that a large portion of the S
was in the form of sulfate, and therefore less reactive, is at least a partial explanation. The sulfur was
evenly distributed through the Ni bed suggesting that it was reacting slowly. Therefore, the Ru catalyst
had been sulfided to a large degree and certainly enough to have a large effect on its gasification and
methanation activity. EDS analysis showed that the sulfur was highly associated with both the Ni and the
Ru. There was also evidence of Ca and P (probably as calcium phosphate) coating the surface of the Ni
scrubber. The Ru concentration in the spent catalyst had been reduced to 3/4th of the starting level,
presumably by carbon deposition in the pores, as there was no evidence of Ru loss to the aqueous stream.

Extended runs were also performed with model compound mixtures simulating the hydrotreater
aqueous byproduct at the high concentration end of the range (313,500 to 410,000 ppm COD). These
experiments which did not include sulfide compounds or alkali impurities operated at high conversion
>99%, even at LHSV of up to 5. Gas composition was also quite consistent throughout the test with little
variation with the minor changes in feed composition. There was a direct linear relationship between
increasing gas flow rate and increasing LHSV. Analysis of the liquid product by GCMS showed there
was no detectable organic material in the aqueous product from the test following the catalytic processing.
Additional analysis by LC showed effectively the same thing, no observable peaks.

These model compound results verified that the major organic components in the byproduct
aqueous stream from hydrotreating bio-oil could be effectively converted to gas product by catalytic
hydrothermal gasification even with a high concentration of organic loading of 16 wt%. Further, the
reaction was not interfered by the inclusion of either glucose or furfural or both. These two components
represent the highly reactive bio-oil components, which are acknowledged to be the prime candidates for
catalyst fouling in the hydrotreating process. This test showed that if even a significant portion of these
components somehow bypass the hydrotreater reaction zone and were found in the aqueous byproduct,
they would not cause problems in the subsequent catalytic gasification for fuel value recovery. However,
these tests were performed without potential interference from trace components such as alkali or sulfide.

The mobile scaled-up reactor system at PNNL was refurbished and put into operating condition
by the addition of a modern data acquisition and control system. The unit was operated with simulated

48
aqueous byproduct to demonstrate its functionality. The only question arising from the test was the
failure of the pressure control system, which will require further improvements to allow long-term
operation in this application.

49
 6.0 References
i. Sealock, L. J., Jr.; Elliott, D.C.; Baker, E. G.; and Butner, R. S. Chemical Processing in High-
Pressure Aqueous Environments. 1. Historical Perspective and Continuing Developments. Ind.
Eng. Chem. Res. 1993, 32, 1535.
ii. Elliott, D.C.; Sealock, L. J., Jr.; and Baker, E. G. Chemical Processing in High-Pressure Aqueous
Environments. 2. Development of Catalysts for Gasification. Ind. Eng. Chem. Res. 1993a, 32,
1542.
iii. Elliott, D.C.; Hart, T.R.; Neuenschwander, G.G. Chemical Processing in High-Pressure Aqueous
Environments. 8. Improved Catalysts for Hydrothermal Gasification. Ind. Eng. Chem. Res.
2006, 45, 3776.
iv. Elliott, D.C.; Phelps, M. R.; Sealock, L. J., Jr.; and Baker, E. G. Chemical Processing in High-
Pressure Aqueous Environments. 4. Continuous-Flow Reactor Process Development
Experiments for Organics Destruction. Ind. Eng. Chem. Res. 1994, 33, 566-574.
v. Elliott, D.C.; Neuenschwander, G.G.; Phelps, M. R.; Hart, T.R.; Zacher, A.H.; and Silva, L.J.
Chemical Processing in High-Pressure Aqueous Environments. 7. Process Development for
Catalytic Gasification of Wet Biomass Feedstocks. Ind. Eng. Chem. Res. 2004, 43, 1999.
vi. Patrick, H.R., Griffith, K., Liotta, C.L., Eckert, C.A., and Gläser, R. Near-Critical Water: A
Benign Medium for Catalytic Reactions. Ind. Eng. Chem. Res. 2001, 40, 6063.
vii. Elliott, D. C.; Sealock, L. J., Jr.; and Baker, E. G. Method for the Catalytic Conversion of
Organic Materials into a Product Gas. U.S. Patent 5,616,154, 1997.
viii. Sealock, L. J., Jr.; Baker, E. G.; and Elliott, D. C. Method for Catalytic Destruction of Organic
Materials. U.S. Patent 5,630,854, 1997.
ix. Elliott, D.C.; Neuenschwander, G.G.; Phelps, M. R.; Hart, T.R.; Zacher, A.H.; Silva, L.J.
Chemical Processing in High-Pressure Aqueous Environments. 6. Demonstration of Catalytic
Gasification for Chemical Manufacturing Wastewater Cleanup in Industrial Plants. Ind. Eng.
Chem. Res. 1999, 38, 879-883.
x. Marinangeli, R.; Boldingh, E.; Cabanban, S.; Fe, Z.; Ellis, G.; Bain, R.; Hsu, D.; Elliott, D.
Pyrolysis Oil to Gasoline—Final Report, CRADA between Battelle, Midwest Research Institute,
and UOP LLC. September 30, 2009.

50
i
Sealock, L. J., Jr.; Elliott, D.C.; Baker, E. G.; and Butner, R. S. Chemical Processing in High-Pressure
Aqueous Environments. 1. Historical Perspective and Continuing Developments. Ind. Eng. Chem. Res.
1993, 32, 1535.
ii
Elliott, D.C.; Sealock, L. J., Jr.; and Baker, E. G. Chemical Processing in High-Pressure Aqueous
Environments. 2. Development of Catalysts for Gasification. Ind. Eng. Chem. Res. 1993a, 32, 1542.
iii
Elliott, D.C.; Hart, T.R.; Neuenschwander, G.G. Chemical Processing in High-Pressure Aqueous
Environments. 8. Improved Catalysts for Hydrothermal Gasification. Ind. Eng. Chem. Res. 2006, 45,
3776.
iv
Elliott, D.C.; Phelps, M. R.; Sealock, L. J., Jr.; and Baker, E. G. Chemical Processing in High-Pressure
Aqueous Environments. 4. Continuous-Flow Reactor Process Development Experiments for Organics
Destruction. Ind. Eng. Chem. Res. 1994, 33, 566-574.
v
Elliott, D.C.; Neuenschwander, G.G.; Phelps, M. R.; Hart, T.R.; Zacher, A.H.; and Silva, L.J. Chemical
Processing in High-Pressure Aqueous Environments. 7. Process Development for Catalytic Gasification
of Wet Biomass Feedstocks. Ind. Eng. Chem. Res. 2004, 43, 1999.
vi
Patrick, H.R., Griffith, K., Liotta, C.L., Eckert, C.A., and Gläser, R. Near-Critical Water: A Benign
Medium for Catalytic Reactions. Ind. Eng. Chem. Res. 2001, 40, 6063.
vii
Elliott, D. C.; Sealock, L. J., Jr.; and Baker, E. G. Method for the Catalytic Conversion of Organic
Materials into a Product Gas. U.S. Patent 5,616,154, 1997.
viii
Sealock, L. J., Jr.; Baker, E. G.; and Elliott, D. C. Method for Catalytic Destruction of Organic
Materials. U.S. Patent 5,630,854, 1997.
ix
Elliott, D.C.; Neuenschwander, G.G.; Phelps, M. R.; Hart, T.R.; Zacher, A.H.; Silva, L.J. Chemical
Processing in High-Pressure Aqueous Environments. 6. Demonstration of Catalytic Gasification for
Chemical Manufacturing Wastewater Cleanup in Industrial Plants. Ind. Eng. Chem. Res. 1999, 38, 879-
883.
x
Marinageli, R.; Boldingh, E.; Cabanban, S.; Fe, Z.; Ellis, G.; Bain, R.; Hsu, D.; Elliott, D. Pyrolysis Oil
to Gasoline—Final Report, CRADA between Battelle, Midwest Research Institute, and UOP LLC.
September 30, 2009.
 Appendix B

Life Cycle Analysis Executive Summary

Michigan Technological University

Note:

This appendix is the Executive Summary provided by Michigan Technological University

for the Life Cycle Analysis on pyrolysis oil upgrading to transportation fuels. This work
was part the IBR project scope.
 Sustainable Futures Institute
Office: 840 Dow Env. Sci. & Engr. Bldg.
Michigan Technological University Phone: 906-487-3612
1400 Townsend Drive Facsimile: 906-487-2943
Houghton, MI 49931-1295 email: sfi@[Link]
website: [Link]

Executive Summary

DOE IBR Pyrolysis LCA: Phase I

LIFE CYCLE ASSESSMENTS OF PYROLYSIS-BASED GASOLINE AND DIESEL FROM

DIFFERENT REGIONAL FEEDSTOCKS: CORN STOVER, SWITCHGRASS, SUGAR
CANE BAGASSE, WASTE WOOD, FOREST RESIDUES, ALGAE, AND ALBIZIA

By

David R. Shonnarda,b, Daniel Fogliattia, Michael Massuccia, Matthew J. Mihalekc, Kareen

Krizzan Encinas Sotod

a Department of Chemical Engineering, Michigan Technological University, Houghton, MI

b Sustainable Futures Institute, Michigan Technological University, Houghton, MI

c now at Environmental Resource Management, Milwaukee, WI

d visiting from Department of Chemical Engineering, University of Sonora, Hermosillo, MX

Submitted to Dan Ellig and Jim Anderson

UOP (a Honeywell Company)

Des Plaines, IL, USA

August 06, 2014

[Link] Page 1
We prepare students to create the future

Michigan Technological University is an equal opportunity educational institution/equal opportunity employer

Table of Contents
Introduction 3
Research Objectives 3
System Boundaries and Description of Product System 4
Feedstock Types 4
Input Data 5
Life Cycle Assessment Methodology 6
Life Cycle Results 7
Greenhouse Gas Emissions 7
Energy Demand 8
Water Consumption 9
Scenario Analyses 10
Conclusions 12
References 13

Page 2
Introduction
This executive summary is for the life cycle assessments (LCAs) conducted for the DOE-funded
Integrated Biorefinery (IBR) Phase I project initiated in May 01, 2010 and ending in July 31,
2012 including the no-cost extension period. Pyrolysis bio-oil is a product of the rapid thermal
decomposition of woody biomass to form a mixture of low molecular weight oxygenated organic
compounds, water, and high molecular weight pyrolytic lignin. Pyrolysis bio-oil is a feedstock
for catalytic upgrading in the presence of hydrogen to produce a biofuel that is a direct hydrocar-
bon replacement for petroleum gasoline and diesel; so called “drop-in” advanced biofuels
PyGasoline and PyDiesel. The Renewable Fuels Standard (RFS) contained in the Energy Inde-
pendence and Security Act (EISA) of 2007 stipulates that 36 billion gallons of biofuels will be
produced per year by 2022, and up to 21 billion gallons can be other than corn ethanol. In order
to qualify as advanced biofuels under the RFS, these replacement fuels must achieve 60% or
more greenhouse gas (GHG) savings compared to petroleum fuels. Key inputs to the PyGasoline
and PyDiesel life cycles need further investigation to reduce uncertainty and generate more
accurate predictions of savings in greenhouse gases (GHGs) compared to petroleum fuels.

This preliminary LCA occurred over a 15 month research program (May 1, 2010 – July. 31,
2011) at Michigan Tech by Matthew Mihalek, a master of science student in the Department of
Chemical Engineering at Michigan Technological University (Michigan Tech) and by his
adviser, Professor David Shonnard. After submitting a draft of his thesis in 2012, Mr. Mihalek
left for a position in the environmental firm Environmental Resources Management in
Milwaukee, WI. This executive summary has been taken from the final thesis of Mr. Mihalek
which was submitted to the Graduate School in August, 2014. The version submitted to
Michigan Tech had confidential data excluded while the version with confidential data is
included as an appendix to this Executive Summary. The research was divided nearly evenly
along two main tasks; i) develop input data through surveys of biomass growers and suppliers
leading to an analysis of GHG emissions from biomass cultivation, harvesting, and transportation
as well as a preliminary analysis of the effects of land management and land use change on the
GHG emissions, and ii) conduct preliminary but complete GHG LCAs of PyGasoline and
PyDiesel based inputs derived from engineering calculations by professionals at UOP. Research
objectives are listed next.

Research Objectives
1. Develop complete “cradle-to-gate” analyses for the production of each feedstock on the
basis of 1 dry metric ton of feedstock.
2. Develop complete “cradle-to-gate” analyses for stabilized pyrolysis oils from each
feedstock on the basis of 1 MJ of stabilized pyrolysis oil.
3. Develop complete “cradle-to-grave” LCAs for each pyrolysis-based biofuel on the basis
of 1 MJ of total transportation fuel (gasoline+diesel).
4. To gain an understanding of the relative importance of biofuel pathway stages by
organizing LCA results for each feedstock-specific pyrolysis-based biofuel around each
stage.
5. To investigate the relative importance of key LCA inputs and assumptions through
scenario analyses.
6. Make recommendations on ways to reduce environmental impacts of pyrolysis-based
biofuels produced from select biomass feedstocks.

Page 3
 System Boundaries and Description of Product System
The system boundary is the complete life cycle from cradle-to-grave. Figure ES1 shows each
stage in the life cycle, including major inputs and outputs, from feedstock production to biofuel
combustion in an engine. In the feedstock production stage, this study includes inputs such as
fertilizers, fuel (including electricity) for harvesting or collecting each feedstock, and
transporting feedstock to the RTP facility. After feedstock production, each pathway has
electricity inputs to reduce the size of the biomass in preparation for pyrolysis, in a stage called
pretreatment. Pyrolysis has inputs like natural gas and electricity used to dry and then heat the
biomass up to approximately 500°C. The pyrolysis oil (pyoil) needs to be stabilized, which has
electricity inputs, chemicals for ion exchange, and ethanol as a rinse solvent. The last stage is
upgrading to hydrocarbon biofuels and the main input is natural gas used for fuel and to produce
hydrogen for catalytic hydro-treatment of the stabilized pyoil. Most of these stages also require
either cooling water, boiler feed water, or rinse water or some combination of water resources.

Ethanol Electricity

Fertilizer Biomass Biomass Biomass

Fuel Production Transport Pretreatment

Electricity Pyrolysis Stabilization PyOil Natural Gas

Natural Gas RTP Process & Transport Upgrading
Steam Steam

Biofuel Biofuel
Transport Combustion

Filter Cake

Figure ES1: System boundary (dashed line) for the LCA with some important inputs and co-
products

Feedstock Types
In this study, many feedstocks were modeled from data provided by several companies and
individuals, as shown in Table ES1. Figure ES2 shows representative regions in the U.S. for
cultivation of the various biomass feedstocks used in this study based on information from
feedstock providers in this study. Waste wood includes all the tree branches and some stumps
and roots produced during lumber harvesting, with input data from a location in Washington
State. Corn stover is the stalk and leaves of corn that is left in the field after corn grain harvest,
and it is assumed to be cultivated in the Midwest of the U.S. Switchgrass can be grown in much
of the United States as shown in Figure ES2, especially in the Southeast and in Mid-Atlantic
states. Sugarcane bagasse is a lignocellulosic waste product from sugar cane processing
assuming cultivation in the southern states of the U.S. and in Brazil, among other countries.
Algae can be grown in regions with ample sun year-round, abundant water supplies, and where

Page 4
Table ES1: Yield and LHV references for each feedstock
LHV sources and
Feedstocks Yield sources and references
references
Switchgrass Ceres (1) UOP (2)
Guinea grass (managed) David Ringuette (3) UOP
Guinea grass (unmanaged) David Ringuette UOP
Wild Algae No data UOP
Open pond Algae (Stratton et al. 2010) UOP
Waste wood U.P. Survey Grays Harbor (4)
Corn stover (Morey et al. 2010) (Morey et al. 2010)
Cane bagasse (U.S. Dept. of Energy 2011) (CARB 2009)
Albizia (Tenbruggencate 2008) UOP
(1) Contacts are Sam Harris and Spencer Swayze
(2) Original contact at UOP for IBR Phase I project is Tom Kalnes (retired)
(3) University of Hawaii consultant to Imperium
(4) Contact is Bruce McComas

Figure ES2: Map with regions of each feedstock in the United States

nutrient runoff causes blooms in rivers and lakes. One promising region is in the southwestern
part of the United States, where land is not suitable for food crops, where sun shines most of
the year, and where deep saline aquifers may provide a water resource. Guinea grass and albizia,
both of which are invasive species, grow wild in Hawaii.

Input Data
Input data on the basis of 1 dry metric ton biomass for each of the feedstocks can be found in
Appendix 3 of the accompanying MS thesis by Mihalek. Input data for production of stabilized

Page 5
pyrolysis oil on the basis of 1 kg stabilized pyoil are in Appendix 4 of the thesis. Input data for
catalytic upgrading of stabilized pyoil on the basis of 1 kg of final hydrocarbon biofuel product
(gasoline+diesel) are in Appendix 5.

Life Cycle Assessment Methodology

The LCA methodology descriptions are found in Chapter 2 of the final MS thesis of Mihalek.
The scope of the analysis is cradle-to-grave, from biomass production to combustion of biofuel.
For all of the inputs to the biofuel life cycle, inventory data for each input were obtained from
the ecoinvent™ database in the LCA software tool SimaPro. Although the ecoinvent™ database
is comprised of European inventory data, we believe that these data are a fair representation of
U.S. conditions because modern production facilities and methods are practiced in both
continents. Electricity use in the various biofuel LCAs was modeled using a U.S. average grid
mix of energy sources (EIA, 2012). In nearly every product LCA, there can be many co-
products created within the same product system. Co-products are allocated a portion of the
environmental burdens from the product life cycle by using different methods, such as energy or
displacement allocation. In this LCA there are multiple co-products, which are shown in Figure
ES1 and are discussed in detail in chapters 4 and 5 of the thesis. This study looked at two ways
of allocating life cycle impacts to the co-products. The first is displacement allocation (also
called system expansion), which assigns all environmental burdens to the transportation biofuels
and then takes a credit for the avoided products displaced in the market by the co-products. For
example, steam is produced in more than one stage of the product life cycle as shown in Figure
ES1. Steam is exported from the product system and is available for use in the industrial sphere
where it may displace steam generated by natural gas, fuel oil, or coal. When displacement
occurs, an environmental credit equal to the impact avoided can be claimed by the main product.
In the results, this credit will show up as a negative emission and helps offset the other emissions
in the process. Displacement allocation is used by the U.S. Environmental Protection Agency to
evaluate biofuels and determine whether they count toward the Renewable Fuels Standard
production targets.

This LCA also employs energy allocation as a scenario, which is favored by the European Union
(EU) biofuel regulatory agencies. Environmental impacts included in this study are greenhouse
gas (GHG) emissions, cumulative energy consumption, fossil energy consumption, and water
footprint. Global warming potentials from the IPCC GWP 100a method were employed for all
greenhouse gases emitted, but the most important greenhouse gases are CO2, CH4, and N2O. The
environmental impacts were tabulated in the thesis at important intermediate stages of biomass
production (basis of 1 metric ton, MT, in chapter 3), stabilized pyoil (bases of 1 MT and 1 MJ in
chapter 4), and hydrocarbon biofuel (bases of 1 MT and 1 MJ in chapter 5). Several scenarios
were also included in this study to account for important study assumptions. In the open-pond
algae feedstock case, one scenario looked into reducing nutrient inputs of synthetic fertilizers
into algae cultivation by recycling nutrients from an anaerobic digestion unit processing lipid-
extracted algae. In the wild algae biofuel analysis, a scenario was investigated for the case when
algae cultivation utilized nitrogen-rich waters from a wastewater treatment plant. In both cases,
environmental impact credits are taken for avoiding 75% of the normal inputs of synthetic
fertilizers in the first scenario and for avoiding expensive denitrification in the wastewater
treatment. Switchgrass cultivation in different states was investigated in a set of scenarios in
which inputs of fertilizers and fuels and yields were varied according to local conditions. A corn

Page 6
stover scenario included a portion of corn cultivation inputs as per the Low Carbon Fuel
Standard methodology from California.

Life Cycle Results

Greenhouse Gas Emissions: The LCA results presented below are organized around each product
stage in order to highlight the most important stages for future process improvement. Figure ES3
gives a graphical representation of the contributions to GHG emissions by each major stage for
each feedstock-based biofuel. Contributions to GHG emissions by feedstock production are
significant but highly variable due to the nature of inputs to this stage. Bagasse, waste wood, and
corn stover require no cultivation, only collection and transport, while wild algae and open pond
algae require very large inputs of electricity and nutrients, respectively. Switchgrass and guinea
grass exhibit intermediate GHG emissions among the various feedstocks, and while these
feedstocks need fertilizer inputs, intensive cultivation reduces transport distances and emissions.
Pyrolysis GHG emissions are similarly highly variable, with some feedstocks having high
emissions while others are lower, and in which the amount of steam credit is an important factor.
Stabilization of pyoil uses ethanol, produces a solid char filtrate and generates emission credits
for the displacement of fossil fuel in the market (coal) by the char. Biofuel upgrading GHG
emissions are also highly variable because the impacts from this stage depend on the details of
the chemistry and the extent of production of C1-C4 hydrocarbons which are used internally to
displace natural gas for generating H2 for hydro-treatment.

90

80

Feedstock Production Pyrolysis

70
Stabilization Upgrading
60

50
GHG g CO2 eq./MJ

40

30

20

10

-10

-20
Albizia Bagasse Guinea Open Pond Corn stover LA Corn Switchgrass Waste Wild Algae
Grass Algae Stover Wood

Figure ES3: Greenhouse Gas (GHG) Emissions Breakdown for Various Biofuels

Page 7
The net GHG emissions are shown in Figure ES4. This figure also shows a horizontal line
representing the biofuel emissions that achieve 60% savings compared to fossil fuel (90.5 g
CO2 eq./MJ fossil fuel), a savings required by the RFS2 (EPA 2010a). Wild algae is the only
feedstock in which the biofuel does not meet the minimum 60% savings target to be considered
an advanced biofuel (EPA 2010a). Wild algae does not qualify mainly due to the large amount
of electricity used in the feedstock production stage and could be greatly reduced if the density of
algae is increased or if “cleaner” forms of electricity are used compared to the U.S. average grid.
An alternative feedstock production pathway for wild algae grown in waste water treatment
plants greatly reduces the GHG emissions for the feedstock production stage by assuming the
avoidance of emissions associated with denitrification of NH3 in waste water (see chapter 4 in
the thesis).

80
74.31

70

60

50
GHG g CO2 eq./MJ

40 36.17 36.17
35.46 34.42

30 26.36

20.28 19.25
20 17.38

10.71
10

0
Albizia Bagasse Guinea Open Pond Corn stover LA Corn Switchgrass Waste Wild Algae
Grass Algae Stover Wood

Figure ES4: Greenhouse Gas (GHG) Emissions Totals for Various Biofuels

Energy Demand: An impact category in this study is the cumulative energy demand (CED) of
each biofuel. Figure ES5 shows the breakdown of CED for each biofuel on the basis of type of
energy. By far, the largest portion of CED is from biomass energy in each feedstock, but non-
renewable fossil energy is also a large contributor, mostly from electricity use and natural gas.
Nuclear energy is the third most important type and is derived from electricity consumption. The
CED of fossil transportation fuels such as gasoline, diesel, and jet are in the range of 1.3;
much lower than the biofuels in Figure ES5. There is a good reason for this difference. Fossil

Page 8
fuels are derived from crude oil, and this feedstock needs only a relatively low intensity of
refining compared to biofuels. Biofuels, for example, require that oxygen be eliminated from the
pyrolysis bio-oil, and this step is very energy intensive. On the other hand, most of the CED is
from renewable biomass energy, which is, as the name states, a renewable resource.

4.00

3.50 Non renewable biomass Renewable others

3.28
Renewable water Non renewable nuclear
3.00 2.8 Non renewable fossil Renewable biomass
MJ/MJ of Biofuel

2.50 2.26
2.35
2.18 2.24 2.14
2.03
2.00

1.50 1.25

1.00

0.50

0.00

Figure ES5: Cumulative Energy Demand (CED) Breakdown of Various Biofuels

The second energy-related impact category is fossil energy demand (FED), which is important
because imported crude oil is non-renewable fossil resource, and achieving energy security will
require that the U.S. use less fossil energy. Figure ES6 shows the breakdown of FED for each
biofuel for major categories of inputs and co-products in the life cycle, and also shows the net
FED result above each bar. Electricity, natural gas, and fertilizers are the main inputs which
contribute to FED, and ethanol is important only for guinea grass and wild algae, presumably
because these feedstocks contain high ash. Large fossil energy credits are from exported steam
and to a lesser extent filter cake char.

Water Consumption: The water consumption footprints for each feedstock-based biofuel are
listed by major life cycle input and output in Tables ES2a and ES2b. The major sources of water
consumption are cooling water for pyrolysis and catalytic upgrading. The major credit for water
generation is in the catalytic upgrading step where water is a byproduct of hydro-treatment and is
sent to wastewater treatment. Wild algae biofuel shows a very large water credit because this
approach is actually a remediation of nutrient contaminated streams, lakes, and ponds. The
removal of wild algae biomass from these aquatic environments is likely to prevent the oxygen

Page 9
loss that would normally accompany algae proliferation. Open pond algae biofuel exhibits
almost ten times greater consumption when compared to other feedstock biofuels because of
evaporation from open ponds. If open pond cultivation utilized deep saline aquifers or ocean
water, then it can be argued this consumption would not count toward the water footprints
because these sources of water are not considered fresh water. In this analysis, energy crops
such as switchgrass and guinea grass are assumed to be grown without irrigation, and therefore
water impacts are minimized. The majority of biofuels exhibit a water footprint of between 1.7
and 3.3 gallons water consumed / kg biofuel.

1.5
Natural Gas Electricity Truck transport Fertilizer
Ethanol Filter cake Steam credit Remaining 1.10
0.60
1 0.33 0.22 0.43
0.27 0.23
MJ of Fossil Energy/MJ of Biofuel

0.39

0.12
0.5

-0.5

-1
Albizia Bagasse Guinea Open Pond Corn Stover LA Corn Switchgrass Waste Wild Algae
Grass Algae Stover Wood

Figure ES6: Fossil Energy Demand (FED) Breakdown for Various Biofuels

Scenario Analyses: A large number of scenarios were investigated in this study to understand the
importance of assumptions and to probe location-specific effects. The reader is directed to the
thesis for more details, but a synopsis is presented here. The main assumption in the LCA
method that was included as a scenario was the allocation method; either displacement (system
expansion) or energy allocation. The results in this Executive Summary are for displacement
allocation. Displacement allocation is the method used by the U.S. EPA to evaluate new biofuel
pathways. Another methodology assumption is whether direct land use change (dLUC)
emissions of CO2 be included in the switchgrass analysis. The dLUC emissions exhibited a wide

Page 10
variability depending on location of switchgrass cultivation, and section 3.1.2 of the thesis and in
the appendix in table A3.4 show the dLUC results. The assumption of grid electricity for wild
algae cultivation (U.S. grid average, coal, natural gas) was investigated in a separate scenario
with significant effect on GHG results for algae cultivation. The reader is encouraged to read the
thesis and appendices for further details on scenario analyses.

Table ES2a: Water use for each feedstock in gallons of water per kg of biofuel assuming 2% and
5% water consumption for cooling water (CW) and boiler feed water (BFW), respectively
Water Use Bagasse Stover Guinea Grass Switchgrass
Feedstock Production
Pyrolysis
Cooling water (25-29°C) 1.79 2.41 2.39 2.10
Heated Water (60-75°C) 6.91E-03 9.57E-03 1.02E-02 8.05E-03
Stabilization
Rinse Water 5.78E-02 7.28E-02 0.18 6.68E-02
Spent Brine -0.11 -0.14 -0.35 -0.13
Water from 98% H2SO4 8.15E-05 1.03E-04 2.50E-04 9.35E-05
Water from 50% NaOH 3.07E-03 3.87E-03 9.37E-03 3.51E-03
Water from 10% NaCL 4.19E-02 5.28E-02 0.13 4.85E-02
Upgrading
Cooling water 9.08E-02 0.11 0.11 0.10
Boiler Feed Water 2.75E-02 3.22E-02 3.24E-02 3.17E-02
Wastewater -0.25 -0.34 -0.33 -0.35
Integrated H2 Plant
Demineralized Water 3.74E-02 4.39E-02 4.64E-02 5.01E-02
WWT using CHG
Cooling water Make-up 4.37E-04 6.11E-04 5.88E-04 6.15E-04
Total (gallons/kg biofuel) 1.70 2.24 2.22 1.94

Table ES2b: Water use for each feedstock in gallons of water per kg of biofuel assuming 2% and
5% water consumption for cooling water (CW) and boiler feed water (BFW), respectively
Water Use Wild Algae OP Algae Wood Stover LA Albizia
Feedstock Production -3546 15.6
Pyrolysis
Cooling water (25-29°C) 1.94 1.62 2.26 2.50 2.70
Heated Water (60-75°C) 7.86E-03 5.89E-03 9.04E-03 1.01E-02 1.13E-02
Stabilization
Rinse Water 0.25 6.69E-02 2.46E-02 1.70E-02 3.08E-02
Spent Brine -0.49 -0.13 -4.89E-02 -3.36E-02 -6.11E-02
Water from 98% H2SO4 3.47E-04 9.44E-05 3.51E-05 2.35E-05 4.32E-05
Water from 50% NaOH 1.30E-02 3.56E-03 1.33E-03 8.99E-04 1.62E-03
Water from 10% NaCL 0.18 4.85E-02 1.79E-02 1.23E-02 2.24E-02
Upgrading
Cooling water 9.03E-02 8.20E-02 1.04 0.11 0.12
Boiler Feed Water 2.73E-02 2.47E-02 0.31 3.45E-02 3.70E-02
Wastewater -0.16 -0.13 -0.34 -0.41 -0.47
Integrated H2 Plant
Demineralized Water 4.31E-02 3.72E-02 4.49E-02 4.70E-02 5.30E-02
WWT using CHG
Cooling water Make-up 2.84E-04 2.36E-04 6.09E-04 7.32E-04 8.26E-04
Total (gallons/kg biofuel) -3544 17.2 3.32 2.29 2.46

Page 11
Conclusions
This is not the first LCA of pyrolysis-based hydrocarbon biofuels in the gasoline and diesel
range to appear in the literature, but this study evaluated a much wider range of biomass
feedstocks than in prior studies. In general, our GHG emission results are in agreement with the
literature for the few feedstocks that overlap with other investigations. The general conclusion
from our study is that pyrolysis-based biofuels can be produced from a wide range of biomass
feedstocks, including algae, agricultural residues, forest residues, and dedicated energy crops,
with waste products being the preferred feedstock from the greenhouse gas emission perspective.
With the exception of the wild algae, all other feedstocks (including open pond algae) exhibited
≥60% savings of GHG emissions compared to conventional fossil transportation fuels. As a
result of these analyses, pyrolysis-based biofuels are shown to be a promising category of
sustainable liquid “drop-in” transportation fuels which deserve further attention for research and
development. Commercialization of pyrolysis-based biofuels, with significant displacement of
petroleum-based liquid transportation fuels, using ubiquitous woody biomass sources in the
United States may lead to a significant reduction in greenhouse gas emissions.

Page 12
References

1. CARB, 2009, Detailed California-Modified GREET Pathway for Brazilian Sugar Cane
Ethanol
2. EIA, U.S. Energy Information Administration, Independent Statistics and Analysis,
”What Is U.S. Electricity Generation by Energy Source?”,
[Link] Web. 15 Apr. 2014.
3. EPA. 2010a. Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel
Standard Program Vol. 75 No. 58.
4. Morey RV, Kaliyan N, Tiffany DG, Schmidt DR. 2010. A Corn Stover Supply Logistics
System. Applied Engineering in Agriculture 26(3):455-461
5. Stratton RW, Wong HM, Hileman JI. 2010. Life Cycle Greenhouse Gas Emissions from
Alternative Jet Fuels: Partner Project 28 report Version 1.2.
6. Tenbruggencate J. Green Acres From green power timber to revenue-producing green
waste, Bill Cowern is harnessing global forces for local gain. September 2008]. Available
from: [Link]
7. U.S. Department of Energy. 2011. U.S. Billion-Ton Update Biomass Supply for a
Bioenergy and Bioproducts industry. R.D. Perlack and B.J. Stokes (Leads), ORNL/TM-
2011/224. Oak Ridge National Laboratory, Oak Ridge, TN. 227p.

Page 13