FINAL TECHNICAL REPORT JANUARY 26, 2026 Processing Municipal Solid Waste for Conversion to Jet Fuel FINAL TECHNICAL REPORT Processing Municipal Solid Waste for Conversion to Jet Fuel Principal Investigator: Bo Li, PhD, GTI Energy Other Key Personnel: Martin Linck, PhD, GTI Energy Jordan Klinger, PhD, Idaho National Laboratory Sheng Dai, PhD, Georgia Institute of Technology Robert Hallenbeck, MEng, Waste Management Nalin Kumar, PhD, NanoRanch/UHV Technologies JANUARY 26, 2026 Prepared by GTI ENERGY for US DEPARTMENT OF ENERGY OFFICE OF ENERGY EFFICIENCY AND RENEWABLE ENERGY BIOENERGY TECHNOLOGIES OFFICE under contract DE-EE0009265 PROJECT PERIOD: October 1, 2020 – September 30, 2025 GTI Energy Business Contact: GTI Energy Principal Investigator: Penny Dubernat, Contracts Manager Bo Li, Senior Project Manager pdubernat@gti.energy | 224-254-3262 Bli@gti.energy | 818-330-1492 TABLE OF CONTENTS EXECUTIVE SUMMARY .................................................................................................................................................. 1 INTRODUCTION............................................................................................................................................................... 3 Background ...................................................................................................................................... 3 Project objectives ........................................................................................................................... 4 Project team .................................................................................................................................... 4 METHODS .......................................................................................................................................................................... 6 Technical approach ........................................................................................................................ 6 Prepare first-pass feedstock .................................................................................................... 6 Characterize nMSW fractions ................................................................................................ 11 Adapt kickdown pump ........................................................................................................... 13 Develop and validate AI sorting algorithm........................................................................ 16 Prepare second-pass feedstock ............................................................................................ 22 Conduct technoeconomic analysis ...................................................................................... 25 Conduct life cycle analysis .................................................................................................... 30 Materials and Equipment ........................................................................................................... 33 Prepare first-pass feedstock ................................................................................................. 33 Characterize nMSW fractions ................................................................................................ 34 Adapt kickdown pump ........................................................................................................... 37 Develop and validate AI sorting algorithm........................................................................ 39 Prepare second-pass feedstock ............................................................................................ 40 RESULTS ........................................................................................................................................................................... 42 Determined composition of as-received nMSW ................................................................... 43 Characterized chemistry of nMSW fractions ......................................................................... 44 Characterized consolidation and permeability of feedstocks ........................................... 46 Achieved decontamination efficiency at 90% ....................................................................... 51 Achieved feedstock variation reduction and purity ............................................................ 52 Demonstrated injection at target U-GAS ® pressure ............................................................ 55 Wood chip testing ................................................................................................................... 56 nMSW testing ........................................................................................................................... 56 Fine powder testing ................................................................................................................ 58 Page ii Cycle time improvement ........................................................................................................ 58 Conducted technoeconomic analysis ..................................................................................... 59 Capital expenditures ............................................................................................................... 60 Operating expenses ................................................................................................................ 62 Levelized cost of jet fuel ........................................................................................................ 66 nMSW preprocessing cost ..................................................................................................... 67 Sensitivity analysis................................................................................................................... 69 Performed life cycle analyses ................................................................................................... 70 Advanced technology readiness level from 2 to 4 .............................................................. 76 DISCUSSION ................................................................................................................................................................... 77 Technoeconomic analysis .......................................................................................................... 77 Life cycle analysis ........................................................................................................................ 77 Recommendations ....................................................................................................................... 78 Conclusions ................................................................................................................................... 79 PRODUCTS AND PUBLICATIONS ........................................................................................................................... 81 GLOSSARY ....................................................................................................................................................................... 82 LIST OF TABLES AND FIGURES ................................................................................................................................ 84 TABLES ........................................................................................................................................... 84 FIGURES ......................................................................................................................................... 85 REFERENCES ................................................................................................................................................................... 87 APPENDIX A: Block Flow Diagrams and Stream Summaries ..................................................................... A-1 APPENDIX B: Simulate and Validate Hopper Flow ........................................................................................ B-1 APPENDIX C: AI-Sorter Scalability........................................................................................................................ C-1 APPENDIX D: Kickdown Pump Scalability ......................................................................................................... D-1 APPENDIX E: Technical Table .................................................................................................................................. E-1 Page iii This page is intentionally left blank. Page 1 EXECUTIVE SUMMARY GTI Energy develops cost-effective technologies for producing diverse power sources. Leveraging expertise in gas-to-liquids chemistry and strategic alliances, it recently partnered with a leading US waste management company to convert tons of municipal landfill into purified liquid fuel. This advancement, combined with processes customized for the aviation sector, offers a viable pathway to producing drop-in quality jet fuel from non-recyclable solid waste. Technoeconomic and life cycle analyses show this solution is on track to compete favorably with industry benchmarks. Non-recyclable municipal solid waste is a blend of materials such as plastics, fiber, rubber, glass, and heavily soiled items. The project team that transformed this underutilized matter into a new energy source undertook a research program with challenging targets set forth by the US Department of Energy. The Idaho National Laboratory and the Georgia Institute of Technology prepared and characterized vast amounts of waste material while nanoRanch/UHV Technologies developed a novel artificial intelligence sorting algorithm, and GTI Energy designed and built a customized kickdown pump for its U-GAS® conversion technology. Results surpassed expectations, with a decontamination efficiency of 90% using an AI sorting tool and a feedstock purity of 96% with >50% variation reduction. Additionally, the kickdown pump, a newly developed and key component of this technology, successfully injected the gasifier at the target pressure of 150 psig, advancing both tools to Technology Readiness Level 4, as determined by the US Department of Energy. Technoeconomic analysis shows fuel production costs are about 9% lower with the kickdown pump than with a conventional lock hopper feed system, with greater savings likely at scale. Life cycle analysis finds a 7–9% reduction in carbon intensity compared to traditional jet fuel. Further consideration of advanced carbon capture and sequestration reveals an improvement ranging from 125% to 133%, resulting in net-negative emissions. Following this successful proof-of-concept testing and laboratory validation, efforts now focus on progressing the AI sorting tool and kickdown pump to Technology Readiness Level 6 through pilot testing. The International Air Transport Association projected a $236 billion global fuel bill for airlines in 2025, underscoring the potential of waste-to-jet fuel technology as a cost-efficient energy solution. With patents pending for complementary processes in aviation fuel production using other feedstocks, GTI Energy is positioning itself as a leading entity in this specialized field. Page 2 This page is intentionally left blank. Page 3 INTRODUCTION Solutions which produce sustainable fuel and achieve economies of scale, such as innovative gas-to-liquids systems, are vital to today’s national and global interests. GTI Energy has been at the forefront of inventing such processes. Its proprietary technologies convert both organic and non-organic material into synthetic liquid fuel in ways that reduce capital costs and are scaled for competitive markets. The research performed and described within this report focuses on overcoming challenges associated with the conversion of large quantities of non-recyclable waste for the production of drop-in quality sustainable aviation fuel. Background In 2011, the Oak Ridge National Lab concluded that the United States has the potential to generate over one billion tons of sustainable biomass annually and serve as a source for renewable transportation fuels, chemicals, and power for the grid (1). It also suggested that utilizing waste streams could increase this capacity further once issues of feedstock supply, quality and processing were addressed. Consequently, the US Department of Energy (DOE) invested in high-impact research and development (R&D) aimed at producing energy from waste biogas, carbon dioxide, and plastics - endeavors often deemed too risky by the private sector. These initiatives specifically targeted the establishment of scientific and engineering collaborations capable of demonstrating high- performing drop-in biofuels and other renewable chemicals at a cost of $3 per gasoline gallon equivalent (GGE), with the additional objective of attaining a modeled mature biofuel minimum fuel selling price below $2.50 per GGE. A primary area of focus involved supporting the advancement of sophisticated sorting and preprocessing methods for non-recyclable municipal solid waste (nMSW). These processes would address material heterogeneity and variability and enable the production of high-purity feedstocks. Within this framework, integration with U-GAS® conversion technology - a commercially validated solution within the GTI Energy portfolio - was emphasized as critical for facilitating large-scale synthetic fuels production. The single-stage, fluidized-bed gasification process underlying U-GAS® operates with high efficiency and low emissions, converting diverse feedstocks into low-to-medium heating value Synthetic gas (syngas). This syngas serves as a foundational product for generating electricity, hydrogen, renewable natural gas, transportation biofuels (gaseous and liquid), and chemicals. Notably, this advanced gasification technology has been evaluated using more than 25 distinct feedstocks, including peat, pulp mill residues, and various waste materials. Specific elements critical to advancing this initiative included integration of an artificial intelligence (AI) sorting tool and implementation of a solids pump designed to provide efficient and dependable feeding technology for preprocessed nMSW. Such innovations would lay the foundation for the next steps towards commercialization. Page 4 Project objectives The primary goal of this project was to develop technology for converting nMSW into liquid aviation fuel with a focus on nMSW preprocessing, the steps to decontaminate and size the feedstock. The project aimed to establish cost-effective methods to turn nMSW into feedstock with 1) chemical and mechanical properties suitable for thermal oxidation in the U-GAS® gasifier, and 2) mechanical properties compatible with injection at the U-GAS® feed pressure of 150 psig using an adapted kickdown pump. Requirements included processing nMSW to a minimum of 95% purity with a 50% reduction in variation, achieving at least 80% decontamination efficiency, and maintaining preprocessing costs below $30 per dry ton. To achieve these objectives, the project was structured into three distinct task structures: 1. Review and approve the project plan and technical baseline. 2. Determine the mechanical and chemical material processing needed to meet conversion standards. Integrate an AI-based sorting approach to improve fractionation and decontamination. Evaluate pump components and analyze the effects on the injection system to assess the suitability of material fractions and blend designs. Develop a complete preprocessing system for use in subsequent technoeconomic and life cycle assessments. 3. Validate and implement a preprocessing system. This included validating a sensor-based AI sorting algorithm and processing various nMSW samples into feedstock for laboratory-scale testing with a kickdown pump. The pump is intended to demonstrate injection at commercially relevant pressures for gasification. Data on preprocessing, pump performance, and conversion costs will inform the technoeconomic and life cycle analyses. Project team Bo Li, PhD, serves as a senior project manager at GTI Energy and the principal investigator of this project. He brings over 24 years of experience in research and development, project management, and process and equipment engineering. He has overseen projects funded by the DOE and private sectors, advancing them from initial laboratory research to commercialization, construction, startup, and commissioning. His expertise spans gasification, hydrogen and decarbonization, process development and integration, technoeconomic and life cycle assessments, and process safety such as Hazard and Operability Study. As a certified Project Management Professional and an Aspen Plus® certified user, he is recognized for his leadership in DOE gasification programs, U-GAS® and R-GAS™ initiatives and demonstration, and his ability to build effective collaborations among cross-functional organizations. Martin Linck, PhD, is an institute engineer at GTI Energy specializing in renewable fuels and advanced energy technologies. With over 20 years of experience, he has transitioned lab research into commercially scalable solutions across energy, jet propulsion, and semiconductor Page 5 manufacturing. Dr. Linck holds 24 US patents (including three for catalyst and process innovations in fuel production) and more than 100 international patents. At GTI Energy, he manages DOE-funded projects and leads the development and implementation of IH2® technology for converting biomass into transportation fuels. Jordan Klinger, PhD, is a researcher at Idaho National Laboratory (INL) with over 15 years of experience in biomass and bioenergy systems. He specializes in feedstock preprocessing, thermal conversion, and flow characterization, using both experimental and computational methods to advance beneficiation and conversion technologies. He has authored numerous peer-reviewed articles with 950+ citations, received the INL Vision Award, several Mentor-of-the Year nominations, and has served as an adjunct assistant professor at Michigan Technological University. Sheng Dai, PhD, is an Assistant Professor at Georgia Institute of Technology with over 15 years of experience in granular materials testing, geomechanics, and computational modeling. He has led more than $4.7 million in federally funded research and developed advanced tools for pressure core characterization and particulate flow modeling. In addition to holding patents related to nano-dots dye and void detection systems, he serves on advisory boards for DOE and US Geological Survey programs and is an Associate Editor for the Journal of Geophysical Research: Solid Earth. He currently holds joint appointments across several schools at Georgia Institute of Technology. Robert Hallenbeck, MEng, is the senior manager of corporate development at Waste Management, Inc., where he leads technology scouting and new partnerships for innovative waste conversion and recycling technologies. During his tenure, he has evaluated the technical, financial, and strategic feasibility of more than 500 collaborations and established the Waste Management Technology Council, a cross-functional body of executives and subject matter experts guiding technology adoption and innovation. He is also the principal architect of “Spectrum Model,” a life cycle framework used to prioritize sustainability and investment decisions. Nalin Kumar, PhD, founded and served as president of NanoRanch/UHV Technologies, Inc. (nanoRANCH). With extensive background in microelectronics and advanced materials and instrumentation, Dr. Kumar obtained 65 issued US patents (10+ pending) and served as principal investigator for numerous federally funded programs, including the Advanced Research Projects Agency–Energy METALS project developing AI-driven XRF sorters for automotive scrap metal recycling. His inventions include nano-crystalline diamond field emitters and aligned carbon nanotube cathodes used in flat-panel displays, x-ray sources, and satellite thrusters. Page 6 METHODS The sections below describe the methodology adopted to preprocess nMSW and validate its suitability for conversion to jet fuel. The approach included staged size reduction, AI-assisted sorting to remove contaminants, physical and chemical characterization, and development of a pressurized feeding system, followed by technoeconomic and life cycle assessments to evaluate process feasibility and sustainability. Technical approach This section describes the processes used to preprocess nMSW and enable its conversion to jet fuel. It includes staged size reduction, AI-assisted contaminant removal, and the development of a novel kickdown pump. The approach also provided the basis for technoeconomic analysis (TEA) and life cycle analysis (LCA) to evaluate cost competitiveness and sustainability. Prepare first-pass feedstock A systematic procedure was developed for transforming raw nMSW into feedstock suitable for gasification (Figure 1). These steps are detailed in this section. The resulting first-pass samples were subsequently distributed to project teams for material characterization, pump component testing, and AI algorithm development. Raw Sorted Processed 1° size delivered manual reduction nMSW rejections 3” into supersacks “as-received” 2° size reduction on-specs, fines, overs Figure 1: Procedure for feedstock preprocessing INL received a full walking floor truck (approximately 10-15 tons) of nMSW from the Waste Management Materials Recovery Facility (MRF) in Salt Lake City, Utah. The material was delivered in bulk and stored in a covered bunker to prevent moisture ingress and contamination. Page 7 Upon receipt, the material was registered in the Biomass Feedstock Library and assigned a globally unique identifier for traceability. The material was covered in the storage bunker with a heavy-duty tarp, held in place with concrete bunker blocks (Figure 2). Figure 2: nMSW received in truck and stored in bunker, with closeup inset A telehandler with an approximate ¾ yard bucket (~0.57 m3) was used to move the nMSW from the bunker to the processing site. First-pass preprocessing involved manual sorting to remove hazardous and non-processable items, including biological waste (e.g. liquid filled containers, diapers, etc.), flammable materials (e.g. fireworks, lighters), metals, and glass, as shown in Figure 3. Each full bucket (scoop) of the telehandler was processed separately to record the variance through the nMSW pile. From each scoop, the removed reject material was weighed, and a Page 8 visual inspection was done to note some common and frequent items being removed. This step was critical to ensure safety and protect downstream equipment. Figure 3: Rejected nMSW The remaining material was subjected to primary size reduction using a Jordan Reduction Solutions (JRS) knife mill equipped with a 3-inch classification screen (Figure 4). The sorted material from the manual sorting step was conveyed manually to the mill and the sized reduced materials processed by the mill was collected in super sacks for intermediate storage. A random sampling between material scoops or supersacks was conducted from the supersack to estimate the quality of the material. Note that overall, it took approximately 9 telehandler scoops to fill 6 super sacks (~0.75m3) with shredded material, suggesting that the bulk density increased by approximately 12%. Figure 4: Supersacks after first-stage size reduction Page 9 Secondary size reduction was performed using 2, 4, and 6 mm heads (Crumbler® Rotary Shear by Forest Concepts) respectively (Figure 5). A magnetic separator was installed above the conveyor to remove residual ferrous contaminants. The feed rate was maintained at a slow speed to avoid clogging in the rotors. The detailed process is described below: the <3” sample was fed manually on a screw conveyer to 6 mm rotor set in a shredder. The sample between 6 mm and 3/32” screen (2.38 mm) screens was considered as 6 mm on-spec sample, while the sample smaller than 2.38 mm was considered as fines (reject) from this process. Subsequently, a portion of the 6 mm on-spec sample was fed manually on a screw conveyer and reduced in size through the 4 mm rotor set, while the sample between 4 mm and 3/32” screens was considered as 4 mm on-spec sample, and the sample under 2.38 mm was considered as fines. Similarly, another portion of the 6 mm on-spec sample was also reduced in size through 2 mm crumble, while the sample between 2 mm and 20 mesh (0.85 mm) screens was considered as 2 mm on-spec sample and 20 mesh was considered as fines. Oversized particles in above steps were recirculated through the crumbles system to improve yield. As the samples were processed, we randomly selected samples from each on spec and fine fractions for further chemical and physical analyses. The following definitions describe the nMSW samples referenced in this report: • 6 mm on-spec samples: preprocessed nMSW with particle sizes between 6 mm and 2.38 mm, obtained after secondary size reduction using the 6 mm head. • 6 mm fines: preprocessed nMSW with particle sizes smaller than 2.38 mm, obtained after secondary size reduction using the 6 mm head. • 4 mm on-spec samples: material obtained by further processing the 6 mm on-spec samples using the 4 mm head, with particle sizes between 6 mm and 2.38 mm. • 4 mm fines: material obtained by further processing the 6 mm on-spec samples using the 4 mm head, with particle sizes smaller than 2.38 mm. • 2 mm on-spec samples: material obtained by further processing the 6 mm on-spec samples using the 2 mm head, with particle sizes between 2 mm and 0.85 mm. • 2 mm fines: material obtained by further processing the 6 mm on-spec samples using the 2 mm head, with particle sizes smaller than 0.85 mm. Page 10 a) b) c) d) e) f) Figure 5: Photos of examined materials a) 6 mm on-spec, b) 6 mm fines, c) 4 mm on-spec, d) 4 mm fines, e) 2 mm on-spec, and f) 2 mm fines Page 11 An overall mass balance of this preprocessing is shown in Table 1 where the 6 mm sample ended up with the most fines (approximately 10%) which was expected as this was the first pass where the dirt and glass were separated. Both 4 and 2 mm samples showed similar mass balance where approximately 92% and 6% ended up with on-spec and fines, respectively. Note that we lost approximately 2%–3% sample during each step, either as oversized, remained inside the mill, or blown away with the wind. Table 1: Mass balance of preprocessed samples Out Sample In kg (wt%) mm kg (wt%) On-Spec Fines Loss 6 231 (100) 202 (87) 22 (10) 7 (3) 4 68 (100) 62 (92) 4 (6) 1 (2) 2 33 (100) 30 (92) 2 (6) 1 (2) The on-spec fraction was dried at 120–170°C to reduce moisture content. Homogenization was achieved using a custom-designed star splitter, which divided each drum into eight representative portions. These were distributed to project partners for characterization and testing. Moisture content, bulk density, and particle size distribution (PSD) were measured to establish baseline physical properties. This preprocessing workflow was designed to simulate industrial- scale operations while maintaining laboratory control and data fidelity. Characterize nMSW fractions Characterization of nMSW fractions was conducted using first-pass and second-pass samples to determine their physical, chemical, and flow properties relevant to thermochemical conversion using the first-pass samples. The characterization included analyses of particle size distribution, bulk density, and moisture content; chemical composition through proximate, ultimate, and heating value determinations; and consolidation and permeability behavior under simulated operating conditions. Physical properties The physical properties of the samples were determined through three main procedures: • Particle size distribution (PSD) was analyzed following the ASAE standard method S319.3. After the sample was sieved for 10 minutes, the sieves were weighed to calculate the cumulative particle passing distributions. • Bulk densities (loose and tapped) were measured using a modified ASAE standard method S269.4. The sample was poured into a cylindrical container, and the loose bulk density was calculated based on the mass and volume. Tapped bulk density was measured after dropping the container five times from a set height, then recalculating the density. Page 12 • Moisture content was determined by oven-drying the sample at 105°C for 24 hours. Chemistry Chemical characterization of nMSW fractions was conducted to evaluate their suitability for thermochemical conversion and to support predictive modeling of feedstock behavior. Samples were cryogenically ground to 2 mm using a Wiley mill with dry ice to prevent thermal degradation and volatilization of organics. The ground material was dried at 40 °C, then split using a riffle splitter to ensure homogeneity. Proximate analysis was carried out in a LECO Thermogravimetric Analyzer 701 (St. Joseph, Michigan) as per ASTM D 7582. Briefly, the sample was heated from room temperature to 107 °C at 6 °C/min heating ramp and held until a constant mass was reached. The mass loss during this time was considered as moisture content. The sample was further heated up to 950 °C at a ramp of 50 °C/min and kept isothermal for 9 minutes. During these steps, an inert atmosphere was maintained by flowing ultra-high purity (UHP) grade nitrogen at 10 L/min. The mass loss during this time was considered as volatile matter (VM). The sample was then cooled down to 600 °C and introduced oxygen (3.5 L/min) instead of nitrogen to allow the sample to be combusted. The temperature was ramped to 750 °C at 13 °C/min and held until a constant mass was reached for the sample. This remaining mass was considered as ash while the fixed carbon was determined by subtracting volatile matter and ash percentages from 100%. Ultimate analysis was carried out in an Elementar Vario EL cube (Ronkonhoma, New York) as per American Society for Testing and Materials (ASTM) standard method D 3176-09. The analyzer was calibrated with a certified Alfalfa standard. During a typical analysis, sample was combusted at 1150 °C in presence of UHP oxygen at a flowrate of 38 mL/min along with a carrier gas (UHP helium) at 230 mL/min. The combusted gas was analyzed using a thermal conductivity detector to determine the elemental carbon, hydrogen, and nitrogen. Oxygen content was calculated by the difference method which is a common method. Higher heating value (HHV) was determined as per ASTM method D 5865 using a LECO AC600 (St. Joseph, Michigan) isoperibolic system. In short, samples were combusted under 450 psi UHP grade oxygen and all the calculations were conducted based on the methods stated in D 5865. Each analysis was conducted in triplicate to ensure statistical robustness. The data were used to assess compositional variability across fractions and to inform blending strategies for feedstock optimization. These measurements also supported validation of AI-based sorting algorithms and numerical models developed by project partners. Permeability The consolidation and permeability of particulated nMSW are essential to feedstock handling. The way that nMSW particles pack together affects the jamming of these materials in various feedstock handling equipment (such as hoppers and screw feeders), as well as heat transfer, reaction uniformity, and overall efficiency during waste-to-energy conversion through gasification, pyrolysis, or incineration. Permeability controls how gases (air, oxygen, steam, Page 13 syngas) flow through the waste bed. Low permeability can cause channeling, incomplete combustion/gasification, or hot spots. Particularly for this project, consolidation and permeability in particulated nMSW feedstock govern the solid pump design to ensure continuous hopper flow and gas pressure induced feedstock densification in the solid pump (see Appendix B for Hopper Flow simulation and validation). After the customized permeability cell was assembled, tested nMSW materials were gradually poured into the permeability cell (i.e., air pluviation) up to 45.72 cm (18 in.) in height. The total mass of the sample was recorded to determine the bulk density and porosity of the tested samples. Then the specimen was mechanically loaded with a constant strain rate of 2.54 cm/min (1 in./min). At each desired height, gas permeability through the compressed specimen was measured. This loading-permeability measurement process is repeated until a maximum loading stress of 5.5 MPa is achieved. The maximum loading stress is chosen to encompass a broad spectrum of stress conditions encountered in multiple industry feedstock handling equipment. For gas permeability measurement, a pressure difference between the inlet and the outlet was created, and meanwhile, the gas flowmeter monitored the flux through the tested specimens. The gas permeability Kg for a specimen with a height H and cross-sectional area A is then calculated as: 2𝐻𝐻𝜇𝜇𝑔𝑔 𝑄𝑄𝑖𝑖𝑖𝑖 𝑃𝑃𝑖𝑖𝑖𝑖 𝐾𝐾𝑔𝑔 = 2 2 (1) 𝐴𝐴 𝑃𝑃𝑖𝑖𝑖𝑖 −𝑃𝑃𝑜𝑜𝑜𝑜𝑜𝑜 where µg = 1.8 × 10-5 Pa·s at 20 °C is the gas (air) viscosity; Qin (m3/s) is the gas flow rate at the inlet; Pin and Pout are the inlet and outlet pressures, respectively. Gas permeability in tested samples was primarily measured at an injection pressure of 70 kPa (10 psi). Additional permeability measurements were conducted using varying injection pressures from 10 to 150 kPa to explore the impact of injection pressure on measured gas permeability. Adapt kickdown pump To enable continuous, pressurized feeding of preprocessed nMSW into a pressurized gasification system (such as GTI Energy U-GAS®), a kickdown pump concept – adapted in place of the originally proposed dry solids pump due to cost constraints – was developed and tested. This technology was designed to overcome the limitations of conventional lock-hopper systems by providing a rapid, reliable, and cost-effective method for feedstock pressurization and delivery. The following section describes the design, operation, and advantages of the kickdown pump. The kickdown pump technology utilizes a unique two-stage ram assembly (Figure 6). The two- stage ram consists of two rams that are nested, one inside the other. By filling the interior volume of the primary cylinder with a pressurized medium, the primary ram can move itself and the secondary ram forward to a first position. The space inside the primary ram can then be energized (i.e. pressurized) and the secondary ram can then be moved to a second position. It should be noted that the figure shows a system where hydraulic or pneumatic pressure is used to actuate the rams, but the assembly could also be actuated via mechanical or even electrical means, if desired. Page 14 Figure 6: Two-stage telescoping ram The second-stage action of the telescoping ram then “kicks” the feedstock through the flapper valve, after the feedstock has been conditioned and made suitable for introduction to the pressurized feed hopper of the process. One major advantage of the kickdown pump technology is that there are no large vessels associated with this concept, nor are any large valves (with slow cycling times) required in order to make it work. Since a vacuum can be drawn on the feedstock in the conditioning chamber in the preferred embodiment of the technology in order to remove air and/or oxygen from the feedstock before it is sent into the process, no compressed inert gas is required (though purging with some type of inert gas may be preferred for certain applications). This will reduce the CAPEX dramatically. In addition, the receiving chamber can be brought up to near-process pressure with the same gas (i.e. H2 or CO2) that comprises the process atmosphere. Thus, no additional flow of any gas may be needed in order to operate the kickdown pump since the pump can potentially be used to introduce some or all of the process gas that would be needed in order to operate the process anyway (this is particularly true in the case of the GTI Energy IH2® process which requires that a stream of make-up hydrogen be supplied to the first-stage reactor in order to carry out the relevant process chemistry). The components of the kickdown pump, including the parts of the two-stage ram, are mechanically as simple as they can possibly be, meaning that minimal cost is associated with the design and construction of these components. As shown in Figure 7, a lab-scale kickdown pump was designed, built and tested through a GTI Energy funded project, where repeated test cycles were carried out without pressure in the receiving chamber on a manual operation basis. The kickdown pump integrated a series of nested pistons, a live seal, a flapper valve and a pressurized process chamber. The loading chamber is a 2 inch bore with length of 4.42 inches long. The pistons and pressurized regions are driven by a custom compressed air system with a supplementary storage tank included in the set up. Some of the key components of the pump were constructed with aluminum to meet Page 15 the fast-track schedule of the above project, so extra care is required to minimize erosion of these components during testing with solid materials. (P1 and P2) a) b) Figure 7: Kickdown pump – a) annotated 3D model; b) actual in laboratory Due to the simplicity and direct action of the piston mechanism, the kickdown pump is able to cycle in as little as about 15 seconds (and possibly much faster in certain applications), instead of the 15-20 minutes needed to cycle a commercial-scale biomass lock hopper system. As a result, the components of the pump might be more than 60 times smaller, while supplying the same amount of feedstock. This massive reduction in physical size should make it possible to realize a significant reduction in system CAPEX and OPEX. The potential to eliminate the need Page 16 for compressed inert gas may help to improve process economics and carbon intensities even further. Develop and validate AI sorting algorithm An AI sorting algorithm was developed and validated to enable accurate identification and separation of plastic fractions derived from nMSW. This effort aimed to improve feedstock quality for thermochemical conversion by removing chlorine-bearing and other contaminant materials prior to gasification. The work was conducted in two phases – algorithm development and system validation – progressing from controlled laboratory datasets to demonstrations using representative mixed plastic streams under controlled conditions. The following sections describe the development of machine learning models, feature optimization, and multi-sensor integration that collectively advanced the AI sorting capability. Development During AI sorting algorithm development phase, the first task focused on building classifiers from X-ray fluorescence (XRF) data collected by nanoRanch from representative plastic bottles sourced from general nMSW streams. Multiple categories of plastic bottles - including PET, HDPE, PVC, PP, and PS - were analyzed individually to capture the compositional variability typically observed in mixed nMSW plastics. All samples were scanned in their intact form, without grinding or preprocessing, to preserve surface characteristics and contamination patterns representative of real waste conditions. 1. To ensure model reliability, extensive data preprocessing was performed prior to training: Columns with ppm-level values (<0.1%) were removed as noise. Here Columns representing elemental or oxide concentration features with values below 0.1% (ppm- level) were removed as noise. 2. A reduced feature set of twelve elements/compounds (e.g., Na₂O, MgO, Al₂O₃, SiO₂, Cl, TiO₂, Fe₂O₃) was retained for analysis. During the course of algorithm development, multiple classical machine learning (ML) models were tested to determine the optimal approach for classifying and decontaminating nMSW fractions by using the preprocessed XRF datasets. Those models include Random Forest (RFC), CatBoost (CBC), XGBoost (XGB), Support Vector Machine (SVM), AdaBoost (ADB), and LightGBM (LGBM). 1. Random Forest Classifier (RFC): Useful for feature importance analysis but showed fluctuating results due to limited sample sizes. 2. CatBoost (CBC): Delivered the most consistent and stable classification results across multiple feature subsets. Served as the benchmark model during Task 6. 3. XGBoost (XGB) and LightGBM (LGBM): Provided high training accuracy but exhibited overfitting, with poor generalization to unseen samples. Page 17 4. Support Vector Machine (SVM): Achieved strong performance for some classes (100% accuracy for certain plastic fractions) but struggled with others. 5. AdaBoost (ADB): Underperformed overall, particularly in handling noise or imbalanced data. To establish chlorine sensitivity thresholds, additional experiments were conducted using PVC samples drawn from the same nMSW plastic set, with chlorine concentrations ranging from 0– 100%. The presence of Cl was observed only in the plastic subset of the dataset, as PVC - the primary chlorine-bearing polymer - was not present in other sample types. Under optimized XRF tube conditions (Omnian, 0.3 mA, 50 kV), the limit of detection for Cl was determined to be approximately 3%, defining the quantitative threshold below which elemental measurements become less reliable. However, the AI classifier relies on relative spectral features of Cl peaks rather than absolute concentration, allowing PVC identification even when chlorine levels fall below the quantitative limit of detection. This provided the spectral benchmark for downstream decontamination and PVC rejection studies. While the classical ML models developed in this phase (e.g., Random Forest, CatBoost, XGBoost) were not directly integrated into the sorter, they played a crucial role in establishing baseline classification accuracy, validating spectral separability among polymers, and identifying key elemental features (e.g., Cl, Si, Ti, Fe). The insights from these experiments guided the feature engineering and data preprocessing pipeline adopted in AI Sorting Algorithm Validation phase and informed the architecture selection for the deep learning phase. Thus, AI Sorting Algorithm Development phase provided the analytical groundwork that enabled successful multi-sensor AI integration in AI Sorting Algorithm Development phase. Table 2 summarizes the comparative performance of the ML models. CatBoost achieved the highest overall accuracy (~87 %), demonstrating the feasibility of AI-driven classification while also revealing the limits of classical ML methods in handling the complexity and variability of real-world waste streams. These findings directly motivated the transition to deep-learning approaches in Task 9 for scalable, multi-sensor deployment. Table 2: Performance summary for machine learning models Model % Accuracy Precision Recall Notes Random Forest 82 0.78 0.80 stable baseline; key feature selection CatBoost 87 0.84 0.86 best overall; high robustness to noise XGBoost 79 0.75 0.78 fluctuated due to small dataset size SVM 73 0.71 0.70 sensitive to scaling AdaBoost 77 0.74 0.76 moderate performance LightGBM 81 0.79 0.79 competitive baseline Page 18 Validation The implementation and validation of sorting algorithms task built on the foundations of AI sorting algorithm development, , extended the AI system to Deep Learning Models which were integrated on a multi-sensor platform combining XRF, NIR, and optical imaging and validated with lab testing. This phase marked the transition from feasibility studies to deployment on pilot-scale sorters at INL and UHV. The analysis of nMSW plastics relied on a multi-sensor approach, combining XRF, NIR, and optical imaging to characterize both the base materials and chlorine-bearing contaminants. XRF spectroscopy revealed unique elemental compositions for each material, enabling differentiation between plastics and identification of contaminants. Type 3 plastics exhibited the highest chlorine levels, serving as a clear marker for PVC (Figure 8). In parallel, a comprehensive high- resolution dataset of NIR signatures was collected using the INL sorter equipped with KUSTA LLA NIR 1.7 and NIR 2.2 spectrometers. These instruments captured high-resolution spectral fingerprints of PET, HDPE, PVC, PP, and PS plastics. Representative waveforms displayed distinct peaks and valleys that serve as reliable identifiers for classification (Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13). Figure 8: X-ray fluorescence of plastic materials Page 19 Figure 9: Type 1 plastic (PET) Figure 10: Type 2 plastic (HDPE) Page 20 Figure 11: Type 3 plastic (PVC) Figure 12: Type 5/7 polypropylene (PP) Page 21 Figure 13: Type 6 polysterene (PS) The integration of these datasets along with optical imaging allowed us to define key identification peaks unique to each plastic category. These spectral markers form the foundation of the classification algorithm, enabling real-time sorting. To support this work, an NIR data collection algorithm was implemented and validated. Demonstrations of the spectral acquisition are available: www.youtube.com/watch?v=F44MAmMJ4jA&feature=youtu.be www.youtube.com/watch?v=2KMIEJlaJUA&feature=youtu.be More than five different neural network models were evaluated, with the best-performing model selected for operational deployment on the “blue plastics sorter.” This provided the technical basis for subsequent validation of PVC removal efficiency. To fully leverage spectral and image data, more than five convolutional neural networks were introduced and evaluated. Among these, ResNet architectures (ResNet-18, ResNet-50) achieved the best results. Models were pretrained on ImageNet and fine-tuned on nMSW datasets, with data augmentation (rotation, scaling, cropping) applied to improve generalization. 1. ResNet-18 and ResNet-50 variants were tested on NIR and optical image datasets of plastic scrap bottles. 2. The ResNet models were trained using transfer learning with ImageNet-pretrained weights, then fine-tuned on nMSW-specific datasets. 3. Input data included both spectral signatures (NIR waveforms) and optical images of PET, HDPE, PP, and PVC plastics. 4. Data augmentation (rotation, scaling, random cropping, normalization) was applied to improve robustness and generalization. 5. ResNet consistently outperformed the classical ML classifiers. Page 22 6. In operational trials on the UHV “blue plastics sorter,” ResNet achieved ~90% identification and rejection efficiency for PVC, surpassing the milestone target of 80%. 7. The model demonstrated stable convergence and robustness across repeated test runs, unlike RFC and other tree-based methods that fluctuated depending on dataset splits. The ResNet model was selected and deployed on GPUs integrated with the UHV and INL sorters. Real-time inference pipelines were optimized using PyTorch with Compute Unified Device Architecture acceleration for fast throughput. The system was capable of classifying and ejecting PVC bottles in under 200 ms per item, meeting the operational requirement for high belt speeds (400–500 ft/min). This provided the technical basis for subsequent validation of PVC removal efficiency. ResNet models achieved ~90% PVC identification and rejection efficiency in operational trials on the UHV “blue plastics sorter,” surpassing the milestone target of 80%. This was a decisive improvement over classical ML methods, which plateaued below the required thresholds. Prepare second-pass feedstock The second-pass feedstock was prepared in a similar process to the first, with three main steps: 1) manual sort; 2) primary size reduction; 3) secondary size reduction as illustrated in Figure 14 schematic. The resulting second-pass feedstocks were distributed to project teams to support the characterization, pump testing, and TEA and LCA analyses. Figure 14: Block flow diagram of preprocessing steps Page 23 The primary size reduction of the received nMSW was conducted using a 2” classification screen in the VecoPlan mill. Prior to feeding the nMSW to the VecoPlan belt, a preliminary sorting was conducted to remove the hazardous material (biological such as liquid filled containers, diapers, etc. In addition to fireworks, lighters, flammable materials, etc.), metals, glasses, rocks, and bulky items for manual cutting before size reduction. From each supersack the removed rejected material was weighed, and a visual inspection was done to note some common and frequent items being removed. A random sampling during this processing was conducted to estimate the quality of the material. The primary size-reduced 2-inch sample was further reduced using a crumbler with 4 mm head. The sample was manually fed onto a conveyor belt, where extraneous metal parts were removed. For additional separation, a magnet was positioned over the conveyor belt. The feeding rate was maintained at a low rate to prevent clogging of the crumbler blades. The output from the crumbler was screened using a screen set with 3/8-inch mesh on top and 3/32- inch mesh on the bottom, as recommended by Forest Concepts, to achieve a nominal particle size of 4 mm. The oversized particles were recycled through the crumbler twice to increase the on-spec quantity. The 4 mm on-spec fraction was defined as material between 4 mm and 3/32" (2.38 mm), Mass balances were recorded for each processing stage to evaluate yield and efficiency. On- spec fractions were dried, homogenized using the star splitter, and characterized for PSD, bulk density, and chemical composition. Purity was assessed by quantifying non-structural ash content using ASTM D7582 and air classification techniques. Structural ash (e.g., Ca, Na, Mg) was distinguished from non-structural ash (e.g., glass, rock) based on physical properties and separation behavior. Figure 15 presents the mass balance of the overall process. Page 24 Figure 15: Mass balance of process Grinding energy consumption during both stages (VecoPlan and Crumbler) was recorded and calculated per dry ton of feed. Table 3 details the grinding energy required for each unit operation, including the energy consumed during the two recycling passes. Table 3: Grinding energy required for each unit operation Unit Operation Energy (kWh/dry ton) Vecoplan 99.7 ± 15.4 Crumbler 129.8 ± 5.8* *Includes energy required during the two recycling passes A cyclic heat treatment (Figure 16) was applied to stabilize the on-spec sample before shipping it to GTI and Georgia Tech (GT). Page 25 Figure 16: Cyclic heat treatment to stabilize nMSW sample The second-pass feedstock achieved >95% purity and >50% reduction in compositional variability, meeting or exceeding project targets. These results confirmed the effectiveness of the preprocessing strategy and validated the feedstock for downstream injection and conversion testing. Conduct technoeconomic analysis The project team conducted a technoeconomic analysis (TEA) for the integrated nMSW to jet fuel plant through U-GAS® pathway, using the second-pass feedstock as input to the gasification island. The nMSW to jet fuel process starts with a Plastics Decontamination section, where AI sorting tools are utilized to separate valuable plastics (mainly PVC) streams from the delivered nMSW material from a MRF facility; the sorted nMSW is then preprocessed into a feedstock suitable for the U-GAS® gasifier thermal conversion process, where it is converted into syngas; finally, the syngas is converted into jet fuel by the Fischer-Tropsch (FT) process. It is worth to note that a densification step is recommended and implemented for both cases – upstream of the kickdown pump or lock hopper feed system, respectively due to the low bulk density of the preprocessed nMSW. Without the densification step, neither feeding system would be practical. Two cases were evaluated – Case 1 employs an innovative kickdown pump technology to feed the preprocessed feedstock into the gasifier, while Case 2 utilizes the traditional lock hopper system to feed the preprocessed solid feedstock into the gasifier. The unique kickdown pumps in Case 1 feed the solid feedstock into the gasifier, which offer several advantages over the conventional lock hopper system in Case 2 such as lower capital cost, lower CO2 emissions due to elimination of the lock hopper, etc. The capacity of each case is 1,000 short tons per day (TPD) preprocessed nMSW feed to the U-GAS® gasifier, with the fuel properties listed in Table 4 below (the moisture content of the nMSW is 1.75 wt%), which are the average properties of the second-pass feedstock prepared by INL. This corresponds to 2,386 TPD raw nMSW (moisture content assumed to be 40% based on literature (2), (3)) feed to plant battery limit. Page 26 Table 4: Fuel properties of preprocessed gasifier feed Fuel Properties of Gasifier Feed using Ultimate Analysis (dry basis) Composition Weight % Carbon 44.35 Hydrogen 7.06 Oxygen 30.69 Nitrogen 0.28 Ash 17.62 High Heating Value (HHV) = 8,607 BTU/lb Note: please refer to Table 14 and Table 15 for more details. Process models were developed to simulate the overall system performance and quantify the key material and energy flows throughout the process. These models enabled the determination of critical inputs - such as feed rate, utility consumption (electricity, steam, water, and cooling requirements), and captured CO₂ rate - as well as key outputs, including nMSW rejects, product yield, composition, and jet fuel production rate. The resulting process data served as the technical foundation for subsequent capital expenditure (CAPEX) and operating expenditure (OPEX) estimations, which were then used to calculate the Levelized Cost of jet fuel (LOCJ) production. Process modeling Process models were configured to simulate the complete process flows and assess the performance of individual unit operations. A hybrid modeling framework combining Aspen Plus and Microsoft Excel modules was employed to capture the distinct characteristics of each process section, as illustrated in Figure 17. Specifically, an in-house proprietary Excel-based simulation tool developed by GTI Energy was utilized to model the U-GAS® gasifier performance, incorporating proprietary correlations and empirical data for predicting syngas composition, carbon conversion, and heat balance. The downstream sections - from partial oxidation (POX) through Fischer-Tropsch (FT) synthesis - were modeled using Aspen Plus, which enabled rigorous thermodynamic and mass-energy balance calculations. The subsequent FT synthetic crude oil (Syncrude) hydrocracking and upgrading steps were represented in Excel based on kinetics and product distribution data obtained from published literature sources (4). This integrated modeling approach ensured consistency between the different process stages and provided the necessary technical basis for subsequent techno-economic analysis. Page 27 Plastics Decontamination, Syngas nMSW POX Solid feed processing, U-GAS® Scrubber Syngas WGS AGR Compression Fischer- Hydrocracking Distillation Jet Fuel Tropsch Figure 17: Modeling approach for converting nMSW to jet fuel Appendix A shows block flow diagrams for Case 1 – kickdown pump and Case 2 – lock hopper configurations (Figure 36 and Figure 37, respectively). Each diagram outlines the process configuration, key unit operations, and major material streams. The corresponding material balance summaries, which include principal operating conditions, stream compositions, and flow rates, are also presented in Appendix A (Table 27 and Table 28). These figures and tables provide the process basis for comparing the two feed system configurations. Additional details on the overall nMSW-to-jet-fuel process are discussed in the following section. Process description The nMSW delivered to the plant from a MRF facility enters a plastics decontamination section where it is sorted by AI-assisted sorting equipment. Plastics such as polyvinyl chloride (PVC) that are highly valuable and also unsuitable for gasification are removed. The sorted feedstock is then processed in a solid feed preprocessing unit to remove non-convertible materials, such as dirt, sand, glass, metals, etc., which is returned to landfill. The preprocessed material is then crushed and densified into briquets before being fed into the gasifier through two kickdown pumps working in parallel in Case 1 (or a traditional lock hopper feed system in Case 2). The low bulk density of the nMSW (~80 – 90 kg/m3) is neither economical nor practical for kickdown pumps or lock hopper feed system. Low bulk density materials tend to bridge in lock hoppers which makes them less suitable for industrial processes; they also require bulky lock hoppers or more kickdown pumps for same gravimetric throughput. The project hence recommends and implements a densification step before feeding the nMSW to the kickdown pumps (or lock hopper feed system). Commercially available densification process such as biomass briquetting can densify the nMSW to a bulk density larger than 500 kg/m3, which is Page 28 used in the nMSW gravimetric throughput estimate. Accordingly, to deliver 1,000 tons per day (TPD) of densified nMSW, a kickdown pump with a volumetric capacity of approximately 1,820 m³/day is required, or alternatively, two pumps rated at 910 m³/day each. To ensure sufficient capacity and operational flexibility, the project team conservatively assumed a total volumetric capacity of 2,000 m³/day, achieved using two kickdown pumps rated at 1,000 m³/day each, for cost estimation and design extrapolation. The nMSW is gasified in a U-GAS® gasifier using oxygen as oxidant and superheated steam as the fluidization medium. The solid feed rate to the U-GAS® gasifier is 1,000 TPD. The gasification occurs at 145 – 150 psig (10 barg) and 1,562 °F (850 °C). The syngas existing the gasifier still contains some light hydrocarbons and higher hydrocarbons (tars), and hence, a partial oxidation (POX) reactor is used to convert them to CO and H2. In the proposed process, an R-GASTM style POX reactor is incorporated due to its robust design, compactness, high mixing efficiency, lower CAPEX and OPEX, where the syngas produced by the U-GAS® gasifier is fed to the top part of the POX reactor together with the required oxygen flow. Immediately downstream of the POX reactor, a syngas cooler (which is mechanically integrated to the POX) is utilized to recover the heat from the hot syngas while generating superheated steam. Portion of the steam is sent back to the U-GAS® gasifier; another portion of steam is added to syngas before the Water Gas Shift (WGS) reactor; the remaining steam is used to drive a turbine to generate power. The ash exits the gasifier at its bottom and is then cooled and collected for disposal. The syngas is then filtered by a candle filter and passes through a syngas scrubber to remove chlorides and carryover particulates. The syngas ratio (H2/CO) out of the scrubber is approximately 1.3, which is insufficient for Fischer-Tropsch (FT) synthesis. Therefore, part of the steam generated by the syngas cooler is added to adjust the H2O/CO ratio to optimize the water gas shift reaction. The syngas, after adding steam, is passed through a Water Gas Shift unit to shift part of CO to H2 so that the shifted syngas has a H2/CO ratio slightly above 2. The resulting syngas contains approximately 27 vol% CO₂, which must be removed prior to Fischer–Tropsch (FT) synthesis, as elevated CO₂ concentrations in the FT feed can adversely affect hydrocarbon product selectivity. CO₂ removal is carried out in a conventional Acid Gas Removal (AGR) unit. In Case 1, a portion of the separated CO₂ stream is recycled to the feed system as transport gas, which enters the gasifier together with the solid feedstock, while the remaining portion is routed to CO₂ sequestration. In Case 2, which employs a traditional lock- hopper feed system, an additional portion of the captured CO₂ is utilized for lock-hopper pressurization and depressurization operations. This step is unnecessary in Case 1, which uses a kickdown pump for direct pressurized feeding. Consequently, Case 1 results in a greater amount of CO₂ being sequestered than Case 2. The syngas is then passed through a Sulfur Removal Unit, which consists of LO-CAT and ZnO beds. Though sulfur is not considered in the composition, the average nMSW composition has about 0.34% sulfur content (3,400 ppmw) (5). The LO-CAT is a liquid redox system that uses a chelated iron system to reduce H2S to elemental sulfur. The sulfur content reduces to 10-20 Page 29 ppm. ZnO beds reduce the sulfur content to <1 ppmv, which is a requirement for the downstream Fischer-Tropsch catalysts. The syngas, which now contains 62% H2, 31% CO, and 4% CO2, is then compressed to 435 psig (30 barg) and fed to the FT synthesis unit. A single-pass CO conversion of 50% is used for the Fischer-Tropsch reactor (6). This conversion requires a recycle to make the overall conversion commercially feasible. A recycle compressor maintains flow in the recycle loop. The recycle ratio is limited by the amount of CO2 in the FT reactor combined feed stream, which is 10%. The products of the FT reactor are based on a cobalt (Co) catalyst for low-temperature Fischer- Tropsch (LTFT). The mass percentages in the product were taken from Klerk (4) and the model compounds used to simulate the FT product are shown in Table 5. Table 5: Fischer-Tropsch product composition for cobalt-LTFT systems Product fraction Model compounds Syncrude Mass % Tailgas CH4 5.6 C2H4 0.1 C2H6 1 Liquefied petroleum gas (LPG) C3H6 3.4 C3H8 1.8 Naphtha C7H14 7.8 C7H16 12 C7H15OH 0.2 Distillate C16H32 1.1 C16H34 20.8 Wax C28H58 44.6 Aqueous product C2H5OH 1.4 C2H5COOH 0.2 The FT reaction is exothermic; hence, steam is produced in the tubular FT reactor and is used to drive a turbine to generate power. The Syncrude product from the FT reactor is further hydrocracked and refined to produce jet fuel and motor gasoline as the main products. Klerk (4) discussed jet fuel refinery designs for FT Syncrude to maximize jet fuel yield. The LTFT jet fuel refinery design was chosen for this study. The refinery design consists of alkylation Page 30 oligomerization, C5-C6 hydro-isomerization, and Zeolite Socony Mobil-5 aromatization for gasoline production and has a hydrocracker and a hydrotreater primarily for jet fuel production. The estimated product yields (expressed as the volume of liquid fuel produced per ton of FT Syncrude processed) are given in Table 6. Table 6: Estimated product yields for FT Syncrude refining section Liquid Fuel Syncrude Product Yield (barrel/ton) Motor-gasoline 2.72 Jet fuel 5.16 With a gasifier feed rate of 1,000 tons per day (TPD), the integrated process produces approximately 20,769 gallons per day (gal/d) of motor gasoline and 39,380 gal/d of jet fuel. These outputs correspond to an overall liquid fuel yield of 61.2 gallons per dry ton of gasifier feedstock. This yield demonstrates the strong carbon utilization efficiency of the system and compares favorably with benchmark data from the National Renewable Energy Laboratory (NREL). Specifically, the NREL 2024 Annual Technology Baseline (7) reports a yield of 65.9 gallons per dry ton for a forest-residue-to-biofuel conversion pathway via Fischer–Tropsch (FT) synthesis. The similarity in yield indicates that the process modeled in this study performs at a level consistent with state-of-the-art biomass-to-liquids technologies, highlighting its technical competitiveness and potential scalability for commercial deployment. For the TEA, a discounted cash flow rate of return analysis (DCFROR) was used to estimate a minimum selling price for jet fuel. Most of the assumptions used in the TEA are from NETL Following completion of the process simulations, the project team developed capital expenditure (CAPEX) and operating expenditure (OPEX) estimates for both cases. Based on these estimates, the Levelized Cost of jet fuel was subsequently determined to assess the overall economic feasibility of each configuration. Conduct life cycle analysis Following the TEA, a preliminary life cycle analysis (LCA) was conducted to evaluate the carbon intensity of jet fuel produced from nMSW through the U-GAS® pathway. The LCA was developed using process material balance data generated from the process simulations described in the TEA section, in conjunction with background data from the GREET model, to quantify the greenhouse gas (GHG) emissions associated with the two process configurations. The well-to-wheels analysis presented in this report accounts for contributions from the upstream, facility, and downstream stages of the fuel production and utilization system. Process boundaries As illustrated in Figure 18, the process boundaries encompass the transportation and receipt of nMSW, which undergoes a plastics decontamination step followed by preprocessing prior to feeding into a U-GAS® gasifier for syngas production. After CO₂ removal, the syngas is Page 31 converted in a Fischer–Tropsch (FT) synthesis unit and subsequently separated into the final fuel products - jet fuel and gasoline. Two design cases were evaluated based on the solid feeding mechanism into the gasifier: Case 1 utilizing kickdown pumps while Case 2 employing a conventional lock hopper system. Polyvinyl chloride (PVC) is separated from the feed during the plastics decontamination step and recycled into valuable secondary products, while rejects from the preprocessing stage are returned to landfill. Although the process yields both jet fuel and gasoline, the Figure 18: Simplified procedure evaluated in life cycle analysis LCA treated gasoline as a co-product using the displacement method to account for its avoided emissions. Carbon balance As part of the LCA for the nMSW-to-jet-fuel pathway, a carbon balance was developed to ensure the completeness, consistency, and accuracy of the modeled system by accounting for all carbon inputs, outputs, and transformations within the defined system boundaries. The balance serves as a mass conservation check to verify that the total carbon entering the system - primarily from nMSW feedstock - is consistent with the total carbon leaving the system in the form of jet fuel, gasoline, process emissions, and solid residues. Establishing a well-closed carbon balance enhances data integrity, helps identify missing or inconsistent streams (e.g., untracked CO2 or carbon in reject fractions), distinguishes between biogenic and fossil-derived carbon, and ensures all relevant process steps - from waste preprocessing and gasification through Fischer-Tropsch synthesis and product upgrading - are accurately represented. Ultimately, maintaining a closed carbon balance strengthens the credibility of the greenhouse Page 32 gas (GHG) results and increases confidence in the calculated carbon intensity of the nMSW- derived jet fuel. The carbon balance for the process is summarized in Table 7 by categories: • Inputs: nMSW, electricity • Products: jet fuel, gasoline • Releases: Flared gases, CO2 vent from the feed system • Wastes: wastewater, nMSW reject streams, U-GAS® ash stream • Sequestration (captured CO2) Table 7: Carbon balances for each case Carbon Balance (kg/hr) Category Case 1 – Kickdown Pump Case 2 – Lock Hopper System Inputs 25,924 25,924 Products 6,137 6,137 Releases 2,714 3,186 Outputs Wastes 10,071 10,071 Sequestration 6,418 5,946 Imbalance -585 -585 Imbalance % -2.26 -2.26 The current LCA shows a carbon imbalance of -2.26% across the full system boundary, which is considered highly satisfactory for a complex, multi-step nMSW-to-jet-fuel model. This level of closure indicates that the majority of carbon flows have been accurately captured and that only minor deviations exist between total carbon inputs and outputs. For systems involving multiple unit operations such as waste sorting, gasification, Fischer-Tropsch synthesis, and product upgrading, an overall carbon imbalance within ±5% is generally regarded as excellent and well within acceptable uncertainty for preliminary LCA analysis. The small discrepancy observed here primarily reflects modeling simplifications, aggregated stream data, and the fact that some process components are scaled up from lower technology readiness level (TRL) systems rather than any significant data gap. These variances also reflect simplifying assumptions applied to selected material streams due to limited analytical data and the scope of available balances: • The carbon content of the sorted PVC stream was assigned based on the chemical composition of polyvinyl chloride (38.4% carbon, all non-biogenic). Page 33 • The carbon content of the reject stream returned to landfill was determined by performing a carbon balance around the preprocessing unit, resulting in a carbon-balanced nMSW preprocessing subsystem. • The downstream section (gasification, Fischer-Tropsch synthesis, and product upgrading) is not yet fully carbon balanced due to limitations in the fidelity of the process simulation. As a result, the overall system (nMSW decontamination and preprocessing, gasification, and FT synthesis) exhibits a residual 2.26% discrepancy between the total carbon entering as feedstocks and the total carbon exiting as products, emissions, and solid wastes. The fidelity of this LCA could be further improved by 1) replacing U.S. average data for nMSW and reject stream compositions with site-specific measurements/analyses, and 2) establishing closed carbon balances at the unit-operation level for all downstream process nodes. Materials and Equipment This section describes the materials and equipment used to preprocess nMSW for conversion to jet fuel. Preprocessing employed shredders, mills, and crumblers for staged size reduction. AI sorters integrating XRF, NIR, optical imaging and deep learning models were used for contaminant removal, supporting preparation of high-purity feedstock for downstream testing. A kickdown pump, mechanically improved and automated, was implemented for further testing to enable rapid, pressurized feeding into the gasifier. Prepare first-pass feedstock The preprocessing workflow utilized a suite of industrial and laboratory-scale equipment (Figure 19). A JRS knife mill with a 3-inch screen was used for primary size reduction. Secondary and tertiary size reduction were performed using a rotary-shear based size reduction technique (Forest Concepts, Crumbler) with rotary heads of nominally 6 mm, 4 mm, and 2 mm sizes. Conveyors were equipped with magnetic separators to remove ferrous contaminants. Material was collected in super sacks and 55-gallon drums for storage and transport. Page 34 Figure 19: Knife mill with conveyor belt pictured above; crumbler below Drying ovens (105–170 °C) were used to reduce moisture content prior to characterization. Bulk density was measured using ASAE S269.4 protocols, and PSD was determined using Ro-Tap sieve analysis per ASAE S319.3. The star splitter was custom-fabricated to divide samples into eight equal portions, ensuring statistical representativeness for downstream testing. Characterize nMSW fractions Physical properties Instrumentation used in the described methodological approach was as follows: • Model RX-29 Ro-Tap separator (for analyzing particle size distribution) • Sieve stack • Analytical balance (for weighing sieves and samples) • Cylindrical container (195 mm diameter) • Oven (capable of maintaining 105°C) • Hard surface (for tapping procedure) Page 35 Chemistry Analytical instrumentation included a LECO TGA701 for proximate analysis (moisture, volatile matter, ash, fixed carbon), an Elementar Vario EL Cube for ultimate analysis (C, H, N, O, S), and a LECO AC600 for higher heating value (HHV) determination. Cryogenic grinding was performed using a Thomas Wiley mill (Model 4) with dry ice to preserve sample integrity. Riffle splitters were used to ensure homogeneity. All analyses followed ASTM standards and were conducted in triplicate. Data were used to evaluate feedstock quality, support AI model training, and validate numerical simulations of feedstock behavior. Permeability Tested materials include five individual fractions (i.e., rigid plastic, cardboard, thin film, paper, and foam) and mixtures (i.e., four designed blends and the second-pass feedstock) preprocessed by INL, as shown in the following photos in Figure 20. Figure 20: Particulated nMSW fractions and blends evaluated The compositions of the mixture samples, i.e., four blends and the second-pass feedstock, are summarized in Table 8. The four blends were designed to investigate the compressibility and permeability of nMSW mixtures across varying compressibility characteristics. Blend 1 (90 % rigid plastic – 10 % foam) and Blend 2 (70 % rigid plastic − 30 % foam) are binary mixtures, using the least and the most compressible streams. Blend 3 (50 % rigid plastic – 30 % cardboard Page 36 – 20 % paper) is a ternary mixture that incorporates rigid plastic with similar material types (i.e., cardboard and paper) used to replace the foam found in Blends 1 and 2. Lastly, Blend 4 (21 % rigid plastic – 33.6 % cardboard – 18.8 % thin film – 22.7 % paper – 3.9 % foam) is a quinary mixture that includes all identified nMSW streams, proportioned according to the original waste stream. Table 8: Composition of tested blends and second-pass feedstock Material Thin Film Rigid Plastic Paper Cardboard Foam Blend #1 (wt%) 18.7 21.1 22.7 33.6 3.9 Blend #2 (wt%) 17.8 25 21.6 31.9 3.7 Blend #3 (wt%) 16.6 30 20.1 29.7 3.5 2nd-pass feedstock (wt%) 18.8 21 22.7 33.6 3.9 The test setup examines the permeability of the nMSW includes a gas supply system, sensors, a customized permeability cell, and a long-stroke mechanical loading system, as illustrated in Figure 21. The customized permeability cell is 61 cm in length and 10.2 cm in inner diameter, made of aluminum. Both the bottom flange and top piston are connected to gas pressure transducers (PX409), with the inlet gas flow measured by a flowmeter (FMA-1613A). Porous stones were placed at both ends of the tested specimens to have a uniform distribution of air pressure at both ends of the specimens. Page 37 Figure 21: Setup for consolidation and permeability testing (a) Schematic includes [1] gas source; [2] valve; [3] pressure regulator; [4] flow meter; [5] pressure transducers; [6] permeability cell; [7] mechanical loading system; [8] gasket; [9] porous stone; [10] nMSW specimen; [11] loading disc; [12] O-ring; [13] spacer (b) Main components labeled in photo Adapt kickdown pump Following approval from the DOE, the operation of the kickdown pump was automated by integrating solenoid valves controlled by a Programmable Logic Controller (PLC) system with programmed sequences. The original operation of the kickdown pump includes a series of manual switches to energize/deenergize solenoid valves for each pneumatic step of the process. To implement a system that could automate most of the procedural steps, a data acquisition cart and series of controllable solenoid valves were introduced to the system. An Allen Bradley PLC was added to control the solenoid valves to enable the automated operation of the kickdown pump. The kickdown pump, prior to automation, was previously shown in Figure 7. The automated kickdown pump used for testing is shown in Figure 22. Page 38 Figure 22: Automated kickdown pump for testing The testing procedure of the kickdown pump was crafted to simulate a semi-automated control sequence that could replicate the operation conditions of live testing within a pressurized process chamber. Prior to each test, the system was prepared using a control program linked to the pump PLC control panel via the lab computer and data acquisition cart. The procedure is listed below including the loading operations and manual overrides necessary for failure conditions and resetting the testing sequence. Kickdown pump operation sequence: 1) On the project computer, open the kickdown pump shortcut (automation program) 2) Connect the ethernet cord from the controller panel to the computer 3) In the program, verify that the computer is communicating with the controller. (All the components will be X’d out if not communicating) 4) Attach house nitrogen supply (150psig) to the cross inlet for low pressure lines 5) Ensure the Process chamber top is fastened properly for pressurization 6) Pressurize the process chamber to 150PSIG 7) Attach the high-pressure regulator to the compressed air cylinder, open the cylinder valve, and set the regulator to the required P2 pressure Page 39 8) After the compressed air cylinder has been set up, open the hand valve downstream of the buffer tank. 9) Verify all hand valves are open except the downstream buffer tank relief. 10) Verify all the regulators are set to the correct pressures 11) Load material into the hopper 12) Click “Initialize Sequence” on the computer (at this point the kickdown pump is ready for an automated cycle) 13) Click “Sequence Start” to run a single cycle of the kickdown pump (if the flapper valve does not reseal, P2 will automatically recycle in attempt to clear the flapper valve) 14) The P2 recycling parameters can be changed on the main screen of the program located in the top left corner 15) Use “Deactivate Sequence Return to Manual” to end the sequence and return to manual control. The key factor required for a successful test is the P2 recycling parameter seen in Step 14. The only intervention that testing engineers must perform is loading the material into the device and initiating the automation sequence to start again. The test that completes the 10-15 cycles is conducted by simply allowing the full automation process to occur with the flapper reseating and the device fully retracting to its starting position. Develop and validate AI sorting algorithm Deep learning models which were integrated on a multi-sensor platform combining XRF, NIR, and optical imaging installed at INL . Figure 23 illustrates the multi-modal sensing architecture of the UHV blue plastics sorter. This system integrates a visible-light vision camera, two NIR hyperspectral cameras, and an X-ray fluorescence detector in sequence to address the limitations of conventional optical sorting. Figure 24 provides a close-up view of the plastics collection bins from the UHV blue plastics sorter, highlighting the physical separation of classified plastic streams after multi-sensor inspection. Page 40 Figure 23: Multi-modal sensing architecture of the UHV blue plastics sorter Figure 24: Plastics collection bins from the UHV blue plastics sorter Prepare second-pass feedstock INL subsequently acquired a Vecoplan MSW shredder (Figure 25) to enhance second-pass sample preparation, offering improved durability and performance compared to the previously used knife mill. This device consists of a set of stationary shredding teeth, and an adjustable rotor that spins to reduce material size. This device is manufactured to accommodate a wide range of materials, and more appropriately withstand contaminants in waste streams like metal, rocks, etc. compared to the prior knife mill the team used. As a result, the Vecoplan shredder Page 41 was used in this second-pass sample preparation to perform the same function as the prior JRS knife mill fit with the 3-inch retention screen. Figure 25: MSW shredder Page 42 RESULTS The milestones that guided the results of this project are summarized in Table 9 and expanded upon later in this section. At the completion of each milestone, GTI Energy provided a report to the US Department of Energy to verify its activities. Corresponding main objectives and tasks are also referenced in this table. Notably, the project team exceeded fractionation and decontamination target metrics that had been defined prior to the project. Table 9: Summary of key milestones and results Milestones Results Metrics (costs: 2014 dollars) Determined composition of all objectives met Figure 26 as-received nMSW Characterized chemistry and Table 10, Table 11, Figure 29, all objectives met permeability of nMSW fractions Figure 30 Demonstrated sorting efficiency exceeded FOA target 90% efficiency at ≥80% Confirmed feedstock purity at ≥95% - 96% purity exceeded FOA targets with 50% reduced variability - >50% reduced variability custom automation for nMSW Adapted kickdown pump all objectives met testing Demonstrated injection at target all objectives met achieved by kickdown pump U-GAS® pressure: 150 psig *in 2014 dollars nMSW preprocessing cost ($*/dry ton): 48.60 without TEA: missed targets; revenue offsets; -46.95 with sensitivity analysis identifies Performed technoeconomic and life revenue offsets strategies for achievement cycle analyses for kickdown pump scenario jet fuel production cost LCA: exceeded fossil-based ($*/GGE): 4.60 targets jet fuel carbon intensity (kg CO2e/kg jet fuel): 3.56 without CCS; -1.26 with CCS Page 43 Milestones Results Metrics (costs: 2014 dollars) Advanced AI tool and pump DOE Technology Readiness TRL2 to TRL4 readiness for next steps Assessment Determined composition of as-received nMSW Fractionation of representative primary size-reduced samples revealed the heterogeneous nature of the as-received nMSW stream (Figure 26). The material was manually sorted into major categories: plastics, paper, woody biomass, metals, foam/cotton/napkins, unsorted residuals, and dirt/glass. Plastics constituted a significant portion of the mass, ranging from 7.6% to 39.3% depending on the sample location (top, middle, or bottom of the super sacks where the material was stored). Paper was the most abundant component, with values between 14.1% and 51.5%, reflecting the high prevalence of packaging and printed materials in residential waste streams. Woody materials and metals were present in smaller quantities, typically below 3%, while foam, textiles, and napkins accounted for less than 5%. Notably, the unsorted and dirt/glass fractions were substantial in some samples, reaching up to 40%, particularly in lower layers of the sacks, indicating stratification during storage and handling. “Unsorted” is listed as a separate category because the particles were too small to be manually separated into distinct fractions. This category primarily consists of thin plastics, rigid plastics, paper, cardboard, and foam/fabric. The reported average and standard deviation represent 15 measurements, providing a quantitative representation of variability within the stream. Page 44 Figure 26: Fractional analysis of nMSW Characterized chemistry of nMSW fractions The proximate analysis, higher heating values (HHV), and ultimate analysis of the as-received nMSW and its preprocessed fractions are presented in Table 10 and Table 11, respectively. Note: each of the analyses were conducted multiple times for each sample to minimize the error limit. The as-received nMSW exhibited high variability in composition, characteristic of heterogeneous municipal solid waste. On average, it contained 73.47 wt% volatile matter, 5.65 wt% fixed carbon, and 20.87 wt% ash, with an HHV of 21.47 ± 2.87 MJ/kg. The ultimate analysis showed 51.16 wt% carbon, 6.28 wt% hydrogen, 0.44 wt% nitrogen, 0.11 wt% sulfur, and 21.14 wt% oxygen (all on a dry basis). Despite the elevated ash levels, the relatively high HHV indicates that a substantial portion of the material possesses considerable energy potential, which can be effectively recovered through proper fractionation and decontamination. The large standard deviations observed in both proximate and ultimate analyses further confirm the significant compositional variability among samples. This heterogeneity reflects the uneven distribution of organic and inorganic constituents such as plastics, paper, food residues, dirt, and glass. Consequently, preprocessing and size reduction are essential steps to improve feedstock uniformity, reduce ash variability, and enhance the suitability of the material for gasification and other thermochemical conversion processes. Page 45 On the other hand, the on-spec samples generated through successive size reduction using the 6 mm, 4 mm, and 2 mm heads exhibited consistent composition and stable energy content across all size fractions. Proximate analysis of on-spec fractions showed minimal variation: volatile matter ranged from 69 to 75 wt%, fixed carbon from 6.8 to 8.0 wt%, and ash from 16.7 to 24.0 wt%, based on mean values. HHV remained consistent at 20.5 to 21.6 MJ/kg, indicating size reduction had little effect on energy potential. The ultimate analysis results were also comparable among the on-spec samples, with carbon contents between 47.9 and 52.1 wt%, hydrogen between 5.4 and 6.3 wt%, and oxygen between 21.2 and 25.3 wt% (dry basis). Furthermore, the reduced standard deviations across both proximate and ultimate analyses confirm that preprocessing improved the homogeneity and consistency of the feedstock. These findings indicate that the on-spec fractions achieved the desired compositional stability and quality suitable for downstream thermochemical conversion testing. Table 10: Proximate analyses and higher heating value Volatile Fixed Carbon Ash HHV nMSW (wt%) (wt%) (wt%) (MJ/kg) As-received 73.47 ± 9.6 5.65 ± 1.25 20.87 ± 10.27 21.47 ± 2.87 6mm on-spec 71.21 ± 4.34 7.67 ± 1.03 21.12 ± 4.94 20.5 ± 1.9 4mm on-spec 69.22 ± 2.72 6.82 ± 1.34 23.96 ± 3.77 20.5 ± 2.01 2mm on-spec 75.3 ± 2.25 7.96 ± 0.51 16.74 ± 1.84 21.6 ± 0.91 6mm fines 47.58 ± 8.34 4.51 ± 1.45 47.91 ± 9.34 14.23 ± 3.84 4mm fines 57.18 ± 4.1 6.35 ± 0.61 36.47 ± 4.37 16.98 ± 1.78 2mm fines 49.93 ± 9.3 6.31 ± 1.09 43.76 ± 10.37 12.6 ± 1.92 Note: nMSW samples and fractions are dry basis Table 11: Ultimate analyses C H N S O nMSW (wt%) (wt%) (wt%) (wt%) (wt%) As received 51.16 ± 7.14 6.28 ± 1.02 0.44 ± 0.17 0.11 ± 0.05 21.14 ± 7.57 6mm on-spec 50.42 ± 5.58 6.27 ± 0.89 0.49 ± 0.16 0.1 ± 0.05 21.16 ± 4.74 4mm on-spec 47.9 ± 3.32 6.12 ± 0.82 0.41 ± 0.11 0.12 ± 0.07 21.5 ± 4.03 2mm on-spec 52.14 ± 3.55 5.39 ± 0.97 0.44 ± 0.07 0.02 ± 0.04 25.27 ± 4.25 6mm fines 34.29 ± 9.39 3.98 ± 1.29 0.86 ± 0.26 0.2 ± 0.07 12.88 ± 5.08 4mm fines 37.49 ± 3.93 4.69 ± 1.03 0.65 ± 0.1 0.16 ± 0.09 20.54 ± 3.71 Page 46 C H N S O nMSW (wt%) (wt%) (wt%) (wt%) (wt%) 2mm fines 35.22 ± 3.14 4.49 ± 2.1 0.7 ± 0.03 0.16 ± 0.09 15.67 ± 7.57 Note: nMSW samples and fractions are dry basis It can be found that, even though following size reduction, the on-spec samples did not exhibit any notable changes in ultimate analysis. However, the fines showed pronounced differences. For example, the carbon content of the 2 mm fines decreased by approximately 30%, while the ash content more than doubled relative to the as-received material. During processing, the as-received material passed through multiple crumble heads, allowing dirt, glass, and other extrinsic inorganic components to fragment into finer particles and separate from the on-spec fraction. As a result, the fines exhibited substantially higher ash concentrations. In contrast, the ash content of the on-spec samples remained relatively stable because only about 6–10 wt% of the material was removed as the fine fraction. The lower standard deviation among replicate analyses further indicates that the preprocessed materials became more uniform and consistent in composition. Characterized consolidation and permeability of feedstocks Constant strain rate (CSR) consolidation Figure 27 shows the normalized consolidation curves for the first-pass feedstock with 3 sizes (i.e., 6, 4, and 2mm), where H=measured height and H0=starting height. Regardless of the starting height, all samples show a concave-downward shape approaching an asymptote with increasing stress (in logarithmic scale). Larger particle size in general requires less load to achieve the same degree of compaction. For example, the 6-mm sample requires ~10 psi stress to achieve a 2-inch final height from initially 6-in (H/H0 = 0.33), while the 2-mm needs approximately 50 psi. Page 47 6mm 4mm 2mm 1.2 1.2 1.2 A (12"-3") A (12"-3") 1 B (8"-2") 1 B (8"-2") 1 C (16"-4") C (16"-4") A (12"-3") D (6"-2") D (6"-2") 0.8 0.8 0.8 B (8"-2") E (18"-6") E (18"-6") C (16"-4") D (6"-2") H/H0 H/H0 H/H0 0.6 0.6 0.6 E (18"-6") 0.4 0.4 0.4 0.2 0.2 0.2 0 0 0 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 Stress [psi] Stress [psi] Stress [psi] Figure 27: CSR consolidation for first-pass feedstock Start and end heights in parentheses. Page 48 The compressibility results of the second-pass feedstock (4 mm) are summarized in Figure 28. Results are comparable to those of the blended samples and the first-pass feedstock. 20 60 18 16 50 14 40 12 H [in] ρb [pcf] 10 30 8 6 20 4 10 2 0 0 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 Stress [psi] Stress [psi] Figure 28: CSR consolidation for second-pass feedstock Gas permeability Figure 29 summarizes the measured gas permeability at 150 psig for the first-pass feedstock and the five fractions (all in 2mm). The results are presented in terms of gas permeability vs. material bulk density (left plot) and the bulk density vs. compression ratio (right plot). The results show a linear trend between logarithmic permeability versus tapped bulk densities and compression ratios of the tested specimens, i.e., gas permeability in these samples decreases exponentially with increasing bulk density (or compression ratio) due to increased consolidation load. This relationship allows for estimating permeability with known compression ratios of the sample. Page 49 Figure 29: Gas permeability at 150 psig Using mixed (original stream) and individual fractions of feedstock. Page 50 Gas permeability measured at 150 psig gas pressure in the second-pass feedstock (4mm) at different bulk densities (or applied compression loads) are summarized in Figure 30. The results are also comparable to those of three tested blends and the first-pass feedstock. 10 10 10 Kg @ 150 psi [Darcy] Kg @ 150 psi [Darcy] Kg @ 150 psi [Darcy] 1 1 1 0.1 0.1 0.1 0.01 0.01 0.01 0 2 4 6 8 0 20 40 60 0.01 0.1 1 10 100 1000 H0/H [-] ρbulk [pcf] Compression stress [psi] Figure 30: Gas permeability for second-pass feedstock Page 51 Achieved decontamination efficiency at 90% The AI sorting algorithm development phase demonstrated that AI-based classification of nMSW fractions was feasible. After preprocessing the XRF datasets to remove noise and selecting twelve key features, several machine learning algorithms were tested. CatBoost consistently gave the most stable performance, while Random Forest was valuable for feature importance analysis but showed fluctuations across runs. Other models such as XGBoost, SVM, and AdaBoost produced mixed results, with some achieving strong accuracy for certain classes but failing to generalize. These tests confirmed that chlorine could be detected at a limit of detection of approximately 3% using optimized XRF settings, but they also highlighted that classical machine learning methods alone could not reach the required sorting efficiency for real-world applications. During the AI sorting algorithm development phase, the algorithms were extended and integrated into multi-sensor platforms at INL and UHV. Near-infrared and optical imaging datasets were collected, and deep learning approaches were introduced to fully exploit the complexity of the spectral data. Among the models tested, ResNet architectures (ResNet-18 and ResNet-50) delivered the strongest performance. Trained on nMSW-specific datasets with data augmentation, the ResNet models proved to be robust and accurate when deployed on the UHV “blue plastics sorter.” As shown in Table 12, controlled trials were conducted with 200 mixed nMSW pieces, consisting of 20 PVC and 180 non-PVC items. The system successfully removed 18 PVC pieces and mistakenly removed 3 non-PVC pieces, leaving 2 PVC items undetected. This corresponds to a decontamination efficiency of 90% (surpassing the milestone target of ≥80%) and a stream purity of 99%, indicating that the AI sorting process was highly effective at separating PVC from non-PVC materials with minimal contamination. These results were further confirmed through demonstrations on the INL sorter, where PVC was reliably separated from PET, HDPE, and PP bottles using real-time inference. Table 12: Sorting data for removal of PVC PVC Non-PVC PVC Non-PVC PVC Decontamination Stream Scanned Scanned Removed Removed Missed Efficiency Purity (# pieces) (# pieces) (# pieces) (# pieces) (# pieces) (%) (%) 20 180 18 3 2 90 99 A demonstration video is available here: www.youtube.com/watch?v=Sa1pSiplgIk&feature=youtu.be At INL, the same framework was validated with additional fractionation runs, confirming reliable identification across multiple plastic types. The following video has this demonstration: Page 52 www.youtube.com/watch?v=Sa1pSiplgIk&feature=youtu.be Overall, the project not only demonstrated the feasibility of AI-driven fractionation and decontamination but also exceeded its performance milestone. The transition from traditional classifiers in AI sorting algorithm development phase to ResNet-based deep learning was decisive, with the system achieving the target sorting efficiency of 80% and reaching peak validated performance of approximately 90% during the same operational test campaign on the UHV blue plastics sorter. As a clear clarification, the reported sorting efficiencies of 80% and 90% refer to the same validation test campaign conducted on the UHV blue plastics sorter. Achieved feedstock variation reduction and purity Particle size distribution (PSD) and bulk density of the second-pass feedstock are shown in Table 13. The proximate analysis, higher heating value and ultimate analysis of the as-received (Vecoplan sample) and on-spec (4 mm) second-pass feedstock samples are shown in Table 14 and Table 15, respectively. The data are consistent with similar analyses, though some variation is expected. Table 13: PSD and density of on-spec sample PSD Bulk Density (mm) (kg/m 3 ) Sample D10 D50 D90 Loose Tapped 4 mm 119.50 ± 6.41 2.74 ± 0.04 4.40 ± 0.03 5.94 ± 0.02 86.76 ± 4.62 on-spec The 4 mm on-spec sample exhibited a narrow particle size distribution with D10 = 2.74 ± 0.04 mm, D50 = 4.40 ± 0.03 mm, and D90 = 5.94 ± 0.02 mm, indicating well-controlled size reduction and classification. The bulk density measurements were 86.76 ± 4.62 kg/m³ (loose) and 119.50 ± 6.41 kg/m³ (tapped), reflecting a moderately low packing density typical of preprocessed nMSW feedstock with irregular particle geometry and residual fibrous content. The consistent PSD and repeatable density values confirm that the size-reduction process produced a uniform on-spec material with predictable handling and flow characteristics, suitable for downstream feeding and reactor testing. Page 53 Table 14: Proximate analysis and HHV of second-pass samples Proximate Analysis, dry basis (wt. %) HHV, dry basis Sample Moisture Volatile Ash Fixed Carbon (MJ/Kg) As-received 1.73 ± 0.31 70.74 ± 10.36 20.5 ± 10.45 8.76 ± 1.98 19.30 ± 3.27 (Vecoplan) On-spec 1.75 ± 0.16 71.77 ± 3.17 17.61 ± 3.54 10.63 ± 0.71 20.02 ± 0.47 (4mm) Table 15: Ultimate analysis of second-pass samples Ultimate Analysis, dry basis, (wt. %) Sample H C N O As-received 7.66 ± 1.06 47.9 ± 6.04 0.43 ± 0.21 23.51 ± 6.83 (Vecoplan) On-spec 7.06 ± 1.09 44.35 ± 6.31 0.28 ± 0.06 30.69 ± 7.00 (4mm) The on-spec (4 mm) samples exhibited notable improvements in feedstock consistency and compositional stability compared to the as-received nMSW (Vecoplan). The average ash content decreased slightly from 20.5 ± 10.45 wt% to 17.6 ± 3.54 wt%, accompanied by a substantial reduction in variability, indicating effective removal of inorganic contaminants during size reduction and screening. The volatile matter and fixed carbon contents remained stable (≈71.8 wt% and 10.6 wt%, respectively), and the HHV showed minimal change - 19.3 ± 3.3 MJ/kg for the as-received material versus 20.0 ± 0.5 MJ/kg for the on-spec fraction - demonstrating that the preprocessing preserved the intrinsic energy potential of the fuel. The ultimate analysis also reflected greater uniformity, with the carbon content averaging 44.4 ± 6.3 wt% compared to 47.9 ± 6.0 wt% in the raw material but with narrower deviations across hydrogen, nitrogen, and oxygen values. Collectively, these results confirm that preprocessing improved feedstock homogeneity and predictability, producing a more stable and energetically consistent fuel suitable for gasification and other thermochemical conversion processes. Variation reduction >50% Variability reduction (VR) was quantified using the range of volatile matter and ash content before and after preprocessing. The variability deduction is calculated based on the following formula: ∅𝑀𝑀𝑀𝑀𝑀𝑀0 − ∅𝑀𝑀𝑀𝑀𝑀𝑀0 −1 𝑉𝑉. 𝑅𝑅. = 1 − � � ∅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 − ∅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 Where, ∅ is volatile matter (VM) or ash. Subscripts “0” and “F” represent the as-received and on- spec material, respectively. The following are the V.R. calculations: Page 54 Based on VM, 85.05−48.68 −1 𝑉𝑉. 𝑅𝑅. = 1 − � � = 61.7% 77.97−64.02 Based on ash, 44.70−11.09 −1 𝑉𝑉. 𝑅𝑅. = 1 − � � = 57.6% 26.13−11.88 Based on volatile matter, VR was calculated at 61.7%, and based on ash content, VR was 57.6%. Both values exceed the project target of 50%, demonstrating the effectiveness of the preprocessing and homogenization strategy in producing a chemically consistent and energetically favorable feedstock. The reduction in ash content and increase in carbon concentration directly improved the fuel quality of the feedstock. The observed variability reduction is particularly important for downstream injection and gasification processes, which require consistent feed properties to maintain operational stability and efficiency. These results validate the hypothesis that mechanical preprocessing, when combined with targeted fractionation and blending, can significantly improve the uniformity and performance of nMSW-derived feedstocks. Furthermore, the data provided critical input for AI-based sorting algorithms and numerical models used by project partners. Purity ≥95% Second-pass feedstock purity was assessed by quantifying the proportion of non-structural ash – primarily composed of glass, rock, and other inert contaminants – relative to total mass. Structural ash, which includes mineral constituents such as calcium, sodium, and magnesium, was not considered removable by mechanical means and was excluded from the purity calculation. Hence the feedstock purity is defined below, 𝑀𝑀𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑀𝑀𝑁𝑁𝑁𝑁𝑁𝑁 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = =1− 𝑀𝑀𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑀𝑀𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 ASTM 7582 was utilized to determine the total ash content for the on-spec samples, followed by separations of the structural and non-structural ash with application of cross-draft air classification method. The remaining fraction, primarily large aggregates glass as compared to the talc-like consistency of structural ash, was classified as non-structural ash. This procedure was repeated five times to minimize variability. Table 16 shows the structural and non-structural ash of on-spec material. Table 16: Structural and non-structural ash of on-spec material Ash Ash Mass Sample Sample (g) (%) Loaded Type ID Non- Net Net Net Non- (g) Total Structural Structural Total Structural Structural 1 1.437 0.2328 0.1187 0.1141 16.2 8.26 7.94 Page 55 2 1.647 0.1565 0.1197 0.0368 9.51 7.27 2.24 3 1.496 0.1840 0.1101 0.0739 12.3 7.36 4.94 4 mm 4 1.519 0.1565 0.1065 0.0500 10.3 7.01 3.29 on- spec 5 1.362 0.1049 0.0834 0.0215 7.71 6.13 1.58 1-5 – – – – – – 4.00 Avg The average non-structural ash content was determined to 4.0%, corresponding to a feedstock purity of 96%. This exceeds the target of 95% purity and confirms the success of the preprocessing and decontamination strategy. The results also demonstrated that the remaining 4.0% non-structural ash could potentially be further reduced through additional screening or density-based separation, if required by downstream conversion processes. These findings validate the preprocessing ability of the system to deliver a high-purity, low-variability feedstock suitable for injection into the U-GAS® gasifier and subsequent conversion to liquid fuels. Achieving 96% purity in a mechanically preprocessed nMSW stream is a significant accomplishment, particularly given the inherent heterogeneity of the input material. This level of purity ensures that the feedstock is suitable for high-efficiency gasification and minimizes the risk of fouling or slagging in downstream equipment. The ability to consistently meet or exceed the 95% purity threshold also supports the technoeconomic viability of the process by reducing the need for additional cleanup or conditioning steps. These results demonstrate that the preprocessing system is capable of producing a high-quality, conversion-ready feedstock from a challenging waste stream. Demonstrated injection at target U-GAS® pressure The project team conducted a series of tests to demonstrate the ability of the kickdown pump to inject solid feedstocks, including wood chips, processed nMSW, and pulverized powder, into a receiving chamber pressurized at 150 psig, representative of commercial scale U-GAS® gasifier operating conditions. The project successfully adapted, automated, and demonstrated a laboratory-scale kickdown pump prototype, advancing the technology from TRL 2 to TRL 4. Testing confirmed the ability to inject diverse solid feedstocks, including wood chips, preprocessed nMSW, and fine powders, into a chamber pressurized at 150 psig through repeated automated cycles. The system achieved cycle times of 12–15 seconds, representing more than a 60-fold improvement over Page 56 conventional lock hopper trains, while maintaining feedstock integrity and reliable flap valve reseating. Wood chip testing Prior to testing with the nMSW, commercially available Marth material was used to commission the kickdown pump, which is a homogeneous, woody biomass feedstock that is commonly found in industry. The average particle size of this material size is smaller than the nMSW, less than 4mm particle size. No additional processing was conducted on this material prior to kickdown pump testing. The purpose of this test is to calibrate pressures and troubleshooting any issues with the kickdown pump hardware and control logic. Kickdown pump operating procedures described above are followed during the testing and no issues found and we successfully demonstrated the injection of the wood chips into the receiving chamber pressurized at 150 psig during February 2025 with multiple testing cycles. Strong “kick” was observed during the testing. No changes to its physical form were observed after the kickdown pump testing, the optimal outcome for processing using the kickdown pump. nMSW testing Upon receiving the second-pass nMSW feedstock from INL (average particle size ~4 mm), the material was conditioned to reduce the risk of damage to the aluminum kickdown pump from severe erosion. Such damage typically results from fine, high-hardness particles – such as metal and glass – scoring internal surfaces or entering sensitive areas of the pump assembly. The conditioning process included: 1) manual inspection to remove glass or other hard objects – materials exceeding 0.25 inches were rejected; 2) passing a magnet across small batches to remove ferrous metals; 3) processing through a 3D-printed high-density separator to separate problematic high-density materials from less dense nMSW; and 4) removing top layers and loading them into corresponding sieves for 10 minutes using a Ro-Tap testing sieve shaker. Material collected from sieves 0.223 through 0.066 inches was mixed and placed into baggies at approximately 18 grams per bag. This weight was determined by testing different loading sizes in the kickdown pump – 18 grams produced the best results. A test conducted in April 2025 successfully completed 22 consecutive semi-automated cycles, where the GTI conditioned nMSW was successfully injected into the receiving chamber pressurized at 150 psig. The only manual operation was the operator loading the materials from the bags into the kickdown pump; all other steps were controlled by the PLC controller. Testing ended when the pressurized air cylinder – a 300-liter compressed air cylinder with a starting pressure of 1500 psig, used to drive the pistons – dropped below 400 psig. The run results were recorded in Table 17. Page 57 Table 17: nMSW testing data Test Cycle1 Cycle Time (s)2 Loaded Material Weight (g) 6 30 19.61 7 3 84 19.18 8 29 19.63 9 30 18.98 10 29 19.12 11 29 18.28 12 29 18.62 13 29 18.77 14 46 17.56 15 29 17.74 163 101 18.48 17 29 17.70 18 29 17.71 19 29 17.51 20 29 17.55 21 29 17.78 223 101 17.61 1. The cycle time was determined from the video recorded, which missed the first 5 cycles. 2. Material loading time (manual loading by operators, ~4 seconds) is not included in the cycle time. 3. Cycles with longer time were due to the flap valve did not reseat properly and the program automatically recycled the pistons until the flap valve was reseated. Further analysis of the data in Table 17 shows that the lab-scale kickdown pump exhibits stable mechanical performance during testing. The recorded video captured cycles beginning with Cycle 6, so the first five cycles were not included in the analysis. Across the recorded cycles, the average cycle time is 45.82 seconds, with most cycles occurring consistently at 29-30 seconds. A few longer cycles were observed at 46 seconds (Cycle 14) and 101 seconds (Cycles 16 and 22), which were attributed to instances where the flap valve did not reseat properly and the control program automatically recycled the pistons until reseating occurred. The reported cycle time does not include the ~4-second manual loading period. The loaded material weight per cycle ranged from 17.51 g to 19.63 g, resulting in an average of 18.32 g. After adjusting for an estimated 4% material spillage during manual loading, the effective mass delivered per cycle is 17.61 g. Based on the average cycle time of 45.82 seconds (calculated from the recorded cycles), the throughput of the lab-scale kickdown pump prototype is 1.353 kg/hr, or approximately 33.2 kg/day, for nMSW feedstock. Page 58 Because the pump displaces a fixed volume per stroke, throughput is strongly influenced by the bulk density of the feedstock. Multiple measurements of the processed nMSW samples indicated a bulk density of 86.76 kg/m³, which is consistent with the mass-per-cycle results and is a primary factor governing achievable throughput under fixed-cycle operation. The nMSW samples before and after the test (Figure 31) were closely examined and no identifiable physical changes were observed. Figure 31: nMSW before (a) and after (b) pump test Shows negligible size difference and compaction. Fine powder testing To broaden the application of the kickdown pump, the project team conducted pump testing with pulverized torrefied forest residual (D50 ~150 microns) since fine powder material is essential for entrained flow gasifiers such as R-GASTM gasifier. Similarly to the test with wood chips and nMSW, the kickdown pump successfully injected the powder into the receiving chamber pressurized at 150 psig. Cycle time improvement The capacity of the kickdown pump for a known material is reverse proportional to the cycle time. The project team conducted additional tests to further reduce the cycle time of the kickdown pump prototype. By optimizing the solenoid valve feedback loop, reducing the scan time of the PLC software (the time the PLC takes to read inputs, execute the program logic, update outputs, and perform internal diagnostics) and supplying sufficient pressure to the prototype, 12-seconds (excluding material loading time) cycle time was demonstrated with Page 59 multiple tests conducted without feedstock. The project team considered the tests are enough to demonstrate the capability of the kickdown pump prototype since there is no significant difference in cycle time with and without materials observed. Conducted technoeconomic analysis For the technoeconomic analysis (TEA), a discounted cash flow rate of return analysis (DCFROR) was used to estimate a minimum selling price for jet fuel. Most of the assumptions used in the TEA are from NETL studies (8) (9), and corresponding references are included. The assumptions used in the TEA are listed below (2014$): • nMSW tipping fee (10): $45/ton. • nMSW transport distance to gasification facility: 40 miles. • Moisture content of incoming nMSW assumed to be 40% based on literature (2) (3). Most of the moisture content is removed in solid feed preprocessing step; gasifier feed has 1.75 wt% moisture. • Operations labor cost: • 22 employees in gasification plant, $130k/y each. • Plastics decontamination unit: 1 Supervisor ($85k/y), 1 Forklift operator ($50k/y), 6 Loaders ($35k/y), and 1 Maintenance Engineer ($80k/y) per sorter machine; 36 sorters are needed. • Solid feed preprocessing unit: 78 employees, assuming 2 employees/machine. $35k/y each (same as Loaders in Plastics decontamination unit). • Investment spread across first 3 years: 10%, 60%, 30%. • Plant life: 30 years. • Year basis: 2014. • Discount rate, the interest rate used to calculate the present value of future cash flows from the project, is set at 8%. • Annual maintenance labor costs: 0.95% of total plant cost (8). • Annual maintenance material costs: 2.19% of total plant cost (8). • Annual administration and general overhead costs: 25% of total labor costs (8). • Insurance and local taxes: 2% of total plant cost (8). • Cost of electricity: $40/MWh. • Capacity factor: 90%. • Ash disposal cost (10): $45/ton. • Selling price of motor-gasoline (11): $3.36/gal. • Gasoline heating value (12): 125,000 BTU/gal. • Jet fuel heating value (12): 135,000 BTU/gal. • 1 gal jet fuel = 1.08 GGE (gasoline gallon equivalent). Page 60 • Cost escalation from bare erected costs was based on GTI Energy internal database of cost factors and prior project experience. • Total overnight costs (TOC) (9): • Pre-production/start-up costs: 6 months operating labor, 1-month maintenance materials, 1-month waste disposal, 2% of total plant cost. • Inventory capital: 0.5% of total plant cost for spare parts. • Land: $3,000/acre, assumed for 100 acres. • Financing cost: 2.7% of total plant cost. • Other owner costs: 15% of total plant cost. • Total As Spent Costs (TASC) (9) = 1.093*TOC. • The U-GAS®, Pox, WGS, and AGR units were costed from equipment cost upwards. • The Fischer-Tropsch section, Refining and Upgrading, Air Separation Unit and Power Block were costed based on correlations from literature (4) and internal GTI Energy estimates. • The Fischer-Tropsch and refining sections were not modeled rigorously, but their capital costs and operating costs have been accounted for. • Off-sites/OSBL costs: Added 15% additional capital costs to account for off-sites/ outside battery limit costs. OSBL costs may vary from 10% to as high as 40% (13) (14). Since some parts of OSBL, such as the air separation unit, turbines, are included in the cost estimation, 15% was used to estimate OSBL costs. Capital expenditures The captial expenditures (CAPEX) for both cases was determined using a combination of detailed scale-up analyses for key subsystems (e.g., AI-based sorter and kickdown pump, see Appendix C and D), scaling estimates derived from commercial technologies (e.g., U-GAS® and POX units) offered by GTI Energy and its subsidiaries, and additional scale-up correlations and cost factors obtained from published references for major process units such as acid gas removal (AGR) and Fischer–Tropsch (FT) synthesis. The CAPEX for Case 1 and Case 2 are shown in Table 18 and Table 19, respectively. Table 18: CAPEX for kickdown pump case (Case 1) CAPEX – Kickdown Pump Case Process Unit $M (2014$) % of TPC Solid preprocessing and feeding (including plastics decontamination, preprocessing and kickdown 95.9 18 pumps) Gasification (U-GAS®, POX) 67.1 13 AGR + syngas clean-up + CO2 compression 132.1 25 Fischer-Tropsch 66.0 13 Page 61 CAPEX – Kickdown Pump Case FT product upgrading and refining 35.0 7 Air separation unit (ASU) 37.8 7 Power block 20.8 4 Off-sites/OSBL 65.1 13 Total Plant Cost (TPC) 519.8 Financing and start-up costs 169.2 Total As-Spent Cost (TASC) 689.1 Table 19: CAPEX for lock hopper case (Case 2) CAPEX – Lock Hopper Case Process Unit $M (2014$) % of TPC Solid preprocessing and feeding (Including plastics 121.3 22 decontamination, preprocessing and lock hopper feeding) Gasification (U-GAS®, POX) 67.1 12 AGR + syngas clean-up + CO2 compression 133.7 24 Fischer-Tropsch 66.0 12 FT product upgrading and refining 35.0 6 Air separation unit (ASU) 37.8 7 Power block 20.8 4 Off-sites/OSBL 72.3 13 Total Plant Cost (TPC) 554.1 Financing and start-up costs 180.0 Total As-Spent Cost (TASC) 734.1 For the kickdown pump configuration (Case 1), the total plant cost (TPC) for the solid feed preprocessing section decreases from $121 million in Case 2 to $96 million, representing a reduction of approximately 21%. This cost advantage reflects the benefit of implementing the kickdown pump system, which eliminates the need for high-pressure lock hoppers and reduces the size and complexity of feed hoppers required in the traditional configuration. In addition to equipment cost savings, the kickdown pump design simplifies the overall feed system layout by reducing the number of pressure vessels, valves, and associated control instrumentation. This simplification not only lowers the capital cost of major components but also decreases installation, piping, and structural steel costs, which are typically significant contributors to total installed cost. Page 62 On a plant-wide basis, the TPC for Case 1 is $519.8 million, representing a 6.2% overall cost reduction compared to Case 2 ($554.1 million). While the relative savings in the feed preprocessing section are more pronounced, the impact at the total-plant level remains meaningful, reflecting the cumulative benefit of system simplification, reduced mechanical footprint, and lower auxiliary equipment requirements. These cost reductions demonstrate the potential economic advantage of the kickdown pump configuration when scaled to commercial capacity and contribute directly to lowering the overall levelized cost of jet fuel. Operating expenses The operating expenses (OPEX) and costs associated with the plant include the following: • Administrative and operating labor • Maintenance – material and labor • Consumables • Fuel/feedstock • Taxes and insurance The operating costs of the plant are categorized into fixed operating costs and variable operating costs. Fixed operating costs are those that remain constant regardless of the level of jet fuel production, such as maintenance labor, supervision, insurance, and administrative overhead. In contrast, variable operating costs are directly proportional to the quantity of jet fuel produced and include expenses such as feedstock, utilities, catalysts, and consumables. Quantities of major consumables – such as feedstock, process water, and utilities – were derived from the heat and material balance. These data ensure consistency between the process model and the economic evaluation. Operating labor requirements were estimated based on the number of operators needed per shift, accounting for plant coverage, supervision, and support personnel. The corresponding labor costs were determined using standard wage rates and industry norms applicable to facilities of similar scale and complexity. The variable and fixed OPEX for Case 1 are shown in Table 20 and Table 21, respectively. The variable and fixed OPEX for Case 2 are shown in Table 22 and Table 23, respectively. Table 20. Variable operating costs for kickdown pump case (Case 1) Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream nMSW Feed and Transport nMSW tipping fee 90,204 kg/h (35.3) Tipping fee, $45/ton Freight rates (15) based on nMSW transport cost 90,204 kg/h 4.7 $0.15/ton-mile Page 63 Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream Solid Preprocessing and Feeding (Including plastics decontamination, preprocessing and kickdown pumps) Power requirement for plastics AI-based plastics 90,204 kg/h 1.4 decontamination process, mainly decontamination compressed air Revenue from PVC Based on $0.4/lb selling price of PVC 2,706 kg/h (19.6) co-product (16) Power requirement for grinders, 243 nMSW preprocessing 52,499 kg/h 4.2 kWh/MT dry feed Waste disposal for 24,477 kg/h 9.6 Based on $45/ton disposal fee reject material Assumed $19.05/ton for briquette Densification 37,799 kg/h 6.3 production (17) Kickdown pump 600 kW power requirement for 0.6 MW 0.2 operation operation of the kickdown pump Gasification (U-GAS®, POX) Revenue from Based on $5.6/ton net revenue from 7,461 kg/h (0.5) gasification ash selling processed ash (18) Air Separation Unit (ASU) Oxygen (for both U- 0.4244 kWh/kg energy requirement 17,820 kg/h 2.4 GAS® and Pox) based on internal ASU load calcs AGR + Syngas Clean-Up + CO2 Compression Power requirement 31,441 kg/h 0.1 Based on IEA GHG, 2017 (19) for AGR Scaled based on amount of sulfur LO-CAT Operational 88 kg S/h 0.2 removed, $199.54/ton sulfur removed (20) Estimated based on stoichiometry of ZnO bed kg reaction, ZnO + H₂S → ZnS + H₂O 3 0.05 replenishment ZnO/h and sulfur content. Input sulfur content : 15 ppm, Outlet : <1 ppm kg Spent ZnO disposal costs based on ZnO disposal costs 3 0.001 ZnO/h tipping fee CO2 compressor for 0.6 MW 0.2 U-GAS® Page 64 Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream CO2 compressor for 3.4 MW 1.1 CCS Fischer-Tropsch Syngas compressor 3.6 MW 1.1 Recycle compressor 0.3 MW 0.1 FT Product Upgrading and Refining Power requirement Based on Unicracking process for for FT Refining 0.5 MW 0.1 Hydrocracking and TIP Isomerization sections process (21) Revenue from Based on selling price of gasoline, 865.4 gal/h (22.9) gasoline sales $3.36/gal Power Block Steam turbine -13.8 MW (4.4) Total (51.0) Table 21: Fixed operating costs for kickdown pump case (Case 1) Fixed OPEX Estimation (2014$) Operating expense $M/Year Notes Considers labor for preprocessing plants as well as Labor 16.2 process facilities Admin and general 5.3 25% of Total Labor Costs overhead Insurance and local taxes 10.4 2% of Total Plant Cost Maintenance 16.3 3.14% of Total Plant Cost Total 48.2 Table 22: Variable operating costs for lock hopper case (Case 2) Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream nMSW Feed and Transport Page 65 Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream nMSW tipping fee 90,204 kg/h (35.3) Tipping fee, $45/ton Freight rates based on $0.15/ton- nMSW transport cost 90,204 kg/h 4.7 mile Solid PreProcessing and Feeding (Including plastics decontamination, preprocessing and lock hopper feeding) Power requirement for plastics AI-based plastics 90,204 kg/h 1.4 decontamination process, mainly decontamination compressed air Revenue from PVC Based on $0.4/lb selling price of PVC 2,706 kg/h (19.6) co-product (16) Power requirement for grinders, 243 nMSW preprocessing 52,499 kg/h 4.2 kWh/MT dry feed Waste disposal for 24,477 kg/h 9.6 Based on $45/ton disposal fee reject material Assumed $19.05/ton for briquet Densification 37,799 kg/h 6.3 production (17) Gasification (U-GAS®, POX) Revenue from Based on $5.6/ton net revenue from 7,461 kg/h (0.5) gasification ash selling processed ash (18) Air Separation Unit (ASU) Oxygen (for both U- 0.4244 kWh/kg energy requirement 17,820 kg/h 2.4 GAS® and Pox) based on internal ASU load calcs AGR + Syngas Clean-Up + CO2 Compression Power requirement 31,441 kg/h 0.1 Based on IEA GHG, 2017 (19) for AGR Scaled based on amount of sulfur LO-CAT operational 88 kg S/h 0.2 removed, $199.54/ton sulfur removed (20) Estimated based on stoichiometry of ZnO bed kg reaction, ZnO + H₂S → ZnS + H₂O 3 0.05 replenishment ZnO/h and sulfur content. Input sulfur content : 15 ppm, Outlet : <1 ppm kg Spent ZnO disposal costs based on ZnO disposal costs 3 0.001 ZnO/h tipping fee Page 66 Variable OPEX Estimation (2014$) Scaling Operating expense Units $M/Year Notes stream CO2 compressor for 0.7 MW 0.2 U-GAS® CO2 Compressor for 3.2 MW 1.0 CCS Fischer-Tropsch Syngas compressor 3.6 MW 1.1 Recycle compressor 0.3 MW 0.1 FT Product Upgrading and Refining Power requirement Based on Unicracking process for for FT Refining 0.5 MW 0.1 Hydrocracking and TIP Isomerization sections process (21) Revenue from Based on selling price of gasoline, 865.4 gal/h (22.9) gasoline sales $3.36/gal Power Block Steam turbine -13.8 MW (4.4) Total (51.3) Table 23: Fixed operating costs for lock hopper case (Case 2) Fixed OPEX Estimation (2014$) Operating expense $M/Year Notes Considers labor for preprocessing plants as well as process Labor 16.2 facilities Admin and general 5.4 25% of Total Labor Costs overhead Insurance and local taxes 11.1 2% of Total Plant Cost Maintenance 17.4 3.14% of Total Plant Cost Total 50.1 Levelized cost of jet fuel A discounted cash flow rate of return (DCFROR) analysis was performed to estimate the levelized cost of jet fuel produced from nMSW via the U-GAS® gasification pathway. The analysis incorporated all capital and operating expenditures, feedstock handling costs, and revenues over the project lifetime, discounted to 2014 value to account for the time value of money. Page 67 The results indicate that the traditional lock hopper configuration (Case 2) yields a levelized cost of $5.03 per GGE ($5.43 per gallon of jet fuel), while the kickdown pump configuration (Case 1) achieves a lower cost of $4.60 per GGE ($4.97 per gallon of jet fuel). This represents an approximate 9% reduction in fuel production cost for the kickdown pump configuration compared with the conventional lock hopper system, highlighting the economic advantage of adopting the kickdown pump technology for large-scale deployment. All costs are expressed in 2014 U.S. dollars (2014$). nMSW preprocessing cost A comprehensive annualized cost analysis was performed only for the plastics decontamination and solid feed preprocessing systems to determine the overall nMSW preprocessing cost, which is already included in the levelized cost of jet fuel (LCOJ) estimate. For each unit operation, detailed estimates of capital investment, operating expenses, maintenance, power consumption, and labor requirements were compiled based on vendor-provided data and equipment specifications. These cost elements were then annualized to represent the typical yearly expenditures throughout the expected operating life of the equipment. The resulting total annualized cost for each system was subsequently normalized by the total solid mass flow rate processed to obtain unit costs expressed on both a per-metric-ton and per- short-ton basis. Based on the available data, the cost estimation of the two systems differed slightly: the plastics decontamination cost was first developed in 2024 dollars and then deflated to 2014 dollars based on Chemical Engineering Plant Cost Index (CEPCI) index, whereas the solid feed preprocessing cost was estimated directly in 2014-dollar terms. The results for these two systems are summarized in Table 24 and Table 25, respectively. It is important to note that the tipping fee was not included in the nMSW preprocessing cost calculation; rather, it was accounted for separately on a plant-wide basis within the overall techno-economic analysis. Table 24: Pre-processing cost for plastics decontamination unit Decontamination raw nMSW, metric Capacity 711,167 ton per year Total capital cost $22,672,563 vendor estimate Annualized CAPEX and OPEX $/Year Notes Total labor cost 10,615,000 vendor estimate Cost of capital (7% of total capital cost) 1,587,079 vendor estimate Repairs (10% of total capital cost) $2,267,256 vendor estimate Supplies (10% of total capital cost) $2,267,256 vendor estimate Page 68 power for AI sorting OPEX for power $1,422,333 equipment Total annualized cost $18,158,92 Total annualized cost/metric ton $25.53 (raw nMSW processed) Total annualized cost/metric ton $18.48 (raw nMSW processed) Total Annualized Cost/ton $16.76 (raw nMSW processed) Total Annualized Cost/dry ton $27.93 (raw nMSW processed, 2014 dollar basis) 1. Moisture content assumed to be 40% based on literature. 2. Capacity factor of 90% applied. 3. All costs in 2024 dollar basis unless noted. Table 25: Preprocessing costs for solid feed unit Solid Feed Pre-Processing Sorted nMSW, metric Capacity 689,832 ton per year*/** Total Capital Cost $14,790,599 Vendor estimate Annualized CAPEX and OPEX $/Year Notes Total Labor Cost $2,730,000 Power requirement OPEX for power $4,188,333 for grinders, 0.253 kWh/kg Cost of Capital (7% of Total Capital Cost) $1,035,342 Vendor estimate Maintenance (10% of Total Capital Cost) $1,479,060 Total Annualized Cost $9,432,735 Total Annualized Cost/metric ton $13.67 (sorted nMSW processed) Total Annualized Cost/ton $12.40 (sorted nMSW processed) Total Annualized Cost/dry ton $20.67 (sorted nMSW processed) 1. Moisture content assumed to be 40% based on literature. 2. Capacity factor of 90% applied. 3. All costs in 2014 dollar basis. Page 69 The combined cost of preprocessing, which includes both fractionation and decontamination, is estimated at $48.60 per ton of nMSW processed (2014$) - comprising $27.93/ton from the Plastics Decontamination unit and $20.67/ton from the Solid Feed preprocessing unit. When revenue offsets are considered: • Tipping fee provides significant credit based on $45 per ton of raw nMSW delivered. • PVC co-product sales add further revenue at an estimated $0.40 per pound. Combined, these offsets total approximately $40.77 million annually at 90% plant capacity, equivalent to $95.55 per dry ton of nMSW. After applying these credits, the net preprocessing cost becomes –$46.95 per dry ton, effectively turning preprocessing into a revenue-positive operation and driving the overall system cost well below the DOE target of $30 per ton for fractionation and decontamination. Sensitivity analysis A univariate sensitivity analysis was conducted for the kickdown pump configuration (Case 1) to examine the influence of individual techno-economic parameters on the overall process economics and the levelized cost of jet fuel. In this analysis, one variable was varied at a time while holding all others constant at their base-case values to isolate its effect on fuel cost. The parameters evaluated included the total plant cost (varied by ±25% of the base estimate), discount rate (6%, 8%, and 10%), nMSW tipping fee ($65, $45, and $25 per ton), selling price of the PVC by-product stream ($0.6, $0.4, and $0.2 per pound), plant capacity factor (95%, 90%, and 85%), number of operational staff in the gasification island (10, 22, and 50 personnel), maintenance cost (±25% of the base case), selling price of the ash stream from the U-GAS® process ($15, $7.5, and –$45 per ton), and electricity price ($30, $40, and $50 per MWh). These ranges were selected to represent reasonable variations in capital cost uncertainty, feedstock supply agreements, by-product market volatility, and operational flexibility typical for early- stage waste-to-jet-fuel projects. The results are presented in Figure 32, which illustrates the relative influence of each variable on the levelized cost of jet fuel. Page 70 Figure 32: Sensitivity analysis for key variables of process for kickdown pump Since this techno-economic analysis (TEA) incorporates several components derived from low TRL systems, the associated cost uncertainty remains relatively high. To account for this variability, the total plant cost was varied by ±25% around the base estimate. Within this range, the levelized cost of jet fuel (LCOJ) fluctuated between $2.9 and $6.3 per GGE, demonstrating that capital investment is the single most influential factor affecting overall project economics. Similarly, the discount rate varied between 6% and 10%, resulting in a jet-fuel cost range of $3.7 to $5.6 per GGE, which underscores the strong dependence of project financial performance on investment risk and financing assumptions. The tipping fee, which represents a key revenue stream for nMSW based conversion projects, was varied from $65 to $25 per ton; the corresponding jet fuel cost ranged from $3.8 to $5.4 per GGE. Ensuring a stable and favorable tipping fee over the project lifetime is therefore critical for economic viability. The selling price of the PVC by-product stream also significantly affects the jet-fuel cost. If the segregated PVC stream is impure or market prices fall to $0.20 per lb, the levelized jet-fuel cost increases to approximately $5.3 per GGE. Performed life cycle analyses The primary focus of this LCA is to quantify the net greenhouse gas (GHG) emissions associated with the nMSW-to-jet-fuel pathway. The analysis identifies relevant emission sources across the upstream, facility, and downstream segments, and all results are reported on the basis of a kilogram of fuel as the functional unit. A 100-year time horizon Global Warming Potential (GWP) is applied to calculate carbon dioxide equivalents (CO₂e) (22), and the gasoline co-product is managed through system expansion (displacement). Life cycle inventory (LCI) assumptions and emission factors were retrieved from GREET 2024 (Rev. 1) (23). Key LCI data used in this study include: • GWPs from IPCC AR6, based on a 100-year time horizon. Page 71 • Transportation emissions for delivery of nMSW feedstock and transport of captured CO₂ for sequestration. • U.S. average electricity emission factors. • Emissions from gasoline production (well-to-gate). • Emissions from jet fuel during transportation, distribution, and combustion. Since the INL study did not include a detailed composition of the raw nMSW feedstock, this analysis assumes feedstock characteristics taken from the US average MSW composition reported in GREET 2024. Accordingly: • The nMSW is assumed to have the same composition as U.S. average MSW. • Avoided emissions are credited for bypassing conventional waste management routes such as landfill disposal and incineration. • The U.S. average MSW biogenic carbon content is estimated at 54 wt%, and the associated avoided emission is 192 g CO₂e/kg MSW. The well-to-wheels GHG emissions for this analysis were calculated outside of GREET due to the custom process configuration and unique material and energy flows associated with the U- GAS® and Fischer-Tropsch units. Table 26 summarizes the process emissions by life cycle stage for both design configurations, including comparisons between cases with and without carbon capture and storage (CCS). The total CO₂e emissions - covering upstream, jet fuel production, transportation, and end-use stages - are reported. Consistent with GREET accounting conventions, negative biogenic carbon emissions are only credited when biogenic-derived GHGs are released. The magnitude of these neutrality credits reflects the carbon released as CO₂, since biogenic carbon originates from atmospheric CO₂. Any carbon emitted as CH₄ (e.g., from biofuel combustion) receives a smaller net credit, reflecting the difference in global warming potential between CO₂ and CH₄ (1 vs. 27, respectively). The baseline, representing conventional petroleum-derived jet fuel, has a high carbon intensity with well-to-wheels emissions of approximately 3.84 kg CO₂e per kg of jet fuel, equivalent to 89 g CO₂e per MJ. Table 26: LCA breakdown Kickdown Pump Case Lock Hopper Case (kg CO2e/kg jet fuel) (kg CO2e/kg jet fuel) Stage Without Without With With CCS CCS CCS CCS nMSW avoided emissions (2.14) (2.14) (2.14) (2.14) Upstream transportation of nMSW 0.04 0.04 0.04 0.04 (Background) electricity consumption 1.58 1.58 1.52 1.52 Page 72 Kickdown Pump Case Lock Hopper Case (kg CO2e/kg jet fuel) (kg CO2e/kg jet fuel) Stage Without Without With With CCS CCS CCS CCS GHG generation/emissions 2.07 2.07 2.43 2.43 GHG emissions biogenic credit (1.36) (1.36) (1.59) (1.59) waste disposal 1.29 1.29 1.29 1.29 Facility waste disposal biogenic credit (0.45) (0.45) (0.45) (0.45) (Foreground) release of capturable CO2 4.85 - 4.49 - capturable CO2 biogenic credit (3.17) (3.17) (2.94) (2.94) CO2 transport to sequestration - 0.02 - 0.02 PVC non-avoidance 0.01 0.01 0.01 0.01 Allocation gasoline co-product (0.26) (0.26) (0.26) (0.26) T&D fuel transport (jet fuel) 0.01 0.01 0.01 0.01 jet fuel end-use 3.20 3.20 3.20 3.20 Use jet fuel end-use biogenic credit (2.09) (2.09) (2.09) (2.09) Total (kg CO2e/kg jet fuel) 3.56 (1.26) 3.51 (0.96) Total (g CO2e/MJ jet fuel) 82.52 (29.18) 81.28 (22.19) CI improvement over baseline 7% 133% 9% 125% The improvement over the baseline petroleum-derived jet fuel is approximately 7–9% without CCS and 125–133% with CCS. The primary contributors to carbon intensity (CI) reduction for the nMSW-to-jet-fuel pathway are the avoided emissions from diverting nMSW from conventional waste management, the biogenic carbon emission credits, and the avoided emissions associated with CCS. These reductions – particularly the avoided emissions and biogenic credits – are critical for enabling nMSW gasification to achieve carbon intensities comparable to or lower than those of fossil-based jet fuel, beyond serving as a low-carbon-intensity pathway when integrated with CCS. Figure 33 and Figure 34 illustrate the stage-level GHG emissions for the nMSW-to-jet fuel process using the kickdown pump and lock-hopper cases. In both configurations, the overall well-to-wheel carbon intensity is shaped by the balance of avoided landfill emissions (−2.14 kg CO₂e/kg jet fuel), process energy use, and biogenic credits applied across multiple stages. Positive contributions arise largely from electricity consumption (1.58 vs. 1.52 kg CO₂e/kg jet fuel) and facility-level GHG generation (2.07 vs. 2.43 kg CO₂e/kg jet fuel), which are modestly higher in the lock-hopper case. These impacts are partially offset by biogenic credits from waste Page 73 disposal and end-use combustion, and more substantially by the biogenic credit associated with capturable CO₂, which is slightly larger in the kickdown pump case (−3.17 vs. −2.94 kg CO₂e/kg jet fuel). Downstream emissions are nearly identical between cases. As a result, both configurations achieve net-negative well-to-wheel emissions when CCS is applied, with totals of −1.26 kg CO₂e/kg jet fuel for the kickdown pump case and −0.96 kg CO₂e/kg jet fuel for the lock-hopper case. kg CO2e/kg jet fuel -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 nMSW Avoided Emissions -2.1 0.0 Transportation of nMSW 1.6 Electricity Consumption 2.1 GHG Generation/Emissions GHG Emissions Biogenic Credit -1.4 1.3 Waste Disposal Increase Waste Disposal Biogenic Credit -0.5 0.0 Release of Captureable CO2 Decrease Captureable CO2 Biogenic Credit -3.2 0.0 CO2 Transport to Sequestration Total 0.0 PVC Non-Avoidance Gasoline Co-Product -0.3 0.0 Fuel Transport (Jet Fuel) 3.2 Jet Fuel End-Use Figure 33: Stage-level GHG emissions for nMSW-to-jet fuel process using kickdown pump Jet Fuel End-Use Biogenic Credit -2.1 TOTAL (Well-to-Wheel) -1.3 Page 74 Page 75 Figure 34: Stage-level GHG emissions for nMSW-to-jet fuel process using lock hopper Page 76 Advanced technology readiness level from 2 to 4 Novel technologies that furthered the goals of the project were measured against criteria in the Technology Readiness Assessment of the DOE. For the following key components, each progressed two levels (Figure 35): • Artificial intelligence (AI) was used to sort and decrease plastic in nMSW which helps to reduce bed agglomeration and de-fluidization in the U-GAS® gasifier. • A kickdown pump was developed, offering a fast, reliable feeding technology for preprocessed nMSW that overcomes issues found in traditional lock hopper systems and screw feeders, such as release of dusty gas and frequent maintenance of expensive metal- seated valves. Technology Readiness Assessment –––––––––––– AI Sorter Level 4 Kickdown Pump Level 3 Project End Level 2 Project Start Figure 35: Progression of technology readiness level over project lifetime Page 77 DISCUSSION This section provides an analysis of the technoeconomic and life cycle results for converting nMSW to jet fuel via the U-GAS® pathway. It examines cost performance relative to DOE targets, identifies key cost drivers, and evaluates GHG reduction potential under different scenarios. The discussion also outlines strategies to reduce costs, optimize processes, and increase revenue, forming the basis for recommendations to improve economic viability and accelerate commercialization. A concluding summary is provided at the end of this section to highlight key findings and future work. Technoeconomic analysis nMSW preprocessing cost The current preprocessing cost for nMSW, excluding revenue offsets, is $48.60 per dry ton, which exceeds the DOE target of $30 per ton. This cost comprises $27.93 per ton for plastics decontamination and $20.67 per ton for solid feed preprocessing. When tipping fees and PVC co-product sales are included, the economics shift dramatically, resulting in a net cost of –$46.95 per dry ton, effectively making preprocessing revenue-positive. However, if these offsets are removed, significant cost reductions are required to meet DOE objectives. Analysis of cost components reveals that labor dominates plastics decontamination, accounting for nearly 58% of its annualized cost, while energy consumption is the largest contributor in solid feed preprocessing, representing about 44% of its cost. Repairs, supplies, and maintenance also add substantial overhead, and capital costs remain significant despite amortization. Levelized cost of jet fuel The TEA demonstrates that converting nMSW to jet fuel via the U-GAS® gasification pathway is both technically feasible and economically promising, particularly when using the kickdown pump feeding configuration. Compared to the lock hopper system, the kickdown pump design reduces total plant cost by approximately 6%, lowers the total as-spent cost from $734 million to $689 million, and decreases the levelized cost of jet fuel from $5.03 to $4.60 per GGE - a 9% reduction. These cost advantages stem from system simplification, reduced mechanical complexity, and lower installation and auxiliary equipment costs. Sensitivity analysis identified total plant cost, discount rate, tipping fee, and PVC by-product price as the most influential variables affecting fuel economics, emphasizing the importance of capital efficiency, stable feedstock agreements, and by-product market optimization. While uncertainties remain due to the inclusion of lower TRL subsystems, the results highlight clear opportunities to enhance economic performance through larger-scale deployment, improved feedstock flexibility, favorable financing, and optimized by-product utilization. Life cycle analysis Overall, the LCA results demonstrate that converting nMSW to jet fuel via the U-GAS® pathway can achieve significant greenhouse gas (GHG) reductions compared to conventional petroleum- Page 78 based jet fuel. Without CCS, the process already provides a 7–9% improvement in carbon intensity due primarily to avoided landfill emissions and biogenic carbon credits. When CCS is incorporated, the pathway achieves net-negative emissions, improving the carbon intensity by 125–133% relative to the baseline. The combination of waste diversion benefits, utilization of biogenic carbon, and carbon capture makes the nMSW-to-jet-fuel system a promising low- carbon alternative to fossil-based jet fuel production. Recommendations nMSW preprocessing cost To reduce costs without relying on revenue credits, the following strategies are recommended. Automation of sorting and material handling can substantially lower labor expenses, which are the single largest cost driver. Energy optimization through efficient grinders, briquetters, and heat integration using waste heat from syngas cooling can reduce power costs. Predictive maintenance programs will minimize repair and upkeep costs, while modular equipment design can control capital investment and allow incremental scaling. Further savings can be achieved by integrating drying and densification steps and co-locating preprocessing facilities near waste sources to reduce handling and logistics costs. Implementing these measures could reduce preprocessing costs by 40–50 percent, bringing them closer to the DOE $30 per ton target even without revenue offsets. This improvement strengthens the economic viability of nMSW-to-jet- fuel conversion and supports the pathway toward commercial deployment. Levelized cost of jet fuel The sensitivity analysis identifies several key variables that can be optimized to enhance overall project economics and move toward the target levelized cost of jet fuel (LCOJ) of $2.5 per GGE. These variables represent both technical and strategic opportunities for cost reduction, process efficiency improvement, and revenue enhancement. The following recommendations summarize potential pathways to achieve the target LCOJ. • Plant scale and location: Locating the facility in a region with an abundant and consistent nMSW supply would enable construction of a larger-scale plant, improve economies of scale and reduce the unit fuel cost. For example, Los Angeles County generated approximately 11 million tons of landfill waste in 2022, equivalent to roughly 30,000 tons per day, which could readily support a commercial-scale facility. • Feedstock flexibility: Blending nMSW with other low-cost carbonaceous feedstocks, such as coal or biomass residues, could stabilize feedstock composition and enhance process throughput, leading to improved gasifier utilization and lower overall production cost. However, each type of feedstock carries its own greenhouse gas emissions, energy inputs, and waste outputs, and hence a separate evaluation is needed for these cases to assess the sustainability and economic viability. • Tipping fee structure: Selecting a project site with favorable tipping fees or negotiating long-term waste supply contracts with municipal or private partners can significantly Page 79 improve project cash flow. Higher tipping fees directly translate into greater revenue for nMSW-based processes. • Financing terms: Securing a long-term, low-interest loan or other favorable financing mechanisms would reduce annual debt service requirements, thereby lowering the levelized fuel cost and improving the internal rate of return of the project. • By-product revenue optimization: Enhancing the purity and market value of the sorted plastic (PVC) stream through improved pre-processing and sorting technologies can yield additional revenue streams. Refining or pelletizing the plastic fraction for resale to recycling or chemical markets could further offset jet fuel production costs. To achieve the target levelized cost of $2.50 per GGE, several pathways for improvement are recommended. Scaling the plant in regions with abundant nMSW supply can enhance economies of scale and significantly lower unit fuel cost. Securing higher and more stable tipping fees or long-term waste supply agreements can improve project revenue stability. Enhancing feedstock flexibility by co-feeding nMSW with other low-cost carbonaceous materials such as coal or biomass residues could improve gasifier utilization and reduce cost, though further evaluation of their life-cycle impacts is necessary. Accessing favorable financing terms - such as low-interest loans or public-private partnerships - would lower capital recovery requirements. Finally, increasing the purity and market value of the PVC co-product through improved sorting and processing could further offset fuel costs. Collectively, these measures would move the process closer to the target fuel cost while improving the economic resilience and long-term sustainability of the nMSW-to-jet-fuel pathway. Future work should focus on these areas to further enhance the financial viability and long-term sustainability of nMSW based jet fuel production projects. Continued research and development are needed to refine process integration, improve feedstock pre-processing, and optimize by- product utilization to maximize overall value recovery. In parallel, efforts to establish favorable policy mechanisms, secure low-cost financing, and develop stable markets for waste-derived fuels and co-products will be essential to enable commercial deployment. Collectively, these advancements will help reduce economic risk, attract investment, and accelerate the transition of waste-to-jet-fuel technologies from pilot to commercial scale. Conclusions This project demonstrates that nMSW can be successfully preprocessed, decontaminated, and converted into a high-quality feedstock suitable for U-GAS® gasification, enabling production of drop-in jet fuel with strong economic and environmental performance. The integrated system combining advanced AI-enabled sorting, staged size-reduction, densification, and a novel kickdown pump feeding mechanism consistently met or exceeded DOE targets for purity, variability reduction, PVC removal, and feed system performance. Technoeconomic analysis shows that the system provides a clear pathway to further reduce production costs toward the long-term target of $2.50 per GGE, with additional opportunities to improve both CAPEX and OPEX at commercial scale. Life cycle analysis shows net-negative well-to-wheel emissions when Page 80 paired with carbon capture and sequestration, highlighting the substantial decarbonization potential of nMSW-derived jet fuel. The benefits of this integrated approach are most pronounced when the system leverages its full suite of innovations – from high-efficiency AI decontamination to pressure-stable densified feedstock and rapid-cycle pump injection – which together overcome long-standing challenges associated with heterogeneous solid waste conversion. By achieving TRL 4 for both the AI sorting tool and the kickdown pump, the project establishes a strong technical foundation for pilot-scale TRL 6 demonstrations. With growing demand for sustainable aviation fuel and increasing pressure to divert waste from landfills, the validated nMSW-to-jet-fuel pathway provides a viable, scalable, and cost-competitive route for decarbonizing the aviation sector while addressing waste management challenges. Continued development will allow stakeholders to optimize cost, improve reliability, and accelerate commercialization of this emerging waste- to-jet-fuel technology. Page 81 PRODUCTS AND PUBLICATIONS Saha, N., Klinger,J., Islam, M.T., Reza, T. (2023). Advanced biorefinery feedstock from non- recyclable municipal solid waste by mechanical preprocessing. Frontiers in Fuels,1. Ikbarieh, A., Dai, S., (2025). Compressibility and permeability of particulated non-recyclable municipal solid waste. Waste Management, Vol 201. Page 82 GLOSSARY ADB AdaBoost AGR acid gas removal AI artificial intelligence ASAE American Society of Agricultural and Biological Engineers ASTM American Society for Testing and Materials BETO Bioenergy Technologies Office BTU British Thermal Unit CAPEX Capital Expenditures CBC CatBoost classifier CCS carbon capture and storage CEPCI Chemical Engineering Plant Cost Index CI carbon intensity Cl chlorine CO carbon monoxide CO2 carbon dioxide CSR constant strain rate DCFROR discounted cash flow rate of return DOE US Department of Energy EERE Office of Energy Efficiency and Renewable Energy FOA funding opportunity announcement FT Fischer-Tropsch GGE gasoline gallon equivalent GHG greenhouse gas GWP global warming potential GREET greenhouse gases, regulated emissions, and energy use in transportation HDPE high-density polyethylene HHV higher heating value INL Idaho National Laboratory IPCC Intergovernmental Panel on Climate Change JRS Jordan Reductions Solutions LCA life cycle analysis LCI life cycle inventory LOCJ levelized cost of jet fuel LECO Laboratory Equipment Corporation LO-CAT liquid oxidation catalyst system LPG liquefied petroleum gas Page 83 LTFT low-temperature Fischer-Tropsch mL milliliter ML machine learning MJ megajoule MRF Materials Recovery Facility MSW municipal solid waste NETL National Energy Technology Laboratory NIR near infrared nMSW non-recyclable municipal solid waste NREL National Renewable Energy Laboratory OPEX operating expenditures OSBL outside battery limits PET polyethylene terephthalate PLC programmable logic controller POX partial oxidation PP polypropylene PS polystyrene PSD particle size distribution PVC polyvinyl chloride RFC Random Forest Classifier SVM Support Vector Machine Syncrude synthetic crude oil TASC total as spent costs TEA technoeconomic analysis TOC total overnight costs TPD short tons per day TRL technology readiness level UHP ultra-high purity UHV UHV Technologies, Inc. US United States of America VM volatile matter VR variability reduction WGS water gas shift XRF X-ray fluorescence Page 84 LIST OF TABLES AND FIGURES TABLES Table 1: Mass balance of preprocessed samples ..............................................................................................11 Table 2: Performance summary for machine learning models ....................................................................17 Table 3: Grinding energy required for each unit operation ..........................................................................24 Table 4: Fuel properties of preprocessed gasifier feed ..................................................................................26 Table 5: Fischer-Tropsch product composition for cobalt-LTFT systems ................................................29 Table 6: Estimated product yields for FT Syncrude refining section ..........................................................30 Table 7: Carbon balances for each case ...............................................................................................................32 Table 8: Composition of tested blends and second-pass feedstock .........................................................36 Table 9: Summary of key milestones and results ..............................................................................................42 Table 10: Proximate analyses and higher heating value ................................................................................45 Table 11: Ultimate analyses .......................................................................................................................................45 Table 12: Sorting data for removal of PVC ..........................................................................................................51 Table 13: PSD and density of on-spec sample ...................................................................................................52 Table 14: Proximate analysis and HHV of second-pass samples ................................................................53 Table 15: Ultimate analysis of second-pass samples .......................................................................................53 Table 16: Structural and non-structural ash of on-spec material ...............................................................54 Table 17: nMSW testing data ...................................................................................................................................57 Table 18: CAPEX for kickdown pump case (Case 1) .........................................................................................60 Table 19: CAPEX for lock hopper case (Case 2) .................................................................................................61 Table 20. Variable operating costs for kickdown pump case (Case 1)......................................................62 Table 21: Fixed operating costs for kickdown pump case (Case 1) ...........................................................64 Table 22: Variable operating costs for lock hopper case (Case 2) ..............................................................64 Table 23: Fixed operating costs for lock hopper case (Case 2)....................................................................66 Table 24: Pre-processing cost for plastics decontamination unit ...............................................................67 Table 25: Preprocessing costs for solid feed unit .............................................................................................68 Table 26: LCA breakdown ..........................................................................................................................................71 Table 27: Stream summary for kickdown pump (Case 1) ........................................................................... A-2 Table 28: Stream summary for lock hopper (Case 2) ................................................................................... A-4 Table 29: Parameter values from laboratory tests for hopper flow simulations .................................B-3 Table 30: Volumetric throughput of a kickdown pump with variable bore size ................................ D-1 Table 31: Technical Table ......................................................................................................................................... E-1 Page 85 FIGURES Figure 1: Procedure for feedstock preprocessing ............................................................................................... 6 Figure 2: nMSW received in truck and stored in bunker, with closeup inset ........................................... 7 Figure 3: Rejected nMSW ............................................................................................................................................. 8 Figure 4: Supersacks after first-stage size reduction ......................................................................................... 8 Figure 5: Photos of examined materials ...............................................................................................................10 Figure 6: Two-stage telescoping ram ....................................................................................................................14 Figure 7: Kickdown pump – a) annotated 3D model; b) actual in laboratory ........................................15 Figure 8: X-ray fluorescence of plastic materials...............................................................................................18 Figure 9: Type 1 plastic (PET) ....................................................................................................................................19 Figure 10: Type 2 plastic (HDPE)..............................................................................................................................19 Figure 11: Type 3 plastic (PVC).................................................................................................................................20 Figure 12: Type 5/7 polypropylene (PP) ...............................................................................................................20 Figure 13: Type 6 polysterene (PS) .........................................................................................................................21 Figure 14: Block flow diagram of preprocessing steps ...................................................................................22 Figure 15: Mass balance of process .......................................................................................................................24 Figure 16: Cyclic heat treatment to stabilize nMSW sample ........................................................................25 Figure 17: Modeling approach for converting nMSW to jet fuel ................................................................27 Figure 18: Simplified procedure evaluated in life cycle analysis .................................................................31 Figure 19: Knife mill with conveyor belt pictured above; crumbler below ..............................................34 Figure 20: Particulated nMSW fractions and blends evaluated ...................................................................35 Figure 21: Setup for consolidation and permeability testing .......................................................................37 Figure 22: Automated kickdown pump for testing ..........................................................................................38 Figure 23: Multi-modal sensing architecture of the UHV blue plastics sorter .......................................40 Figure 24: Plastics collection bins from the UHV blue plastics sorter .......................................................40 Figure 25: MSW shredder ..........................................................................................................................................41 Figure 26: Fractional analysis of nMSW ................................................................................................................44 Figure 27: CSR consolidation for first-pass feedstock.....................................................................................47 Figure 28: CSR consolidation for second-pass feedstock ..............................................................................48 Figure 29: Gas permeability at 150 psig ...............................................................................................................49 Figure 30: Gas permeability for second-pass feedstock ................................................................................50 Figure 31: nMSW before (a) and after (b) pump test ......................................................................................58 Figure 32: Sensitivity analysis for key variables of process for kickdown pump ...................................70 Page 86 Figure 33: Stage-level GHG emissions for nMSW-to-jet fuel process using kickdown pump ........74 Figure 34: Stage-level GHG emissions for nMSW-to-jet fuel process using lock hopper .................75 Figure 35: Progression of technology readiness level over project lifetime ...........................................76 Figure 36: Block flow diagram of nMSW to jet fuel conversion - kickdown pump (case 1) .......... A-1 Figure 37: Block flow diagram of nMSW to jet fuel conversion - lock hopper (case 2) .................. A-3 Figure 38: Hopper flow modeling .........................................................................................................................B-1 Figure 39: Determining material properties for numerical simulation ...................................................B-2 Figure 40: Determining material constitutive parameters for cardboard stream ...............................B-3 Figure 41: Hopper flow testing setup ..................................................................................................................B-4 Figure 42: Hopper flow results for first-pass feedstocks ..............................................................................B-5 Figure 43: Hopper flow results for 2 mm first-pass feedstock with hopper openings .....................B-6 Figure 44: Hopper flow results for the second-pass feedstock with hopper openings ...................B-6 Figure 45: Numerical (left) vs. experimental (right) hopper flow for 2 mm rigid plastics ................B-7 Figure 46: Numerical vs. experimental hopper flow for second-pass feedstock (4 mm) ................B-8 Figure 47: AI sorter with 20,000 metric tons/year capacity ........................................................................ C-1 Figure 48: Price of a kickdown pump vs. maximum volumetric throughput....................................... D-3 Figure 49: Kickdown pump hydraulic power demand vs. maximum volumetric throughput ....... D-4 Page 87 REFERENCES (1) 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. (2) McPhail, A., Griffin, R., El-Halwagi, M., Medlock, K., Alvarez, P. J. J. (2014). Environmental, Economic, and Energy Assessment of the Ultimate Analysis and Moisture Content of Municipal Solid Waste in a Parallel Co-Combustion Process. Energy Fuels, Vol 28. (3) Farhana, S. S., Weekes, J. S., Melikechi, N., Hossain, S, Fluman, D. A., Rana, M. M. (2016). In- situ Moisture Content Measurement of Fresh Municipal Solid Waste by Wenner Resistivity Method. The Journal of Solid Waste Technology and Management, Vol 42. (4) De Klerk, A. Fischer-Tropsch Refining. 1st ed. Weinheim, Germany: Wiley-VCH, 2011. (5) Leal-Quirós, E. (2004). Plasma Processing of Municipal Solid Waste. Brazilian Journal of Physics, Vol 34. (6) Pandey, U., Putta, K. R., Rout, K. R., Blekkan, E. A., Rytter, E., Hillestad, M. (2022). Staging and Path Optimization of Fischer-Tropsch Synthesis. Chemical Engineering Research and Design, Vol 187. (7) National Renewable Energy Laboratory (2024), Aviation Fuels | Transportation | 2024 | ATB. (8) National Energy Technology Laboratory (2022), Comparison of Commercial, State-of-the-Art, Fossil-Based Hydrogen Production Technologies. Skone, T.J., Lewis, E., McNaul, S., Jamieson, M., Henriksen, M., Matthews, H., Walsh, L., Grove, J., Shultz, T., Stevens, R. DOE/NETL- 2022/3241. (9) National Energy Technology Laboratory (2021). Quality Guidelines for Energy Systems Studies: Cost Estimation Methodology for NETL Assessments of Power Plant Performance. Theis, J. NETL-PUB-22580. (10) National and Regional Trend on Tipping Fee. https://reencle.co/blogs/news/national-and- regional-trend-on-tipping-fee. (11) U.S. Regular All Formulations Retail Gasoline Prices (Dollars per Gallon). https://www.eia.gov/dnav/pet/hist/leafhandler.ashx?n=pet&s=emm_epmr_pte_nus_dpg&f= m (12) United States, Department of Transportation, Bureau of Transportation Statistics. National Transportation Statistics (NTS). (2019). https://doi.org/10.21949/1503663. (13) Compass International. 2025 Front End/Conceptual Estimating Database. Compass International, 2025. Page 88 (14) Northwestern University Chemical Process Design Open Textbook, available at https://processdesign.mccormick.northwestern.edu/index.php/Estimation_of_capital#OSBL_( Outside_Battery_Limits)_Plant_Costs. (15) Method. Trucking Rates Per Mile – Pricing Guide. Available at: https://www.method.me/pricing-guides/trucking-rates-per-mile/ (16) SoCal Recycling. SoCal Recycling Rates. Available at: https://socalrecycling.com/recycling- rates/ (17) Biofuel Machines. Briquettes Manufacture Cost and Revenues Analysis. Available at: https://www.biofuelmachines.com/Briquettes-Manufacture-Cost-and-Revenues- Analysis.html (18) National Energy Technology Laboratory (NETL). Assessment of Gasifier By-Product Utilization. Available at: https://netl.doe.gov/research/Coal/energy- systems/gasification/gasifipedia/assessment-byproduct-utilization (19) IEAGHG (2017). Techno-Economic Evaluation of SMR Based Standalone (Merchant) Hydrogen Plant with CCS. Available at: https://ieaghg.org/publications/techno-economic- evaluation-of-smr-based-standalone-merchant-hydrogen-plant-with-ccs/ (20) Parker, M. (2010). Method for Removing Hydrogen Sulfide from Sour-Gas and Converting It to Hydrogen and Sulfuric Acid. Ph.D. Dissertation, Stanford University, Stanford, CA (United States). (21) Handbook of Petroleum Refining Processes, Third Edition; McGraw-Hill Education, New York, New York (United States), 2004 (22) Intergovernmental Panel on Climate Change (IPCC). Sixth Assessment Report. March 2023. Available at: https://www.ipcc.ch/assessment-report/ar6/ (23) Argonne National Laboratory. GREET 2024 (Rev 1). Available at: https://greet.anl.gov/ APPENDIX A: BLOCK FLOW DIAGRAMS AND STREAM SUMMARIES Figure 36: Block flow diagram of nMSW to jet fuel conversion - kickdown pump (case 1) A-1 Table 27: Stream summary for kickdown pump (Case 1) Stream ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Temperature °C 25.0 25.0 25.0 30.0 85.0 24.0 45.0 850.0 399.4 148.9 850.0 25.0 850.0 69.0 400.0 90.0 280.0 400.0 336.0 45.0 38.0 44.9 220.0 220.0 219.6 219.6 219.6 25.0 400.0 25.0 25.0 150.0 Pressure bara 1.0 1.0 1.0 1.0 1.0 1.0 11.0 1.0 44.0 11.0 11.0 11.0 10.5 45.0 44.0 10.0 9.5 44.0 9.0 1.0 1.0 8.0 30.0 29.5 27.5 27.5 27.5 45.0 44.0 1.0 1.0 100.0 Mass Rate kg/hr 90,204 2,706 87,498 24,499 25,199 37,799 7,924 7,461 17,293 13,362 68,917 4,446 73,363 46,284 46,284 73,363 57,374 10 57,374 31,441 23,516 25,934 25,934 52,207 34,562 26,273 7,565 52,799 52,799 2,408 4,851 23,516 Component Flow nMSW kg/hr 90,204 87,498 37,799 PVC kg/hr 2,706 Reject / Ash kg/hr 24,499 7,461 Gasoline kg/hr 2,408 Jet fuel kg/hr 4,851 CO kg/hr 12,738 0 22,691 22,691 22,691 18,724 0 18,724 18,724 15,185 15,185 11,543 0 CH4 kg/hr 3,667 0 324 324 324 324 0 324 324 491 491 373 0 C2H4 kg/hr 280 0 0 0 0 0 0 0 0 9 9 7 0 C2H6 kg/hr 378 0 0 0 0 0 0 0 0 88 88 67 0 N2 kg/hr 235 301 78 404 404 404 404 0 404 404 1,686 1,686 1,282 0 COS kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 CS2 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 CO2 kg/hr 7,924 30,040 0 28,705 28,705 28,701 34,934 7,924 23,516 3,493 3,493 14,567 14,567 11,074 0 23,516 C2H2 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 HCN kg/hr 2 0 0 0 0 0 0 0 0 0 0 0 0 H2S kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 H2 kg/hr 1,350 0 2,417 2,417 2,417 2,702 0 2,702 2,702 2,075 2,075 1,577 0 CH2CO kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 H2O kg/hr 25,199 17,293 19,174 0 18,819 46,284 46,284 18,819 2,835 10 283 0 283 283 9,923 0 0 0 52,799 52,799 O2 kg/hr 13,127 0 4,368 0 0 0 0 0 0 0 0 0 0 0 NH3 kg/hr 32 0 2 2 1 1 0 1 1 6 6 4 0 METHANOL kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 PROPYLEN kg/hr 0 0 0 0 0 0 0 0 0 298 298 227 0 PROPANE kg/hr 0 0 0 0 0 0 0 0 0 158 158 120 0 N-HEPTAN kg/hr 0 0 0 0 0 0 0 0 0 1,052 0 0 1,052 1-HEPTEN kg/hr 0 0 0 0 0 0 0 0 0 684 0 0 684 1-HEPTNL kg/hr 0 0 0 0 0 0 0 0 0 18 0 0 0 C16ALKEN kg/hr 0 0 0 0 0 0 0 0 0 96 0 0 96 C16ALKAN kg/hr 0 0 0 0 0 0 0 0 0 1,823 0 0 1,823 C28ALKAN kg/hr 0 0 0 0 0 0 0 0 0 3,909 0 0 3,909 ETHANOL kg/hr 0 0 0 0 0 0 0 0 0 123 0 0 0 PROPIACD kg/hr 0 0 0 0 0 0 0 0 0 18 0 0 0 C6H6 kg/hr 339 0 0 0 0 0 0 0 0 0 0 0 0 C7H8 kg/hr 20 0 0 0 0 0 0 0 0 0 0 0 0 C10H8 kg/hr 415 0 0 0 0 0 0 0 0 0 0 0 0 C6H6O kg/hr 181 0 0 0 0 0 0 0 0 0 0 0 0 A-2 Figure 37: Block flow diagram of nMSW to jet fuel conversion - lock hopper (case 2) A-3 Table 28: Stream summary for lock hopper (Case 2) Stream ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Temperature °C 25.0 25.0 25.0 30.0 85.0 24.0 45.0 850.0 399.4 148.9 850.0 25.0 850.0 69.0 400.0 90.0 280.0 400.0 336.0 45.0 38.0 44.9 220.0 220.0 219.6 219.6 219.6 25.0 400.0 25.0 25.0 150 150.0 Pressure bar 1.0 1.0 1.0 1.0 1.0 1.0 11.0 1.0 44.0 11.0 11.0 11.0 10.5 45.0 44.0 10.0 9.5 44.0 9.0 1.0 1.0 8.0 30.0 29.5 27.5 27.5 27.5 45.0 44.0 1.0 1.0 11 100.0 Mass Rate kg/hr 90,204 2,706 87,498 24,499 25,199 37,799 7,924 7,461 17,293 13,362 68,917 4,446 73,363 46,284 46,284 73,363 57,374 10 57,374 31,441 21,785 25,934 25,934 52,207 34,562 26,273 7,565 52,799 52,799 2,408 4,851 1,731 21,785 Component Flow nMSW kg/hr 90,204 87,498 37,799 PVC kg/hr 2,706 Reject / Ash kg/hr 24,499 7,461 Gasoline kg/hr 2,408 Jet fuel kg/hr 4,851 CO kg/hr 12,738 0 22,691 22,691 22,691 18,724 0 18,724 18,724 15,185 15,185 11,543 0 CH4 kg/hr 3,667 0 324 324 324 324 0 324 324 491 491 373 0 C2H4 kg/hr 280 0 0 0 0 0 0 0 0 9 9 7 0 C2H6 kg/hr 378 0 0 0 0 0 0 0 0 88 88 67 0 N2 kg/hr 235 301 78 404 404 404 404 0 404 404 1,686 1,686 1,282 0 COS kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 CS2 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 CO2 kg/hr 7,924 30,040 0 28,705 28,705 28,701 34,934 9,656 21,785 3,493 3,493 14,567 14,567 11,074 0 1,731 21,785 C2H2 kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 HCN kg/hr 2 0 0 0 0 0 0 0 0 0 0 0 0 H2S kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 H2 kg/hr 1,350 0 2,417 2,417 2,417 2,702 0 2,702 2,702 2,075 2,075 1,577 0 CH2CO kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 H2O kg/hr 25,199 17,293 19,174 0 18,819 46,284 46,284 18,819 2,835 10 283 0 283 283 9,923 0 0 0 52,799 52,799 O2 kg/hr 13,127 0 4,368 0 0 0 0 0 0 0 0 0 0 0 NH3 kg/hr 32 0 2 2 1 1 0 1 1 6 6 4 0 METHANOL kg/hr 0 0 0 0 0 0 0 0 0 0 0 0 0 PROPYLEN kg/hr 0 0 0 0 0 0 0 0 0 298 298 227 0 PROPANE kg/hr 0 0 0 0 0 0 0 0 0 158 158 120 0 N-HEPTAN kg/hr 0 0 0 0 0 0 0 0 0 1,052 0 0 1,052 1-HEPTEN kg/hr 0 0 0 0 0 0 0 0 0 684 0 0 684 1-HEPTNL kg/hr 0 0 0 0 0 0 0 0 0 18 0 0 0 C16ALKEN kg/hr 0 0 0 0 0 0 0 0 0 96 0 0 96 C16ALKAN kg/hr 0 0 0 0 0 0 0 0 0 1,823 0 0 1,823 C28ALKAN kg/hr 0 0 0 0 0 0 0 0 0 3,909 0 0 3,909 ETHANOL kg/hr 0 0 0 0 0 0 0 0 0 123 0 0 0 PROPIACD kg/hr 0 0 0 0 0 0 0 0 0 18 0 0 0 C6H6 kg/hr 339 0 0 0 0 0 0 0 0 0 0 0 0 C7H8 kg/hr 20 0 0 0 0 0 0 0 0 0 0 0 0 C10H8 kg/hr 415 0 0 0 0 0 0 0 0 0 0 0 0 C6H6O kg/hr 181 0 0 0 0 0 0 0 0 0 0 0 0 A-4 APPENDIX B: SIMULATE AND VALIDATE HOPPER FLOW Efficient and reliable feeding of nMSW into pressurized gasification systems requires a clear understanding of bulk solid flow behavior under various operating conditions. To support the design and optimization of the nMSW feeding system, a comprehensive hopper flow study was conducted combining physical experiments and numerical modeling. The study aimed to characterize the flowability of individual nMSW fractions, their blends, and the second-pass preprocessed feedstock, and to evaluate the influence of particle size, hopper geometry, and outlet opening size on mass flow rate and jamming tendencies. Continuum-based hypoplastic constitutive modeling (Gudehus–Bauer model) was employed within a coupled Eulerian– Lagrangian (CEL) finite-element framework to simulate the flow behavior, with parameters calibrated from laboratory tests such as oedometer and ring-shear measurements. The experimental data obtained from dynamic hopper discharge tests provided essential validation for the numerical simulations and offered insight into the key factors controlling the flow consistency and feed stability of nMSW-derived materials. Modeling The hopper flow simulations use both 2D plane-strain slices of the wedge-shaped hopper and 3D models with geometries identical to that of the hopper used in physical testing (Figure 38). The simulations use continuum models, such as the hypoplastic Gudehus–Bauer (G-B). To address the mesh-tangling issue of the Finite Element Method (FEM) when simulating large deformation, the coupled Eulerian–Lagrangian (CEL) FEM scheme was applied, which was realized by two steps: a deformation step, where the material deformed with the mesh, followed by a remapping step, where the deformed mesh was returned to its original position and the material (deformed) properties were interpolated back to the “fixed” mesh. The inlet and outlet boundaries are maintained at atmospheric pressure, and the impacts of different wall frictions are investigated in the simulations. Parameters Unit CS friction angle, φc ° Max void ratio, ei0 - Min void ratio, ed0 - Void ratio at CS, ec0 - Granulate stiffness, hs kPa Pressure-sensitive exponent, n - Peak-critical stress transition exponent, α - Stiffness variation exponent, β - Figure 38: Hopper flow modeling B-1 Determine material parameters for the models The hypoplastic Gudehus–Bauer model requires the input of eight material parameters, which can be grouped into three categories: (1) the internal friction angle (ϕc) and exponent (α), which dominate the shear behavior; (2) the granulate hardness (hs), exponent (n), and parameter (β), which determine the compression behavior; and (3) the minimum, critical, and maximum void ratios at zero pressure (ed0, ec0, and ei0), which define the range of the void ratio. These parameters were calibrated using laboratory experiments and single-element FEM simulations, with the detailed calibration procedure reported in our earlier publications. As illustrated in Figure 39, ed0 and ec0 were obtained from the minimum/maximum density tests; hs and n were derived from oedometer tests; and the ring-shear friction angle (ϕcrs) was measured from Schulze ring-shear tests. Then, using single-element FEM simulations, n and ei0 were further calibrated against oedometer results, ϕc from the critical-state shear behavior, and α from the peak-stress behavior observed in the ring-shear tests. Note that the internal friction angle at the critical state (ϕc) was mapped from the ring-shear friction angle (ϕcrs) by numerically determining the Lode angle. (a) Packing (b) Compressibility 250 200 ρb [kg/m3] 150 100 50 ρbmin 0 0 5 10 σv [kPa] 250 Reloading 200 ρb [kg/m3] 150 Unloading Loading 100 5 6 7 8 9 10 11 log(σv/Pa) [-] Figure 39: Determining material properties for numerical simulation B-2 Figure 40 shows the parameter calibration results for the cardboard stream with different sizes. (a) Packing 300 300 ρbmax 2mm ρbmin ρb [kg/m3] ρb [kg/m3] 200 200 4mm 100 6mm 100 0 0 0 5 10 0 4 8 σv [kPa] Grinder size [mm] (b) Compressibility (c) Angle of repose 300 60 250 2mm ρb [kg/m3] 50 AoR [º] 200 4mm 40 150 6mm 100 30 1 10 100 0 2 4 6 8 σv [kPa] Grinder size [mm] (d) Angle of repose photos 2mm 4mm 6mm Figure 40: Determining material constitutive parameters for cardboard stream (using three sizes of blades) Values of the eight parameters in the hypoplastic model for the five individual streams, four blends, and the second-pass feedstock are summarized in Table 29. Table 29: Parameter values from laboratory tests for hopper flow simulations Rigid Thin Cardb Blend Blend Blend 2nd-Pass Parameters Unit Paper Foam plastic film oard #1 #2 #3 Feedstock CS friction angle, φc ° 42.1 50.2 45.3 42.2 49.9 26.1 39.9 37.0 45 Max void ratio, ei0 - 2.793 3.463 3.503 2.573 2.129 1.245 1.214 1.400 1.191 Min void ratio, ed0 - 1.956 0.684 0.535 1.266 0.264 0.394 0.384 0.367 0.411 Void ratio at CS, ec0 - 2.327 3.148 2.919 2.339 1.936 1.037 1.011 0.967 1.083 B-3 Rigid Thin Cardb Blend Blend Blend 2nd-Pass Parameters Unit Paper Foam plastic film oard #1 #2 #3 Feedstock Granulate stiffness, hs kPa 541.5 1.93 3.49 2.39 2.65 20.0 21.0 20.3 30 Pressure-sensitive - 0.383 0.449 0.458 0.405 0.397 0.5 0.5 0.5 0.5 exponent, n Peak-critical stress - 0.15 0.5 0.5 0.5 0.4 0.01 0.01 0.01 0.3 transition exponent, α Stiffness variation - 0.2 1.00 0.10 1.00 1.00 0.10 0.01 0.01 0.1 exponent, β Testing The experimental setup for physical measurements of hopper flows is shown in Figure 41. It consists of a drum to contain the tested materials (both a single-drum container and a two- drum container were used, with the latter accommodating a larger quantity of material), a dynamic hopper, a receiving bin to collect the outflowed materials, a camera, and a scale to record the outflowed mass over time. Drum Dynamic hopper Receiving bin Scale Camera Data loging Figure 41: Hopper flow testing setup One-drum system (right) and two-drum system (left). Figure 42 summarizes the hopper flow results, showing the outflowed mass from the dynamic hopper over time for the first-pass feedstock in three particle sizes. Different colors in each plot represent repeated tests using different batches of materials with the same particle size. For the B-4 6 mm materials, there were four jammed cases (shown as blue curves in the left plot), in which no additional mass flowed out of the hopper over time. The remaining eight trials eventually discharged completely from the hopper. Among these eight samples, however, several exhibited temporary jamming, characterized by short periods where the outflow mass did not increase with time (i.e., plateaus in the curves). No jammed cases were observed for the 4 mm and 2 mm materials, although some of the 4 mm tests showed temporary clogging that later resolved spontaneously to continue flow. For example, the green-colored data in the 4 mm sample indicated approximately 7 lb of material discharged during the first 15 seconds, followed by an additional 7 lb over 65 seconds as the outlet was partially jammed. After about 80 seconds, the clogging cleared, and the remaining material flowed until the drum was emptied. In contrast, all 2 mm material samples demonstrated smooth and continuous outflow, with no permanent or noticeable temporary jamming observed. 6mm 4mm 2mm 120 50 50 2 drums 45 45 100 40 40 35 35 Mass flowed [lbs] 80 Mass flowed [lbs] Mass flowed [lbs] 30 30 60 25 25 20 20 40 15 15 10 10 20 5 5 0 0 0 00:00 04:00 08:00 12:00 16:00 20:00 0 20 40 60 80 100 120 0 20 40 60 80 Time [mm:ss] Time [second] Time [second] Figure 42: Hopper flow results for first-pass feedstocks Figure 43 summarizes the hopper flow results for 2 mm first-pass feedstock through the dynamic hopper with 2, 3, and 4 inches of openings. The inset figures illustrate where the testing materials are jammed (shaded zone) at the hopper outlet, delaying the flow or jammed completely and stopping the flow. B-5 4-inch Opening 3-inch Opening 2-inch Opening 50 70 70 45 60 60 40 50 50 35 Mass flowed [lbs] Mass flowed [lbs] Mass flowed [lbs] 30 40 40 25 30 30 20 15 20 20 10 10 10 5 0 0 0 0 100 200 300 0 100 200 300 0 20 40 60 80 Time [second] Time [second] Time [second] Figure 43: Hopper flow results for 2 mm first-pass feedstock with hopper openings The hopper flow results of the second-pass feedstock with different hopper opening are summarized in Figure 44. The results are consistent for each hopper opening size. Once the opening size exceeds 3 inches, the feedstock flows relatively fast: approximately 3–4 lbs/sec. 40 40 40 35 35 35 30 30 30 Mass flowed [lb] Mass flowed [lb] Mass flowed [lb] 25 25 25 20 20 20 15 15 15 10 10 10 5 2-inch outlet 5 3-inch outlet 5 4.25-inch outlet 0 0 0 0 20 40 60 80 100 0 5 10 15 20 0 5 10 15 20 Time [sec] Time [sec] Time [sec] Figure 44: Hopper flow results for the second-pass feedstock with hopper openings Validating Figure 45 compares numerical simulations and experimental results for hopper flow of 2 mm rigid plastic particles. The left panel shows simulated mass flow over time for hopper openings of 2, 3, and 4 inches, illustrating the strong influence of opening size on discharge rate. The inset images depict flow patterns within the hopper. The right panel presents experimental B-6 measurements (colored squares) alongside numerical predictions (blue line) for a hopper with a 4.” opening, demonstrating close agreement between model and experiment. Figure 45: Numerical (left) vs. experimental (right) hopper flow for 2 mm rigid plastics Figure 46 compares the outflowed mass over time from physical testing and numerical simulations for second-pass feedstock (4 mm), including cases with and without wall friction. It should be noted that the friction between the drum wall and feedstock was not experimentally determined in this study, although it is certain that wall friction is nonzero. This friction depends on factors such as material properties, surface texture, and drum geometry. The wall friction angle (δ) can be estimated using standard methods such as a Jenike shear cell or the inclined plane test. Based on the empirical Jenike flow rule, δ ≈ 0.7φ = 0.7 × 45° = 31.5°, where φ represents the internal friction angle of the feedstock material. As shown in Figure 46, incorporating wall friction in the simulation significantly improves agreement with experimental results, reducing the discrepancy in discharge rate and better capturing the observed flow behavior B-7 Figure 46: Numerical vs. experimental hopper flow for second-pass feedstock (4 mm) Left: zero wall friction; Right: wall friction coefficient = 0.7 (δ = ~35° friction angle). Summary The integrated experimental and modeling investigation successfully captured the complex hopper discharge behavior of prepressed nMSW feedstocks. Results demonstrated strong correlations between particle size, bulk density, and flowability, with smaller particle sizes (≤ 2 mm) showing stable mass flow and minimal jamming, while coarser fractions (≥ 6 mm) exhibited intermittent or complete blockage. The hypoplastic Gudehus–Bauer model, calibrated with laboratory derived material parameters, reproduced the observed flow trends and provided predictive capability for evaluating hopper geometry and outlet size effects. These findings provide a validated framework for designing and scaling up feed systems for heterogeneous solid waste streams, supporting reliable operation of downstream gasification and conversion processes. Future efforts will focus on refining the hypoplastic material model to better represent the multi- phase and compressible nature of nMSW derived feedstocks, including the effects of particle shape irregularity, inter-particle cohesion, and moisture content. Additional large-scale experiments under pressurized conditions will be performed to validate hopper flow at pilot scale. Integration of real-time flow monitoring and machine-learning-based predictive control will be explored to enhance feed consistency and prevent jamming events during continuous operation. B-8 APPENDIX C: AI-SORTER SCALABILITY Following the successful validation of the AI-based sorting algorithms, a technoeconomic analysis was conducted to assess the scalability and commercial viability of the system. The analysis was based on the design and demonstration of a 2-meter wide AI-enabled plastics decontamination system operating at belt speeds of 400–500 feet per minute. Under these conditions, a single sorter is capable of processing approximately 20,000 metric tons per year of nMSW when operated continuously at 24/7 with 80% uptime (Figure 47). To reach industrial throughput levels, three parallel sorters were modeled, resulting in a combined capacity of 60,000 metric tons per year. The total CAPEX for this configuration was estimated at approximately $2.0 million, which included the cost of sorters, installation (vibratory feeders and loaders), and facility upgrades such as power, water, and compressors. All costs mentioned here is 2024$. Figure 47: AI sorter with 20,000 metric tons/year capacity C-1 APPENDIX D: KICKDOWN PUMP SCALABILITY With the success of the laboratory scale prototype kickdown pump, the final task was to estimate the cost of the commercial scale kickdown pump. To contend with the throughput of the current solid fuel handling technology used in the U-GAS® process, our upscaled kickdown pump(s) must handle a minimum total gravimetric throughput of 1,000 short tons per day (TPD) of nMSW, which is volumetric throughput multiplied by the bulk density of the feedstock. Scale-up considerations For scale up estimations, cycle time is conservatively kept at 15 seconds even though commercial scale industrial piston-based pumps can achieve quicker cycle times (as low as a few seconds), especially because they are hydraulically driven and may only require one stroke per cycle. Since faster cycle times have been achieved in industry, yet the original expectation was that cycle time would increase with bore size due to greater piston inertia, the cycle times of larger kickdown pumps have been held constant at 15 seconds for the purpose of simplicity in the preliminary throughput calculations. Loading length, or hopper window length, are held at 1m, but are eventually adjusted to match the specifications of industrial piston pumps gathered during industry benchmarking stage. Table 30 lists the preliminary volumetric throughput for a kickdown pump with different bore sizes. Table 30: Volumetric throughput of a kickdown pump with variable bore size Kickdown Pump Volumetric Throughput Machine Size Diameter Loading Length Loading Volume Cycle Time Throughput (in) (in) (m3) (s) (m3/day) Prototype 2.00 4.42 0.00 15.00 1.31 4” bore 4.00 39.40 0.01 15.00 46.71 6” bore 6.00 39.40 0.02 15.00 105.10 7” bore 7.00 39.40 0.02 15.00 143.06 8” bore 8.00 39.40 0.03 15.00 186.85 9” bore 9.00 39.40 0.04 15.00 236.48 10” bore 10.00 39.40 0.05 15.00 291.95 12” bore 12.00 39.40 0.07 15.00 420.41 14” bore 14.00 39.40 0.10 15.00 572.22 D-1 Kickdown Pump Volumetric Throughput Machine Size Diameter Loading Length Loading Volume Cycle Time Throughput (in) (in) (m3) (s) (m3/day) 16” bore 16.00 39.40 0.13 15.00 747.39 18” bore 18.00 39.40 0.16 15.00 945.92 20” bore 20.00 39.40 0.20 15.00 1,167.80 22” bore 22.00 39.40 0.25 15.00 1,413.04 Density considerations/densification The GTI processed nMSW used in kickdown pump testing has a bulk density of 86.76 kg/m3. By multiplying volumetric throughput in Table 2 with bulk density, the gravimetric throughput of a single 22” bore kickdown pump is found to only reach 135 TPD of nMSW. If the loading length of a 22” bore kickdown pump is matched to that of a 22” bore piston pump found in industry (92.5”), then gravimetric throughput becomes 316 TPD of nMSW. At this capacity, the U-GAS® process would require four-multimillion-dollar machines to contend with the process’ current solids handling technology. The low bulk density of the nMSW is neither economical nor practical for kickdown pumps or lock hoppers. Low bulk density materials are prone to bridge in lock hoppers which makes them not preferable for industrial processes; they also require bulky lock hoppers or more kickdown pumps for same gravimetric throughput. The project hence recommends a densification step before feeding the nMSW to the kickdown pumps (as well as lock hoppers). Commercially available densification process such as biomass briquetting can densify the nMSW to a bulk density larger than 500 kg/m3, which is used in the nMSW gravimetric throughput estimate. Therefore, in order to deliver 1,000 TPD densified nMSW, a kickdown pump with volumetric capacity of 1,820 m3/day or two kickdown pumps with 910 m3/day each. On the safe side, the project team used 2,000 m3/day or two kickdown pumps with 1,000 m3/day each to extrapolate the kickdown pump cost. CAPEX estimate – industry benchmarking Instead of using a bottom-up estimate approach by designing a commercial scale kickdown pump and soliciting fabrication quotes, a project team decides to determine the kickdown pump cost by benchmarking the volumetric throughputs with similar industrial equipment – this procedure is less time consuming and can produce more accurate estimates. Due to higher fidelity of similarity, concrete pumps and industrial piston pumps designed to handle solids and non-Newtonian fluids such as sewage, slurries, and landfill trash are used for cost estimates. Most of the relevant piston pumps follow the design of a dual piston concrete pump, which uses two hydraulically driven pistons side by side to continuously deliver material. D-2 In this design, the pistons are out of phase so that one is filled while the other pushes material to the outlet of the machine. Besides this design difference, the price of the entire machine should match that of a kickdown pump because the hydraulic system, PLC, materials of construction, machine weight and skid requirements, and piston inertia are expected to match a kickdown pump of the same bore size. The pistons of the kickdown pump pistons are nested, but the hydraulic power requirements should be similar to the dual piston design because each full cycle of the machine includes one full stroke from each piston. The cost of ancillary hydraulic equipment used to switch the output between each piston in the dual piston design is assumed to be offset by the cost of different ancillary equipment in the kickdown pump design. The project team contacted two leaders of the hydraulic piston pump industry and obtained the prices of a few commercially available standard models. The price versus the max volumetric throughput of the machine is plotted in Figure 48 below. One datapoint represents the price of an industrial machine and the calculated throughput of a kickdown pump having the same delivery volume at a rate of 15s/cycle. Machine price as a function of the volumetric throughput of the kickdown pump exhibits a linear relationship. In contrast to Table 2 above, the length of the loading hopper matches the manufacturer specifications (the max volumetric throughput is calculated from the bore diameter of the machine, the loading length and a cycle time of 15 seconds). Figure 48: Price of a kickdown pump vs. maximum volumetric throughput Cycle time of 15 seconds and loading length using manufacturer spec (2025 cost basis). To allow some redundancy, two kickdown pumps, each having an 11.8” bore and 122” long loading window, are selected. Each such pump can deliver 1,250 m3/day or 625 metric ton per day (687.5 short tons per day) densified nMSW (assuming bulk density of 500 kg/m3). Two pumps combined can deliver 1,000 TPD densified nMSW with 30% surplus capacity. Cost of the two pumps is $1.8 million dollars (2025 cost basis). D-3 Power requirement estimate Commercial scale piston pumps utilize hydraulic power packs ranging from 5.5 to 1600 kW, but the hydraulic power pack size ultimately depends on the required output pressure and is therefore application specific. The applications requiring the greatest hydraulic power are typically concrete and mining because the pump must continually push material into a pipe that has significant backpressure. The hydraulic power requirements of the kickdown pump are expected to be lower than those of concrete pumps because densified nMSW is much lighter than concrete (around 2,500 kg/m3), and comparatively, back pressure in a kickdown pump is much less (150 psig required for U-GAS® gasifiers). Figure 49 shows data collected from two major concrete pump vendors that present a specific application where one of the models of interest is used. One datapoint represents the power requirement of an industrial machine and the calculated throughput of a kickdown pump having the same delivery volume at a rate of 15s/cycle. Based on the linear trend, the hydraulic requirement of the configuration that optimizes capital expenditure, two 11.8” bore size machines, would be less than or equal to 300 kW/machine or 600 kW. Figure 49: Kickdown pump hydraulic power demand vs. maximum volumetric throughput Scalability Summary Following the successful laboratory-scale demonstration of the kickdown pump, the project team conducted a commercial-scale scale-up and cost evaluation to assess its applicability for nMSW feed delivery to U-GAS® gasifiers. Throughput modeling indicated that a commercial system must handle a minimum of 1,000 short tons per day (TPD) of densified nMSW to match current solid fuel handling performance. Benchmarking against industrial hydraulic piston and concrete pumps revealed a linear cost correlation with volumetric throughput, leading to an estimated total capital cost of $1.8 million (2025 $) for two 11.8-inch bore units with a combined capacity exceeding 1,000 TPD. Power analysis further indicated that hydraulic requirements would not exceed 300 kW per pump, significantly lower than comparable concrete applications. D-4 These results confirm that the kickdown pump technology is technically scalable and economically competitive with existing solid-feed pressurization systems. D-5 APPENDIX E: TECHNICAL TABLE Table 31: Technical Table Parameter/ Final Description Unit Final Attained Value Performance Target FOA Required KPP Metrics A feedstock stream having a >95% purity for gasification (95% of material being convertible with specified physical and chemical properties). This is primarily accomplished through the removal of substantial inorganic and unconvertable materials in the nMSW - rocks, soil, etc. and other components Purity of stream wt. % >95 96 that do not meet minimum quality specifications for the gasification process (decontamination). Such materials will be processed through simple rock traps, size exclusion, and screening techniques, as well as materials fractionation. This is performed through density-based separations and AI control mechanical picking arms as a last separation. >50% reduction in feedstock variability as fed to the reactor (a consequence of compaction and compounding in the dry-solids pump). One major challenge in feeding and handling of nMSW are the non-ideal properties leading to inability Reduction in to convey material or handle material uniformly due to the tendency to >50 based on VM and segregate during transport leading to dynamic material properties changes [5]. % >50 variability ash Upon material flow segregation, this can impart significant variability of material composition as-fed to the gasifier. With incorporation of the active hopper and solids pump, the material is effectively compounded into a uniform feed unit and allowed for process optimization around consistent properties. In comparison to the nMSW baselining activity, the densified material plug is expected to achieve this reduction in material variability with E-1 Parameter/ Final Description Unit Final Attained Value Performance Target respect to the overall material bulk density as well as the chemical composition as determined by proximate and ultimate analysis. Decontaminatio % >80 90 n efficiency 48.60 without revenue Added cost of $/dry <30 offsets; -46.95 with technology over Delivered Cost Goal of $86/ton, including collection, handling, ton revenue offsets (2014$) preprocessing, and transportation costs Physical and numerical modeling No. Feedstock 3 To predict gas permeability (in/sec) at 150 psi through feedstocks after of permeability (blende 5 (blended) consolidation with ±20% accuracy fracti prediction d) ons No. 3 Mass flow To predict mass flow (flow volume/flow area/sec) of feedstock out of hopper of (blende 5 (blended) prediction with ±20% accuracy fracti d) ons Pump Specific KPP Metrics WBS 4.1-4.3: Measured by GIT in test rig for fractions and blends, at different <0.075 Feedstock formats (size, shape), consolidations and pressures. Convert to gas velocity in/se with 0.9-1.7 based on 4 mm permeability through material at comparable consolidation attained with moving wall test c nMWS final feedstock rig (WBS 5.1-5.3). blends E-2 Parameter/ Final Description Unit Final Attained Value Performance Target Based on 4 mm final feedstock, 403.3 for 2 >6 in/s inch hopper opening, Pump inlet feed WBS 5.1-5.3: Inlet hopper flow measured by GIT. Flow speed = volumetric flow in/se for 857.9 for 3 inch hopper rate / flow area. c nMSW opening and 827.4 for blends 4.25 inch hopper opening 150 psig to Pump discharge meet WBS 11.2: Measured by GTI in pump discharge vessel. psig 150 pressure U-GAS require ments Pump TRL WBS 5 and 11 for pump sizing, adaptation and testing. TRL 4 4 Gasifier Specific Feedstock KPP Metrics Volatile matter Range of properties as tested % 3-69 71.77 ± 3.17 Particle size Mills or crusher modifications for nMSW mm <6.7 2.74 - 5.94 Chlorine Chlorine content damage to refractory, HCL formation, Metallurgy issues in ppm <1000 Not available concentration scrubbing, catalyst poison, wastewater issues. Fluorine HF formation can cause corrosion ppm <250 Not available concentration E-3 Parameter/ Final Description Unit Final Attained Value Performance Target Inorganics speciation (CaO, Can reduce ash fusion temperature so limit the fluidized bed reactor operating % <2 Not available Fe2O3, Mg and temperature Na) E-4 1700 S. Mount Prospect Rd. Des Plaines, Illinois 60018 www.gti.energy