ABSTRACT
Battelle has demonstrated a novel and potentially breakthrough technology for a direct coal-to-liquids (CTL) process for producing jet fuel using biomass-derived coal solvents (bio-solvents). The Battelle process offers a significant reduction in capital and operating costs and a substantial reduction in greenhouse gas (GHG) emissions, without requiring carbon capture and storage (CCS). The results of the project are the advancement of three steps of the hybrid coal/biomass-to-jet fuel process to the technology readiness level (TRL) of 5. The project objectives were achieved over two phases. In Phase 1, all three major process steps were explored and refined at bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a synthetic crude (syncrude); and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, all three subsystems of the CTL process were scaled up to a pre-pilot scale, and an economic analysis was carried out.
A total of over 40 bio-solvents were identified and prepared. The most unique attribute of Battelle’s bio-solvents is their ability to provide much-needed hydrogen to liquefy coal and thus increase its hydrogen content so much that the resulting syncrude is liquid at room temperature. Based on the laboratory-scale testing with bituminous coals from Ohio and West Virginia, a total of 12 novel bio-solvent met the goal of greater than 80% coal solubility, with 8 bio-solvents being as good as or better than a well-known but expensive hydrogen-donor solvent, tetralin.
The Battelle CTL process was then scaled up to 1 ton/day (1TPD) at a pre-pilot facility operated in Morgantown, WV. These tests were conducted, in part, to produce enough material for syncrude-upgrading testing.
To convert the Battelle-CTL syncrude into a form suitable as a blending stock for jet turbine fuel, a two-step catalytic upgrading process was developed at laboratory scale and then demonstrated at pre-pilot scale facility in Pittsburg, PA. Several drums of distillate products were produced, which were then distilled into unblended (neat) synthetic jet fuel and diesel products for a detailed characterization. Based on a detailed characterization of the synthetic jet fuel, a 20% synthetic, 80% commercial jet fuel blend was prepared, which met all specifications. An analysis of the synthetic diesel product showed that it has the promise of being a drop-in fuel as super-low (less than 15 ppm)-sulfur diesel fuel.
A detailed economic analysis showed that the Battelle liquefaction process is economical at between 1000 metric tons/day (MT/day) and 2000 MT/day. The unit capital cost for Battelle CTL process for making jet fuel is $50K/daily bbl compared to $151K/daily bbl for indirect CTL, based on 2011 dollars. The jet-fuel selling cost at the refinery, including a 12% capital cost factor (which included profit), for the Battelle CTL process is $61/bbl ($1.45/gallon). This is competitive with crude oil price of $48/bbl. At the same time, the GHG emissions of 3.56 MT CO2/MT fuel were lower than the GHG emissions of 3.79 MT CO2/MTfuel for petroleum-based fuels and 7.77 MT CO2/MT fuel for indirect CTL. Thus, the use of bio-solvents completely eliminates the need for carbon capture in the case of Battelle CTL process. The superior economics and low GHG emissions for the Battelle CTL process has thus sparked worldwide interest and some potential commercialization opportunities are emerging.
1.0 EXECUTIVE SUMMARY
Battelle has demonstrated a novel and potentially breakthrough technology for a direct coal-to-liquids (CTL) process for producing jet fuel using biomass-derived coal solvents (bio-solvents). The Battelle process offers a significant reduction in capital and operating costs and a substantial reduction in greenhouse gas (GHG) emissions, without requiring carbon capture and storage (CCS). The results of the project are the advancement of three steps of the hybrid coal/biomass-to-jet fuel process to the technology readiness level (TRL) of 5. The project objectives were achieved over two phases. In Phase 1, all major process steps were explored and refined at bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a synthetic crude (syncrude); and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, all three subsystems of the CTL process were scaled up to a pre-pilot scale.
Biomass-Derived Bio-solvents. While the goal of Phase 1 was to identify and prepare 6 hydrogen-donor bio-solvents, we actually prepared a total of over 40 bio-solvents. The raw materials for these, mostly non-edible bio-solvents, are believed to be readily available around the world. In some cases, the biomass feedstocks are commercially available from specialty chemical companies though none have previously been used for CTL processing. In many cases, commercially-available feedstocks had to be modified by Battelle to provide the desired solvation and other physical properties. The most unique attribute of Battelle’s bio-solvents is their ability to provide much-needed hydrogen to liquefy coal and thus increase its hydrogen content so much that the resulting syncrude is liquid at room temperature.
Direct Coal Liquefaction. Coal liquefaction tests were done in several different reactor systems. The majority of parametric testing, especially to down-select preferred bio-solvents, was done at Battelle, using a 0.5L autoclave system. The resulting products were analyzed to determine the coal solubility, defined as the yield of THF-soluble fraction. The viscosity of the THF-free syncrude was also measured at 50°C to assess the degree of hydrogen transferred from the bio-solvent to coal-derived liquids. The initial testing was done with a bituminous coal from West Virginia, followed by more extensive testing on an Ohio bituminous coal. The various bio-solvents were compared to tetralin, which is a well-known hydrogen-donor solvent, as well as soybean oil, which was referenced in prior art. The Battelle tests showed that a total of 12 novel bio-solvent met the goal of greater than 80% coal solubility, with 8 bio-solvents being as good as or better than tetralin. On the other hand, soybean oil gave solubility below 70% and the product was very viscous. The viscosity of the product with preferred bio-solvents was an order-of-magnitude lower than with soybean oil. The solubility goal was also met with a sub-bituminous coal from the Powder River Basin. The Battelle parametric testing was supported by microcatalytic-reactor testing by Pennsylvania State University (PSU).
The Battelle CTL process was scaled up to 1 ton/day (1TPD) at a pre-pilot facility operated in Morgantown, WV by Quantex. These tests were conducted, in part, to produce enough material for syncrude-upgrading testing. The Quantex plant required several changes to adequately carry out liquefaction. The preliminary data show that the solubility and syncrude viscosity are comparable to those attained during autoclave testing at Battelle.
Upgrading the CTL Syncrude to Distillates. To convert the Battelle-CTL syncrude into a form suitable as a blending stock for jet turbine fuel, a two-step catalytic upgrading process was developed at laboratory scale and then demonstrated at pre-pilot scale facility operated by Intertek. For the first step (Stage-1), a number of commercially-available hydrotreatment catalysts were tested for removal of heteroatoms, most notably sulfur, nitrogen, and oxygen. A proprietary, sulfided catalyst was selected for demonstration. For Stage-2, two proprietary catalysts for cracking and hydrogenation of Stage-1 product were successfully demonstrated at pre-pilot scale. Several drums of distillate products were produced, which were then distilled into unblended (neat) synthetic jet fuel and diesel products for a detailed characterization.
Characterization of Distillate Fraction as Jet Fuel or Diesel. A synthetic jet-fuel produced from the Battelle CTL process was evaluated by UDRI to determine potential suitability for use in aviation applications. Efforts focused on testing of the specification and limited Fit-For-Purpose (FFP) properties of the neat synthetic fuel and a blend with petroleum-derived aviation fuel, and followed recommended protocols for certification of synthetic fuels for commercial and military applications. Analyses of the neat synthetic fuel indicated it was not feasible to use the current formulation as a direct “drop-in” fuel as a couple of the properties did not conform to required Jet A/JP-8 specification requirements. Overall, the results indicate that the synthetic fuel has the potential for use as a synthetic blending feedstock for aviation applications. Based on the analyses and testing, it appears feasible to make slight modifications to the syncrude upgrading process to better tailor the final synthetic fuel to aviation applications. An analysis of the synthetic diesel product showed that it has the promise of being a drop-in fuel as super-low (less than 15 ppm)-sulfur diesel fuel.
Economic Analysis. A detailed economic analysis was carried out to show that the liquefaction process is economical at between 1000 metric tons/day (MTPD) and 2000 MTPD, which is an order of magnitude smaller than commercially-available indirect CTL processes. The elimination of the need for gaseous hydrogen and a catalyst in the Battelle liquefaction process leads to process simplifications that greatly reduce capital and operating costs. A plant design using 4 Battelle CTL plants and a single syncrude upgrading plant producing 32,000 barrels (bbl) per day of jet fuel (and/or diesel) was compared to a 19,000 MTPD FT-technology based indirect CTL plant producing 50,000 bbl/day (BPD) of jet fuel plus diesel plus naphtha. The unit capital cost for Battelle CTL process is $50K/daily bbl compared to $151K/daily bbl for indirect CTL, using 2011 costing basis required by DOE. The jet-fuel selling cost at the refinery, including a 12% capital cost factor (which included profit), for the Battelle CTL process is $58/bbl ($1.38/gallon). This is competitive with crude oil price of $46/bbl. The selling price for the indirect CTL process was much higher at $95/bbl. An analysis also showed that the unrefined syncrude from the Battelle CTL process could also be sold to petroleum refineries at $32/bbl. No premiums were placed on either the syncrude or the jet fuel or diesel from the Battelle CTL process for having an ~40% bio-content. This however was a major factor in meeting the GHG reduction goals.
Greenhouse Gas (GHG) Analysis. A GHG emissions analysis was performed by Prof Bhavik Bakshi of The Ohio State University (OSU). The total GHG emissions from well (for petroleum) or coal mine/biomass-feedstocks source-to-wheels (e.g., well-to-wheels [WTW] for petroleum-based jet fuel) were estimated. While the baseline WTW value for petroleum-based jet is 3.79 MT CO2/MT fuel, the Battelle-CTL jet fuel GHG emissions were somewhat lower at 3.56 MT CO2/MT fuel. On the other hand, the GHG emissions on the same basis for FT-based jet fuel was much higher at 7.77 MT CO2/MT fuel. For the FT process to meet the Section 526 of EISA 2007 goal of being no worse than petroleum-to-jet baseline, about 90% of the pre-combustion GHG emissions from CTL will need to be controlled by CCS. However, no CCS will be required for the Battelle CTL process.
The superior economics and low GHG emissions for the Battelle CTL process has thus sparked worldwide interest and some potential commercialization opportunities are emerging. Thus, the project goal of demonstrating a fast, straight forward path to commercialization has also been achieved through this project.
2.0 PROJECT OBJECTIVES
Battelle has invented a potentially breakthrough direct coal-to-liquids (CTL) technology using biomass-derived solvents. The objectives of this project were as follows:
The project objectives were accomplished over a 3-year, 5-task R&D effort to advance the hybrid, direct CTL process for jet fuel to a TRL of 5. The three major Subsystems of the process – biomass to bio-solvent conversion, coal dissolution and demineralization to produce a syncrude, and hydrotreatment/ hydrogenation of the syncrude to jet fuel – were developed and tested in batch/lab-scale, bench-scale, and then at pre-pilot scale. The project objectives were achieved over two phases. In Phase 1, all major process steps were explored and refined at continuous bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a syncrude; and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, these same process steps were scaled-up to continuous, pre-pilot scale, allowing realistic estimates of process economics and GHG emissions reduction, thus defining the path for widespread process commercialization in a short time period. The process meets the requirements of Section 526 of Energy Independence and Security Act of 2007 (EISA 2007) without requiring CCS, and it should help reduce the dependence on imported petroleum crude for jet fuel production. More information on each task is provided below.
Phase 1
Task 1 - Lab/Bench-Scale Coal Liquefaction and Syncrude Hydrotreating/Hydrogenation. Several combinations of coals and bio-solvents were tested at laboratory- and bench-scale to determine preferred operating conditions for the scale-up of the coal liquefaction Subsystem to 1 ton per day (1 TPD) pre-pilot scale in Task 2. Additionally, a two-stage catalytic system was tested for upgrading the syncrude to jet fuel. Several catalysts were screened at laboratory-scale to determine the preferred conditions for scale-up to one barrel per day (BPD) pre-pilot scale in Task 3. This task was supported by four Battelle subcontractors: Pennsylvania State University (PSU), University of Dayton Research Institute (UDRI), Advanced Research Associates (ARA), and Quantex.
Task 2 – Pre-Pilot-Scale Coal Liquefaction and Syncrude Production. A 1 TPD coal liquefaction pre-pilot plant was tested at the Quantex facility in Morgantown, WV. Several hundred gallons of syncrude were produced for upgrading to jet fuel in Task
Task 3 – Pre-Pilot-Scale Syncrude Hydrotreating /Hydrogenation to Jet Fuel. The syncrude from Task 2 was upgraded to a distillate product, which was fractionated into jet fuel and diesel fractions at the Intertek facility near Pittsburgh, PA, employing catalysts and operating conditions determined in Task 1. The jet fuel was analyzed against the commercial Jet-A specifications. Some “Fuel-fit-for-use” testing was also performed, with the testing conducted by UDRI. A diesel product was also characterized.
Task 4 – Conceptual Plant Design and Process Economics. Battelle completed a comprehensive conceptual plant design and utilized the design for a techno-economic analysis (TEA), following DOE/NETL guidelines.
Task 5 – Greenhouse Gas (GHG) Emission Analysis. A GHG life-cycle emissions analysis was performed, using DOE/NETL and USAF guidelines, to demonstrate progress towards meeting requirement of Section 526 of EISA 2007.
This is a Final Report on the 2-phase project.
4.0 DIRECT CTL BACKGROUND AND DESCRIPTION OF THE BATTELLE CTL PROCESS
4.1 Direct CTL Background
The U.S. DOE-supported direct CTL programs in the 1970s and 1980s included solvent-refined coal (SRC), H-Coal, Exxon Donor Solvent (EDS), and other two-stage liquefaction processes. The DOE’s focus was on direct CTL since it had a significantly higher thermal efficiency than indirect CTL using the coal gasification plus Fischer-Tropsch (FT) route. Most R&D and demonstration efforts were stopped in the late 1980s due to escalating cost estimates for CTL and very low petroleum crude prices. During the last decade, CTL has received renewed interest, though the focus of recent R&D has been indirect CTL. In many respects, direct CTL is more appealing than indirect CTL relative to the priority objectives of this project. First, it is possible with direct CTL, by carefully dissolving, depolymerizing, and hydrotreating coal, to produce acceptable jet fuel without blending in petroleum-based jet fuel. Indirect CTL, e.g., via FT route, converts a highly aromatic coal structure to a linear, paraffinic structure, which is unacceptable as JP-8 or Jet-A, as those fuels require a minimum of 8% aromatics, and hence must be blended with petroleum-based jet fuel, per Military (Mil-DTL-83133H w/Amendment 1–Tables A-II & B-II) and Commercial (ASTM D7566-12A–Table 1, Part 2) specifications. Second, a carefully controlled direct CTL process, such as the one Battelle is developing, is thermodynamically more efficient in terms of yield than the indirect CTL approach of converting coal to synthesis gas and then recombining to make condensable liquids. The higher expected thermal efficiency of direct CTL drives a reduction in greenhouse gas (GHG) emissions per gallon of jet fuel produced.
4.2 Battelle’s Novel Direct CTL Process
Battelle has developed a hybrid CTL process that uses a significant amount of biomass for direct CTL in its innovative approach. The hybrid CTL process includes three basic steps: (a) biomass conversion to a high hydrogen-donor (H-donor) bio-solvent; (b) coal dissolution in novel biomass-derived solvent without molecular H2; and (c) syncrude hydrotreating/hydrogenation to jet fuel.
State-of-the-art direct CTL processes first quickly depolymerizes/dissolves the coal, typically in a coal-derived solvent, and then these slowly hydrotreat the solution to break up large molecules, remove heteroatoms (e.g., S, N, O), and increase the H/C atomic ratio. The resulting syncrude can be further refined by hydrotreating to various distillate fuels, a portion of which may be jet fuel. Battelle has investigated ways to overcome several disadvantages of current direct CTL. First, the GHG emissions for current CTL jet fuel are about twice that for petroleum jet fuel, so it would be necessary to capture 90% CO2 at the CTL plant to meet the CTL GHG emissions reduction goal. Second, a straightforward process for dissolving coal and biomass in a solvent has not been practical, partly because of the high moisture content of biomass. Third, the H/C atomic ratio in a typical bituminous coal is about 0.80, while it is about 1.90 for jet fuel, so a large amount of H2 must be added to coal, which contributes to high GHG emissions, and drives up capital and operating costs for current direct CTL processes. In particular, with processes that both dissolve coal and hydrotreat in one step, the high temperature (~450°C) and pressure (~2500 psig) requirements make the reactor costs uneconomically high, cause equipment erosion, and utilize H2 inefficiently due to excessive production of lighter hydrocarbons. Finally, the yield of jet fuel vs. diesel is significantly less in these current direct CTL processes than for FT-based jet fuel production.
To address these concerns, Battelle has developed a hybrid CTL process that uses a significant amount of biomass for direct CTL in an innovative approach, as shown in Figure 1.
As currently conceived, the Battelle Direct CTL process has three major Subsystems including:
(1) biomass conversion to high hydrogen-donor solvent; (2) coal dissolution in biomass-derived solvent without molecular H2; and (3) 2-stage catalytic hydrotreating/hydrogenation to jet fuel and other distillates. In Subsystem 1, biomass, derived primarily from non-food sources, is converted to a bio-solvent in processing plants that are economical at smaller scale (≤100 TPD). The resulting bio-solvents, with a H/C atomic ratio above about 1.40, are delivered to a larger coal dissolution/demineralization facility (≥1,000 TPD) in Subsystem 2. Based on data on solvent refining of coal published by Longanbach and Chauhan of Battelle in late 1970s, and more recently confirmed by West Virginia University as well as by Battelle project team member Quantex, coal can be dissolved quickly (in ˂10 minutes) at mild conditions (~400°C and 500-800 psig) with addition of only 0.3–0.5% hydrogen, by weight of coal, which increases the H/C molar ratio from about 0.80 to 0.86 [1-4]. The bio-solvents can be engineered to alter the nature and quantities of the cyclic/aromatic and linear species with desired hydrogen-donor properties.
Based on the low hydrogen-addition requirements to dissolve coal, as little as 10% of bio-solvents based on weight of coal is sufficient without requiring any gaseous H2, and thus minimizing CO2 emissions at the CTL plant (Subsystem 2). However, larger amounts of bio-solvent input will not only help meet the GHG reduction goal without CO2 capture, but also reduce the viscosity of the synthetic crude (syncrude), improving separation of ash and unconverted coal. Further, the larger quantity of bio-solvent helps produce a jet fuel that has a more manageable balance of aromatic and non-aromatic species, thus overcoming a significant limitation of current direct CTL. The solvent/coal weight ratio is expected to be about 2.5, so a portion of the solvent, including some coal-derived liquids, will need to be recycled after solid/liquid separation.
In Subsystem 3, the syncrude from the liquefaction plant can be catalytically hydrotreated/hydrogenated at either a petroleum refinery or a facility dedicated to maximizing jet fuel production. This is the location where all the molecular H2 will be added. The overall, gaseous H2 requirements for the Battelle CTL process are expected to be substantially less than for the H-Coal or EDS processes. We also expect the yield of jet fuel to be much higher than from H-Coal or EDS processes.
Battelle’s approach is to use novel, biomass-derived solvents with high hydrogen-donor capacity. These bio-solvents can be engineered to have significant amounts of cyclic/aromatic compounds (>20%) and a controllable H/C ratio, with good hydrogen-donor capabilities. The objective was to achieve as good hydrogen donor performance as with the well-known solvent tetralin, used in the EDS process as well as tested for direct CTL for the last 75+ years, but with solvents that are biomass-derived and cheaper than tetralin.
4.3 Potential Benefits of the Battelle Direct CTL Process
Battelle’s hybrid CTL process offers these specific advantages: (a) straightforward system integration of proven process steps; (b) improved process reliability due to mild liquefaction operating conditions (less than 800 vs. 2500 psig); (c) elimination of CCS at coal liquefaction site as well at the syncrude refining site; (d) significant reduction in molecular H2 requirement for syncrude refining; (e) increased aromatic content and density of jet fuel close to that of JP-8; (f) short time period to commercialization; (g) significant reduction in the capital and operating costs; and (h) substantial reduction in GHG emissions.
The Battelle CTL technology helps achieve the GHG emissions reduction goal of this project, unlike state-of-the-art CTL technologies, without requiring CCS. The key reasons for this are as follows:
5.0 LAB/BENCH-SCALE COAL LIQUEFACTION
The objectives of this effort were as follows:
The Battelle CTL process combines coal, coal-derived recycle solvent, and an additional bio-solvent, and then heats the agitated mixture at elevated temperatures and pressures. After the mixture is cooled, mineral matter and undissolved organic matter in coal are removed in a solid-liquid separation step. Finally, the de-ashed, liquefied coal is fractionated in a distillation column to separate the liquefied coal or synthetic crude (syncrude) from the recycle solvent and heavy oil. The heavy oil may be sold as a by-product, or coked to recover more liquefied coal and the coke sold. If desired, the heavy oil may also be hydrocracked with the rest of the syncrude. Other potential, high-value, products from the heavy oil are: (a) binder pitch; (b) anode-grade coke; (c) needle coke; or (d) polyols or other chemicals. The centrifuge cake may be sold as an asphalt additive, burned to generate heat or power, or gasified to generate syngas.
Presented below is a brief discussion of the selection of the three major feedstocks.
5.1.1 Coal
Three coals were selected: (1) a West Virginia (WV) high volatile A, bituminous coking coal1 from Leer Mine; (2) an Ohio high volatile A, bituminous, coal from Waterloo Coal Company (sample obtained from Bramhi Coal Company), and (3) a Powder River Basin sub-bituminous coal from the Black Thunder Mine. One ton each of the first two coals were acquired and a portion ground to -25 mesh size at Quantex. A drum of the Black Thunder subbituminous coal also was obtained and ground to -25 mesh size at Quantex.
WV Bituminous Coal. During the previous West Virginia University (WVU) coal-liquefaction program, they successfully processed a Lower Kittanning seam, high-volatile A, coking, bituminous coal. Unfortunately, the coal mine in West Virginia where this coal was obtained from has now closed. In reviewing options, Quantex identified a coal from a Lower Kittanning seam about 30 miles from the original site. This coal is from the Leer Mining Complex, located in the town of Grafton in Jackson County, WV; a map showing the location of the mine is presented in Figure 2. The mine is owned by Arch Coal Inc. One ton of this coal was acquired. Four 5-gallon pails of the coal were initially ground by Quantex to a size smaller than 25 mesh (designated -25 mesh) to support chemical analysis and small scale liquefaction tests. More of this coal was ground to -25 mesh to support the Quantex 1 ton per day (TPD) unit testing.
1 A high volatile A bituminous coal has a fixed carbon content, on a moisture and ash (MAF) basis, of less than 69 wt%, and volatile matter content, on an MAF basis, of greater than 31 wt%, and a higher heating value (HHV) equal to greater than 14,000 Btu/lb on an MAF basis.
Ohio Bituminous Coal. There are three major seams that underlie southeastern OH, western PA and northern OH; see Figure 3. They include the Pittsburgh, Upper Freeport, and Lower Kittanning seams. The Lower Kittanning seam is the largest and is the same seam as was found successful in the WVU liquefaction tests that employed hydrogenated soybean oil [3, 4].
Therefore, the team looked for a Lower Kittanning seam, high-volatile A, coking, bituminous coal in OH and identified existing mines with coal preparation plant in the area. Coal from the Waterloo Coal Plant, was identified. The location of the preparation plant is shown in Figure 4. Leer Coal does not sell directly to the public, but a vendor (Bramhi Coal Company) was identified. Four pails of the Ohio coal were initially ground by Quantex to -25 mesh for analysis and testing. More of this coal was ground to -25 mesh to support the Quantex 1 TPD liquefaction unit testing.
Wyoming Subbituminous Coal. WVU also successfully treated lower rank, low-ash coals in their liquefaction process. One of the subbituminous coals processed was Black Thunder coal, surface mined in Wright, WY. The location of the mine is shown in Figure 5 (this figure also shows the mine is located in the Powder River Basin covering southeast Montana and northeast Wyoming).
Fortunately, this coal is still commercially mined by The Black Thunder Coal Company (owned by Arch Coal, Inc.). A 55-gallon drum of a fresh sample of this coal was obtained.
The Leer (WV) and Waterloo (OH) coals were analyzed for ultimate and proximate analysis, sulfur forms, and Free Swelling Index. The Geisler plasticity was also determined to provide a measure of how easily the coal softens and flows upon heating. Results are shown in Table 1. The analyses are reported on a dry basis, except for moisture content and hydrogen, for which the as received (AR) value is also reported. Typical Black Thunder analyses (obtained from Arch Coal, Inc.) are also included in this table.
Coal parameters expressed on a dry or MAF basis are often used to characterize coal. The coal values on these bases are noted in Table 2.
5.1.2 Coal-Derived Recycle Solvent
Coal tar distillate (CTD) is used to start the liquefaction process. After a period of time sufficient coal liquids will be generated in the process so they can displace the CTD and very little (≤ 10% based on wt of coal) additional CTD will be required. For the initial liquefaction tests at Battelle, PSU, and Quantex, a Koppers CTD was used. The typical Koppers CTD, used to make carbon black, has a boiling range of 230ºC to 360ºC. However, the Koppers CTD supplied to Quantex a couple of years ago has about 50% boiling above 400ºC, as indicated by the simulated distillation (SimDis) test conducted by PSU (see Figure 6). These two start-up solvents were compared so we could select the correct one for further testing. In any case, this CTD is expected to contain significant quantities of cresol, naphthalene, naphthol, and anthracene.
5.1.3 Bio-solvents
In the initial work at West Virginia University (WVU), a combination of coal, CTD (recycle solvent), and hydrogenated soybean oil was used to liquefy the coal [3, 4]. It is believed by Quantex that the soybean oil facilitated coal depolymerization and the resulting coal dissolution. On the other hand, in Battelle’s direct coal liquefaction process, we use a bio-solvent that helps dissolve coal as well as serves as a hydrogen donor to species generated during coal depolymerization/ dissolution. In this fashion, we expect to increase the H/C atomic ratio of coal-derived syncrude without using any molecular hydrogen (H2). One example of a well-known hydrogen donor-solvent is tetralin, which is expensive and typically not bio-based.
A major effort on this project was the selection, procurement, modification, and liquefaction screening of a number of bio-solvents. Over 40 bio-solvents were prepared and screened. The majority of these bio-solvents were prepared utilizing non-edible biomass, employing proprietary treatment methods. Originally, about 6 bio-solvents were targeted as the testing began with the West Virginia coal. Next, the Ohio coal was tested, during which time many more bio-solvents were identified or became available. As such, most of the bio-solvents were tested on Ohio coal. A very limited amount of testing, with a preferred bio-solvent, was carried out on the Wyoming coal. Some of the bio-solvents were a mixture of two bio-solvents to achieve better hydrogen transfer as well as to facilitate coal depolymerization.
5.2 Laboratory-Scale Coal Liquefaction Testing
5.2.1 Test Objectives
The goal for this effort was to demonstrate the feasibility of Battelle’s hybrid, direct CTL process for producing a syncrude using novel bio-solvents without using any molecular H2. The primary liquefaction testing was performed in Battelle’s 0.5 L high temperature, high pressure autoclave in order to achieve a coal solubilization of 80% or greater. Microreactor liquefaction tests were also run at PSU in parallel to Battelle’s work to better understand the kinetics of the liquefaction process.
5.2.2 Lab-scale Coal Liquefaction Test Procedures.
The test procedure for the autoclave tests is shown schematically in Figure 7.
Shown in Figure 8, below, is a photo of the autoclave as well as the 0.5-L cup (i.e., the base of the autoclave).
After a run, water was passed through a Hastelloy-C coil located within the reactor. This cooled the liquids and allowed us to reduce the pressure to atmospheric level. The retaining bolts were unscrewed and the autoclave head that held the mixing assembly and cooling coil was removed. A low-boiling inert liquid (tetrahydrofuran, THF) was sprayed onto the head, mixing shaft and propeller, and the coil to recover all remnants. Additional THF was added to the cup (the more viscous the product, the more solvent was used).
The contents of the cup were also transferred to a 1 L Erlenmeyer flask. The solution was heated to boiling while being stirred to separate the liquid fraction from the solid particles. This step also speeded up filtration. Once boiling was achieved, it was poured into the top of a pressure-filter. In some cases the Syncrude produced after liquefaction was low viscosity, water-like liquid and was easily transferred. In other cases the product was a thicker, more viscous fluid. In these cases, THF alone could not reduce viscosity enough for effective filtration. For this reason, hot (150°C) dimethylformamide (DMF) was utilized. The recovered liquid and undissolved coal and mineral matter were pressure filtered at up to 40 psig (easy to filter products required less pressure to filter). More inert fluid, THF or DMF, was used to try to recover as much coal material as possible from the filter. The filter cake was washed again with THF and the cake placed in a vacuum oven to remove the wash solvent. The filtrate was placed in an evaporating flask and put in a specialized vacuum heated apparatus called a Rotovap operated at 50°C and 20 amHg (20 Torr) for 1 hour. If DMF was used, a specialized Rotovap, called a Kugelrohr, was used at 80°C and ~1 mm Hg pressure. The solvent-free filtrate and filter cake were then weighed.
Provided below is an example run. The following was weighed out into the autoclave cup:
Solubility of coal is defined as 100 minus (the “quantity of filter cake minus coal ash fed” divided by the quantity of moisture, ash free (MAF) coal fed) times 100, or
Or, in more detail
Using the numbers in our example, solubility was calculated as follows:
The filter cake was ashed at 500°C for 240 minutes. The ash content was determined to be 36.9 wt%. We compared this measured ash value to a theoretical solubility versus ash content curve, see Table 3 and Figure 9. The mass balance for this example was 96.4%.
This relationship was calculated by assuming (1) the mineral matter in the coal equals its ash content, (2) 100 % of the ash in the coal ends up in the filter cake, (3) all of the non-solubilized coal ends up in the filter cake, and (4) no residual coal liquids are retained with the filter cake. So, for 0 wt% solubility, all of the coal would end up in the filter cake and the predicted ash content would be the moisture-free ash content of the coal (8.48 wt%). At 100% solubility, only the ash would be retained in the filter cake, and the ash content of the filter cake would be 100 wt%.
For the measured 36.9 wt% ash content, the predicted solubility was between 84% and 85 %, which is in reasonably good agreement with the calculated value of 87.12 %.
5.2.3 Results for West Virginia (WV) Coal
The reactions were performed following the autoclave procedure described above. The coal was minus 25 mesh with no further treatment. We initially began running reactions at 400ºC for 30 minutes with WV coal and Koppers CTD but found that the resulting syncrude was very viscous and only one candidate bio-solvent performed reasonably well. Because of the higher viscosity of filtrate, we decided that a higher temperature was needed to help depolymerize the coal to a greater degree. We next ran at a temperature of 425ºC. Tetralin performed well at this temperature yielding 86.8% dissolution of coal. However, after running the best bio-solvent at that point (BS-3) at times of 30 min, 60 min, and 90 min, we found that we were getting some coking. This was supported by previous observations at Quantex indicating they had seen coking at temperatures greater than 420ºC. This led us to testing at 410ºC. With Leer coal and Koppers CTD, the best bio-solvent yielded a dissolution value of 86.7%, which is same as for tetralin. At 415ºC the same material gave a dissolution of 85.5%; essentially the same value as at 410ºC. Under these same conditions Tetralin was run twice and gave dissolution values of 84.3% and 86.6% at 415ºC and 410ºC, respectively. From this point onwards, we selected 410ºC as the standard operating temperature for feedstock screening. The liquefaction yields for the several bio-solvents, using WV coal, are compared with soybean oil and tetralin in Table 4. The yields for other bio-solvents were below 50%.
The data on WV coal showed that the percent solubility of coal, as defined by solubility in hot THF, as well as the viscosity and filtration behavior of the product are highly dependent on the solvent used. For example, the product based on use of tetralin was of a lower viscosity and the product work-up, including dissolution in hot THF and filtration were easy. At the same time, the solubility with tetralin was quite high at 86.6%. Soybean oil base case gave a relatively good dissolution yield at 76.5%, but a stronger solvent (DMF) had to be used and the product was very viscous. After re-extracting the filtrate with hot THF, we found that 23.0% of the filtrate was THF insoluble. Thus, the corrected THF solubility with soybean oil was only 58.9%. On the other hand, three bio-solvents, BS-3, BS-9, and BS-15, provided as high dissolution yields as with tetralin and the products were of lower viscosity and relatively easy to dissolve in hot THF. This result was expected since soybean oil is not an H-donor.
Based on the initial results from 31 autoclave tests, BS-3 (Autoclave Test No. WV-13) and BS-15 (Autoclave Test No. WV-34) appeared to be the best bio-solvents. One of these tests, WV-13, was selected for further analysis of the filter residue and syncrude, which included a portion that will normally be returned to the front-end of the process as the CTD recycle solvent. The ultimate analysis of the filter residue and syncrude, are compared with that of coal and CTD solvent in Table 5.
A mass balance for Test No. WV-13 shows about 96% closure for filter residue, syncrude, and moisture from feed. The remaining 4% includes losses including some C1-C3 gases vented from autoclave. Based on elemental mass balances, the estimated composition of syncrude, free of CTD solvent, compared with coal, is shown in Table 6.
The results on the WV coal show that coal liquefaction resulted in 98% mineral matter removal and 34% oxygen removal. The calculation for sulfur removal is complicated as the bio-solvent, which becomes part of the syncrude, is free of sulfur. However, based on the types of sulfur present in coal, it is predicted that the liquefaction process removes only the pyritic and sulfate sulfur, but not the organic sulfur. Similarly, no nitrogen removal is expected, as all of the nitrogen is organic nitrogen. The H/C atomic ratio increased by about 25%.
5.2.4 Results for Ohio Coal.
After completing the tests with WV coal, we began testing the Ohio coal. A total of 151 batch tests were carried out. The primary reason for more testing on Ohio coal was that many more bio-solvents became available during Ohio coal testing.
The preferred bio-solvents were compared with tetralin and soybean oil, as shown in Table 7.
As shown in Table 7, seven (7) bio-solvents performed as well as or better than tetralin. A total of 11 bio-solvents gave over 80% coal solubility, compared with only 66.9% for soybean oil, which served as the base case. The BS-19A and BS-32 bio-solvents were actually used for larger-scale testing at Quantex. The liquid product from the BS-19A test, called Syncrude 2, was evaluated for lab-scale hydrotreating/hydrogenation. Based on cost considerations, two bio-solvents, BS-27B and BS-32, were down-selected for scale-up in Task 1.03.
5.2.5 Results for Wyoming Coal.
Black Thunder coal was found to be 6% ash and 22.2% water. We ran four tests through our dissolution process. After correcting for moisture content, we ended up using a liquid to coal ratios of 2.5-2.8 and ran at temperatures of 400°C or 410°C for 30 minutes. The bio-solvents used were soybean oil, BS-19A, BS-27, and BS-32. The soybean oil run could not be worked up due to a very thick tar product plugging the filter. We tried using hot THF and hot DMF but the material would still not filter. The reaction with BS-19A went well and produced a low viscosity oil. But, even with THF this material would not filter due to the solid material being more like a tar/thick oil than a solid as seen with other coals. It is believed this reaction would do well with a centrifuge, as used in continuous tests at Quantex. The solid gummed so well that the liquid solvent mixture was decanted. The solid portion was rinsed with hot solvent and continued to show a tar-like consistency. This test with BS-19A calculated to 91.9% solubilized coal. The filtrate was of very low viscosity and this would be expected due to the fact that we did not remove the water from the coal, making our real liquid to coal ratio equal to 2.6 instead of 2.0.
5.2.6 Parametric Testing Results
Effect of Time and Temperature. During early studies on WV coal, temperature was varied from 400°C to 425°C. We began our work at 400°C, but found that we could only obtain good data for one candidate, the BS-3 solvent. The soybean oil and other candidates produced too thick of tars even without removal of the extraction solvents. Because of these issues, we then switched and ran at 425°C. This temperature gave lower viscosity oils, but also gave lower yields. When time was increased from 30 to 90 minutes on the BS-3, the solubilization dropped by 40.0 percent. This meant that we were coking our mixtures at 425°C. This was supported by WVU work which saw coking above 420°C. On the other hand, at 410°C, the same BS-3 solvent solubilized 86.7% of coal. When this temperature was increased to 415°C the solubilization was 85.5%. Therefore, at that time, we settled on a reaction temperature of 410°C. Later on, while working with the Ohio coal, we decided to evaluate lower temperatures due to the belief that no coking would take place at 400°C and below. Based on PSU work (see Section 5.2.7), we also knew that solubility was better at lower temperatures, but typically at longer times. Due to lack of the normal coal tar distillate at the time, we had to use a light coal tar distillate which was known to perform less favorably, but was believed to be acceptable for providing the trends relative to time and temperature. The results are in Table 8 below.
As seen above, there was little difference in the solubility between 390°C at 30 minutes and 400°C at 10 and 30 minutes. However, lower solubility was observed at 390°C and 10 minutes. This suggested that time is more critical when working at lower temperatures. Once we had obtained more of the desired coal tar distillate (CTD), we repeated two tests shown above. The results are shown in Table 9 below.
As expected, the solubilities at 400°C for 30 minutes were quite good with the new, heavier CTD. The solubility was good even at 390°C when time was extended to 45 minutes. However, the reaction is likely not complete as the viscosity was still high. If lower temperature (e.g., 390°C) is desired, a time of 60 minutes or greater is likely needed.
Effect of Bio-solvent. The various bio-solvents have varying degrees of performance based on their potential to dehydrogenate, and thus serve as an H-donor. As seen in Table 7, BS-15 and BS-19 seem to have the greater potential to reduce the final viscosity of the syncrude. Also, based on the molecular structures, BS-9, BS-15, BS-26, BS -27, and BS-32 have the highest potential for dehydrogenation, so it was expected that they would perform well, and which they did. Another surprising matter is that the physical state of the bio-solvent before the reaction does not necessarily effect the final syncrude state. For example, BS-24, BS-27, and BS-19 are solids at room temperature. They were blended with soybean oil in order to maintain them as liquids. However, the non-blended samples produced lower viscosities than after blending with soybean oil. This further confirms that soybean oil, which was tested by WVU and therefore served as a base case, doesn’t transfer hydrogen to depolymerized coal unlike BS-19, BS-24, and BS-27. Some selected comparisons of the novel bio-solvents with soybean oil relative to coal solubility and viscosity of syncrude are shown in Table 10.
Table 10. Effect of Bio-Solvent Type on Coal Solubility and Syncrude Viscosity for Ohio Coal.
Solvent Recycle. We evaluated replacing Koppers CTD with our own syncrude five different times. When we took our syncrude from Run OH-43 and replaced 100% Koppers CTD in run with BS-19A, the coal solubility dropped from 84.5% to 51.0%. Similarly, when we took syncrude from Run OH-60 with BS-24 and replaced 50% Koppers CTD in run with BS-24, the coal solubility dropped from 76.6% to 48.3%. However, when we took syncrude heavies from Quantex Run #3 with BS-19A and replaced 50% Koppers CTD in run with BS-19A, the coal solubility increased from 82.0% to 89.7%. Similarly, when we took syncrude from Quantex Run #3 with BS-19A and replaced 100% Koppers CTD in run with BS-19A, the coal solubility increased from 82.0% to 84.7%. The reason the first two failed when compared to those run with Quantex Run #3 heavies is likely due to the low-boiling fractions being included in the first two. The Quantex Run #3 heavies is the material that passed through the wiped film evaporator (WFE) to remove everything boiling below ~290°C. This means that the material contains everything >290°C boiling point. It is known that heavier coal tar distillate typically solubilizes coal better than lighter fractions. By removing the lighter fractions, we made a better solvent out of our syncrude. Replacement of 10% of the BS-24 with hydrotreated syncrude from Quantex Run #3 improved the coal solubilization from 76.6% to 80.1%, which is due to production of tetralin type of molecules during hydrotreatment. Finally, in Run OH-140, we replaced 50% of the Coppers CTD with centrate from Quantex Run #6A and used Stage-1 product from syncrude hydrotreatment as a H-donor solvent. This reaction resulted in 85.2% coal solubility, indicating that the bio-solvent can be regenerated via syncrude hydrotreatment.
5.2.7 Microcatalytic Reactor Testing at PSU
Test Equipment and Procedure. A series of coal dissolution experiments were performed in vertical tubing reactors (microreactors). Having an internal volume of approximately 25 milliliters (mL), these microreactors are fitted with a pressure gauge, adjustable pressure relief valve and isolation valve. A photograph of a microreactor is shown in Figure 10.
A blend of coal, bio-oil, and coal-derived solvent, having a total weight of approximately 7.50 grams, was loaded into the reactor’s body for each test. The reactor was then connected to a gas manifold, pressurized with nitrogen, and tested for leaks. The reactor was vented and again pressurized. This step was repeated two times to purge all air from the reactor. Finally, the reactor was pressurized to the reaction pressure and the pressure relief valve adjusted as necessary. The reactor was then plunged into a preheated sand bath held at the desired reaction temperature. The bath temperature was measured by a thermocouple placed in close proximity to the reactors. Each reactor and its contents were agitated at approximately 200 cycles per minute to promote mixing between the coal, bio-solvent, and coal-derived solvent throughout the test.
After the reaction time had been reached, the reactors were removed from the sand bath and quenched in cold water. Product work-up was similar to typical coal liquefaction product work-up [5]. Any gas products were vented and the solid or liquid by-products were flushed from the reactor with THF. The solution was sonicated for 20 minutes to promote dissolving of the coal dissolution products. THF insolubles were then separated from solution by filtering the THF through glass fiber filter paper having particle retention of 2.2 µm. The filter paper and solids were allowed to dry for 24 hours in a fume hood before being transferred into a glass fiber thimble. The thimble had particle retention of approximately 0.8 µm. Solvent extraction was performed for 24 hours using the THF collected during the previous filtration. The thimble and its contents were dried for 24 hours in a fume hood and then dried under vacuum for at least 1 hour at 110°C. The final weight of the THF insolubles was determined and used in Equation 1 to calculate the percent conversion of the coal on a dry, ash-free basis.
The THF solvent was removed from the THF solubles by rotary evaporation combined with final evaporation under vacuum for 1 hour at 70°C.
Coal Dissolution Results. A total of 38 coal dissolution experiments were performed in the microreactors. These experiments were intended as a screening tool to evaluate the effect of different bio-oil solvents, coal-derived solvents, chemical solvents, and operating conditions (reaction temperature and reaction time) in promoting coal conversion. The operating conditions and coal conversion (on a dry-ash free basis) for each microreactor experiment is provided in Table 11. A slight correction (0.1 to 0.2 weight %) for coal conversions reported in the previous quarterly report was made to adjust for coal analysis subsequently determined at PSU. Only the Leer (WV) coal was used in the microreactor testing. The Bramhi (Ohio) coal was used in the large lab reactor testing discussed in Section 5.3.
The first microreactor test was performed using soybean oil blended with QRS coal-derived solvent. Attempts to separate the solid and liquid products from this test using dichloromethane were not successful, as also observed in Battelle’s testing. Therefore, a decision was made to perform the next series of tests using BS-2 followed by recovery with THF. With the exception of PSU-4, the first five tests in this next series were performed using BS-2 blended with the QRS coal-derived solvent. These initial tests were used to develop the protocol for recovering the coal dissolution products from the microreactors and separating the unreacted coal and mineral matter from the liquid products. The highest level of coal conversion using BS-2, 50.5 weight % MAF), was achieved at 385°C with a reaction time of 60 minutes.
Tetralin, a hydroaromatic solvent capable of donating hydrogen to cap free radicals, was used in PSU-4 to compare with the coal conversions achieved with the bio-oils. This test resulted in the highest coal conversion (78.1 wt% MAF) achieved in the microreactor experiments.
Although the microreactors were agitated during each test, a concern was raised regarding the effect of reactor orientation on mixing and coal conversion. Previous coal dissolution studies have used both vertical (see Figure 10) and horizontal reactor bodies. To evaluate the effect of reactor orientation, an experiment (PSU-5) was performed in a horizontal reactor body. The coal conversion of PSU-5 (conversion – 48.2 wt% MAF was compared with PSU-3, the same test performed in a vertical reactor body (conversion – 50.5 wt% MAF). The difference between these tests was 2.3 wt% with the vertical reactor achieving the higher coal conversion. Because the vertical reactor orientation achieved the higher conversion, it was determined that this design did not limit internal mixing and coal conversion. Therefore, continued dissolution testing was performed in the vertical reactor design. Also, an analysis of the liquid products by GC SimDis confirmed that the reactor’s orientation did not significantly affect the liquid product’s boiling point distribution.
Testing performed at Battelle determined that the coal conversions achieved using BS-3 were greater than those with BS-2. Therefore, Battelle requested that all further testing be performed using BS-3. Several series of tests were then performed varying the reaction temperature, reaction time, and coal-derived solvents in combination with BS-3. Figure 11 shows a plot comparing the percent coal conversion versus temperature for different coal-derived/bio-oil solvent combinations. These tests were all performed at a reaction time of 30 minutes. The data indicate that the best temperature to run reactions using either coal-derived solvent (Koppers or QRS) with BS-3 is 400°C for a 30-minute reaction time. Lower coal conversions were achieved at 385°C, while no significant increase in coal conversion was gained by running at 415°C. While no advantage in coal conversion was gained by operating at 415°C, an analysis of the liquid product’s boiling point distribution also showed no benefit at the higher temperature.
Two series of experiments were also performed at a 30-minute reaction time replacing BS-3 with tetralin. The purpose of these tests was to evaluate the benefit of introducing a strong hydrogen donor solvent. Only a slight improvement in coal conversion was observed at 385°C. However, a greater increase in coal conversion was seen with tetralin at higher reaction temperatures (400°C and 415°C). This increased conversion can likely be attributed to the donation of additional hydrogen from the tetralin to free radicals, thus limiting coking of the coal and bio-oil mixture.
Figures 12 and 13 plot the percent coal conversion versus reaction temperature for each reaction time studied. Figure 12 compares tests performed using a solvent blend of Koppers and BS-3, while Figure 13 compares tests performed using a solvent blend of QRS and BS-3. These show that the coal-derived solvent from Quantex (QRS) performs better than the coal-derived solvent provided by Koppers. It also appears that a reaction time of 30 minutes is insufficient to achieve a maximum coal conversion at 385°C. While the shorter reaction time requires higher reaction temperatures, the coal conversion at longer reaction times (60 or 90 minutes) decreases at higher temperatures. This observation can possibly be attributed to repolymerization of the liquid products or thermal degrading of the bio-oil.
The best reaction condition appears to be when using tetralin with QRS (70.4 wt%), but the bio-solvent used thus far was getting decent results (~60-65 wt%). An additional bio-solvent (BS-19A) was tested in the microreactors at a solvent-to-coal ratio of 2:1 and higher. Although the coal conversions achieved with BS-19A were not as high as those observed with BS-3, the higher solvent-to-coal ratio did yield increased conversions. The higher solvent-to-coal ratio of 3:1 was also tested with tetralin (66.4 wt%). This test yielded similar increases in coal conversion to those using BS-19A.
The final series of tests performed in the microreactors was a determination of repeatability for this type of experiment. The repeatability was determined by performing an experiment in triplicate (PSU-13-1, PSU-13-2, and PSU-13-3) using the Koppers coal-derived solvent and BS-3 at a solvent-to-coal ratio of 2:1. The mean coal conversion for these three tests was 61.2 wt% with a repeatability of ± 0.6 wt%. This value can be used when comparing the coal conversion of a dissolution test with any other dissolution test.
5.3 Bench-Scale Coal Liquefaction Testing at PSU
5.3.1 Test Equipment and Procedures.
A bench-scale coal dissolution system was constructed by PSU for this task. The system, shown in Figures 14 and 15, contains three one-liter continuously stirred tank reactors (CSTR(s)) operated in parallel to increase the processing volume. These larger reactors permit greater quantities of coal dissolution liquids to be produced for further analysis or upgrading. Each CSTR was equipped with a magnetic stirrer that had two impellers mounted on the internal shaft. The mixer’s speed was continuously monitored by a tachometer. The reactors were housed within an electrical heater that provided the energy to heat the reactor and coal/coal tar distillate/bio-oil slurry up to the reaction temperature. The internal reaction temperature was monitored throughout the test by a type-K thermocouple inserted into a thermowell located inside the reactor.
A test was performed in the large reactors by individually filling one of the CSTRs with approximately 800 grams of coal/coal tar distillate/bio-oil slurry. This is equivalent to roughly 750 mL in volume. The slurries were manually prepared and loaded into each CSTR. The reactors were then closed, the headspace purged with nitrogen, and the reactor pressurized with additional nitrogen up to 600 psig. The reactor was heated to the reaction temperature and held there for the required residence time. After the residence time had been reached, the dissolution products were drained into a product/settling tank through heated transfer lines (see Figure 15). Light ends were allowed to pass from the product tank, through a heat exchanger, and were collected in a separate condensate tank. Air loaded backpressure regulators were initially used to maintain pressure within each CSTR prior to venting to the atmosphere. The backpressure regulators were subsequently replaced by manual metering valves for much of the large reactor testing. The filter, shown in Figure 16, allows pressure filtration of the liquid products to remove unreacted coal and mineral matter greater than 10 µm in size. Product from either shakedown runs or actual tests performed at a solvent-to-coal ratio of 2:1 were too viscous to be filter. Therefore, the dissolution products were collected from the product and condensate tanks prior to filtration. Typically, a 15-gram sample was collected as the dissolution products flowed from the product tank. This sample was then filtered and extracted following the same procedure used in processing the microreactor samples.
5.3.2 Bench-Scale Test Results
Several initial shakedown runs were performed in the CSTRs using the Ohio (Bramhi) coal, Koppers coal tar distillate, and BS-19A at the same weight percentages used in the microreactor testing. Performed at a reaction temperature of 410°C and a reaction time of 30 minutes, these tests produced a tarry product that had a consistency of a petroleum resid at room temperature. Attempts to reheat this material for filtering were not successful. We have learned that getting the material to flow in a uniform manner after allowing the reaction products to cool down does not work well, even when heating the material to 125-150°C. Three options for lowering the viscosity to permit filtering were discussed. They included: 1) increasing the solvent-to-coal ratio to 2.5:1, 2) mixing THF with the products in a 50:50 blend, 3) ship the product to Battelle for filtering. For time’s sake, a decision was made to send the product from the first set of runs to Battelle for processing.
The next series of runs in the large reactor system were conducted using different bio-oil solvents. These tests are listed in Table 12. Not until test PSU-LR5 was a sample of the dissolution products processed and the coal conversion determined at PSU. Battelle also analyzed a portion of the dissolution product and we determined a significantly greater coal conversion (75-85 wt%). The reason for this difference is not apparent, but remains a point of interest.
5.3.3 Reaction Differences Between Microreactors and Bench-Scale Reactor
The purpose of microreactor testing was to evaluate the relative effects of different bio-solvents and operating conditions, such as reaction temperature and reaction time, and coal solubility. There are several operating differences between the microreactors and the larger CSTRs. First, the microreactors are agitated at approximately 200 cycles per minute, while internal impellers mix the CSTRs. It’s assumed that the impellers provide better mixing of the slurry (and improved mixing may be better with higher solvent to coal ratios), but the importance of this difference hasn’t been investigated. The second difference lies in the heating rate. It’s predicted that the contents of the microreactors reach the reaction temperature in ~3 minutes after the reactor is plunged into the preheated sand bath. However, the heavy-walled CSTRs take approximately 75 minutes to reach the reaction temperature (400°C). These differences may make it difficult to directly compare results between reactor types. However, each type of reactor should provide a relative ranking for experiments performed in them.
Assuming that the extended time interval required for the CSTRs to reach the reaction temperature was allowing free radicals to repolymerize, yielding heavier products, an additional series of tests (PSU-LR7, PSU-LR8, and PSU-LR9) were performed using shorter reaction times of 0, 15, and 30 minutes. As shown in Table 3, these experiments indicate that a reaction time of 0 or 15 minutes is insufficient to maximize coal conversion. In fact, reaction times greater than 30 minutes may be required when performing reactions in the CSTRs.
5.3.4 Reaction Differences in PSU CSTRs and Battelle Batch Autoclave
There were also differences between CSTR reactor conditions at PSU and batch autoclave at Battelle. As discussed in Section 2.2, PSU tried reactions at lower residence times to see if PSU was reacting the materials took with the longer heat up times. However, PSU found that these experiments indicate that a reaction time of 0 or 15 minutes is insufficient to maximize coal conversion. In fact, reaction times greater than 30 minutes may be required when performing reactions in the CSTRs. Another issue is PSU operated reactors at lower pressures than Battelle; PSU used controllers to keep pressures at ~500 psig (as suggested by Battelle), while Battelle allowed the internal pressure to reach the maximum pressure, ~600-1200 psig. PSU also found that Battelle’s batch autoclave has a faster heat up rate and an internal cooling loop, so heating and cooling could take place relatively quickly. PSU could not increase the heating rate, but PSU was able to remove the heater from the reactor at the end of the residence time and drop liquids out quickly in order to mimic Battelle’s cooling profile. As discussed in the previous section, the differences may make it difficult to directly compare results between reactor types. However, each type of reactor should provide a relative ranking for experiments performed in them. Future work may require increased reaction pressures.
5.4 Down-Selection of Feedstocks and Operating Conditions for Continuous Coal Liquefaction Testing
The objective of this subtask was to document the rationale used in down-selection of the pre-pilot scale operating conditions for conversion of coal to jet fuel. The Program Plan outlines two series of pre-pilot runs. The first involves the preparation of Syncrude using the Quantex 1ton/day (1 TPD) coal liquefaction facility located in Morgantown, WV. The second run, to convert the syncrude to distillation fuels, will be conducted at the Intertek 1 BPD hydrotreating system. A discussion of the rationale for bio-solvent(s) and conditions used for the Quantex pre-pilot unit is provided below.
Battelle conducted over 100 batch liquefaction tests, PSU conducted over 30 microreactor semi-continuous tests, and Quantex conducted three continuous runs to gather the data needed to down-select the preferred set of operating conditions. As noted earlier, we set a minimum of 80% solubility as our acceptance criteria. Subsequently, we added viscosity, bio-solvent availability, bio-solvent usage and raw-material cost as additional discriminators.
5.4.1 Initial Liquefaction Testing
Based on our work in the 1970s and 1980s, along with input from Quantex, we selected the initial range of operating conditions as follows:
In our testing we found that two types of bio-solvents were useful. One that could provide hydrogen to the coal matrix, such as a hydrogen-donor solvent, and one that could promote depolymerization. In order to maximize the donor-solvent capabilities, it was found that an optional pretreatment could, in some cases, be used to optimize performance. The process is shown schematically in Figure 17.
We conducted a series of tests, covering the range of liquefaction operating parameters. Some parametric testing was also conducted on high-priority bio-solvents. After review of the solubility data, we noticed that a test might produce a similar solubility, but the liquefied coal could appear dramatically different. Some products were a thick sludge and some were low-viscosity syrup-like liquids. Our experience indicated that lower viscosity product was superior, so viscosity determined at 50°C was added to our down-select criteria.
We initially tested a promising bio-solvent along with a depolymerization solvent. If successful, we progressively reduced the content of the depolymerization solvent and even dropped the bio-solvent-to-coal ratio to find the lowest, but still effective, bio-solvent-to-coal ratio.
The final criterion was bio-solvent availability and cost. In some cases we could purchase the bio-solvent from an established company, or we could easily convert a commercially available chemical. Cost was considered, but not in a rigorous manner. Some of the very best bio-solvents were also the most expensive. But, some relatively expensive bio-solvents could be used at relatively low proportions and still achieve good performance.
5.4.2 Initial Testing at Quantex.
We initially conducted three full-scale exploratory tests at the Quantex facility; one with Leer coal and two with Ohio coal. In these tests we fed -20 mesh Ohio (supplied by the Bramhi Coal Company), operated at approximately 400°C, 400 psig, and maintained a liquid to coal ratio of 1.9 to 2.0.
The coal was screw fed into a mixing pot that was co-fed with CTD (from a 250 gallon, room temperature tote), and bio-solvent (from a room temperature drum). The mixture was mixed in a special pump and then transferred to a preheater where most of the moisture was driven off. Next, the slurry was sent to a pump to bring the pressure up to the desired operating pressure. The coal was liquefied in two digesters. Excess pressure, caused by the vaporization of trapped water in the coal or the production of light hydrocarbons from thermal devolatilization, was purged to maintain the desired pressure. After liquefaction, the slurry was cooled to 100°C and then centrifuged to remove the bulk of the unreacted coal and residual mineral matter. The solids stream (the centrifuge “cake”) was sent to a drum. The centrate (liquid fraction from the centrifuge) was fed to a wiped-film evaporator (WFE) which served as a single-stage, vacuum-distillation column with a single cut point. Material boiled above or below this cut point. The below cut-point material exited the top of the WFE where it was cooled and stored in drums. The greater-than cut point material exited the bottom of the WFE and discharged directly to a drum.
Feed pump and feed line plugging was observed when using a 2:1 liquids to coal ratio. Coking in the two reactors (called digesters) was observed when operated at temperatures of 410°C. Plumbing modifications were instituted to reduce plugging. At Quantex’s request, it was decided to operate the multi-ton run using a 2.5:1 liquids to coal ratio to further reduce the chance of plugging. The temperature used to control the digester temperature was the bulk fluid temperature (measured via a thermocouple inserted in a well extending into the digesters). If a layer of material built up on the digester walls, reducing heat transfer, the wall temperature was automatically raised to achieve the desired bulk temperature. Thus, it was possible to have a 410°C or even 420°C wall temperature, and that led to coking in the earlier runs. To overcome this, the temperature set point was lowered in future runs, as discussed in Section 6. This may result in reduced time at temperature. If in the longer-time tests it is found that the centrate viscosity is too high, we may need to reduce the feed rate, or raise the liquid level in the digesters, in order to increase residence time.
In the planned longer-time Quantex tests, we hoped to simulate commercial operation as closely as possible. One major difference between what we do in the lab and what we expect to do in a commercial plant relates to the use of CTD. In a commercial plant we would use CTD only for startup, using WFE bottoms instead of CTD. Tests in the batch autoclave where the equivalent of WFE bottoms was used to displace CTD resulted (in some cases) in comparable, or even superior solubility rates. So our plan is to start with a high proportion of CTD, but in subsequent runs to displace it with recycled CTD (RCTD) prepared from a mixture of WFE overhead and bottoms.
5.4.3 Top Biomass-Derived Solvents.
The top ten biomass derived solvents, along with their solubility of Ohio coal, viscosity, and availability are noted in Table 13.
To accommodate the existing feeding system at the Quantex site, a 50:50 mixture of bio-solvent 27B and CTD was prepared at Battelle and then shipped to Quantex in four 55-gallon drums for Quantex Run #4. The contents of these drums, when heated to ~50°C, will have a viscosity of 3,650 cP (at 70°C, 359 cP, and at 90°C, 89 cP), so it can be fed in a manner similar to SBO that has been used in prior tests. Findings for BS-27B are noted in Table 14.
We chose BS-27B for the main test series at Quantex because of its excellent performance and excellent availability (it can be purchased from a commercial vendor). To support those tests, a series of parametric tests were conducted at two temperatures (390 and 400°C) and two residence times (10 and 30 minutes). Unfortunately, we had used up all of our high quality (Carbon Black Oil) CTD, and were forced to use an inferior “light” CTD. This resulted in an average drop in solubility of 15 to 17%. The parametric test results are shown in Table 15. As can be seen, at 390°C, even with longer reaction times, we obtained a difficult to filter product. Solubility was higher with the longer reaction time. In contrast at 400°C, even at 10 minutes reaction time, we produced an easy to filter product. At this higher temperature, solubilities were unchanged with longer reaction times.
The down-selected operating conditions for the Quantex pre-pilot scale run are as follows:
5.4.4 Planned Quantex Run.
A typical Quantex batch size is 300 lb of as-received (AR) coal, enough for about 6 hours of operation. Quantex does not have a sufficient number of trained staff to run 24 hours a day, so operation between 3 and 6 hours (1 to 2 batches) a day was planned.
If we assume the coal has 8.7 wt% ash (dry basis), and the centrifuge will capture 98% of the ash/mineral matter, the coal solubility will be 87%, and the centrifuge cake ash level will be 36.9%, we can make a material balance. The assumptions are shown in Table 16 and the results are shown graphically in Figure 18.
As noted in Figure 18 above, the process requires 605 lb/hr of RCTD. The relative split between the WFE overhead and bottoms is controlled by the selected set point temperature. The WFE is operated at a modestly-high vacuum 15-in. water (28 mm Hg) so that has been factored into the WFE set point operating parameter calculations. Based on batch single-stage distillation tests conducted under vacuum, we believe that set point such 320°F (160°C) at 15-in. water vacuum (28 mm Hg) is equivalent to 554°F (290°C) atmospheric pressure temperature and would meet these requirements.
In this configuration, all of the WFE overhead (92 lb) goes to the product stream. About 66% of the WFE bottoms (605 lb) are recycled as RCTD, leaving 223 lbs of bottoms to go to product. Total product per 300 lb coal lot is 315 lb or about 38 gallons.
6.0 PRE-PILOT-SCALE COAL LIQUEFACTION AND SYNCRUDE PRODUCTION
6.1 Introduction
Our team partner Quantex Energy was selected to produce large quantities of liquefied coal (Syncrude) in their pre-pilot plant scale coal liquefaction facility. To obtain information on scale up of the batch liquefaction tests conducted by Battelle and PSU, we elected to perform some tests to make enough material for syncrude upgrading in Subtask 1.04, described earlier in Section 5.4.
6.2 Description of the Quantex Pre-Pilot-Scale Coal Liquefaction Plant
A schematic of the Quantex liquefaction system is shown in Figure 19. A photograph of the system is shown in Figure 20.
The five components that were used in the testing to date are noted below:
6.2.1 Coal Preparation
Coal is obtained from the coal company in either 1-ton synthetic-fiber “Super Sacks” of in metal 55-gallon drums and transported to the Quantex site. In a special coal-grinding building, coal is conveyed up from the feed bin to a hammermill fitted with a metal screen. The mill produces coal with a size 90% smaller than 20 mesh (0.0331 inches, 0.841 mm); after it is ground it falls into 55-gallon drums. There is no drier including with the Quantex setup, but a small level of moisture removal is achieved during the handling and grinding process. Enough coal for the next day’s processing is typically ground and stored and be ready for liquefaction.
Representative samples of the Leer, West Virginal coal and Bramhi, Ohio coal were obtained and were analyzed for proximate and ultimate analysis along sulfur forms; results were presented earlier, in Section 5.1.1.
6.2.2 Preparation of the Battelle Bio-solvent and Coal Tar Distillate
The bio-solvents were prepared at Battelle using proprietary processes. Coal tar distillated (CTD) is a liquid obtained through pyrolysis of coal. It is distillation into different fractions before sale. Samples of three different CTD fractions were obtained from Koppers. Their relative effectiveness was evaluated in batch autoclave tests at Battelle and “carbon black oil” was determined to be the best.
6.2.3 Co-feeding of Coal, Bio-solvent, and Coal Tar Distillate
The pre-ground coal was screw-fed at a controlled rate of 100 lb/hr using a volumetric feeder. There it is mixed with CTD, from a 250-gal tote, and bio-solvent fed from a 55-gallon drum via a metering pump. There they were mixed using a special feed-mixing pump.
Depending on the set up of the feed-mixing pump, there can be limited or extensive grinding of the coal during its operation. Due to the grinding, the coal is preheated from ambient temperatures (80°F, 27°C) to over 150°F (66°C). The output of the mixing pump is sent to a slurry-preparation vessel. A pump at the bottom of the slurry-prep vessel sends the pre-heated coal/solvent/CTD mix to a valve that allows a portion to be sent back to the top of the slurry-prep vessel and a portion sent to the dewatering vessel. A similar mixing action is achieved as the coal temperature is raised about the boiling point of water and a portion of the water is removed. These gases are cooled and condensed and the liquids are sent to an oil/water separator. The gases are sent to a water scrubber for cleaning and removal of an organic mist.
After slurry prep and dewatering, the slurry is cooled and passed to a digester feed pump. Here the pressure is raised to the desired operating pressure.
6.2.4 Liquefaction of Coal
The output of the pressurization is sent to the first of two 10-gallon digesters. The digesters have level- control sensors so that when the volume exceeds 50%, a valve is automatically opened to allow a portion to be removed. A pump at the bottom of Digester 1 sends the fluid to a splitter. It allows a portion to be returned to the top of the first unit while the other portion is sent to Digester 2. The digesters do not have agitators, so this internal recirculation provides mixing. The output of Digester 2 output flows by pressure difference to a downstream heat exchanger/cooler. By adjustment of the level set points, different residence times can be achieved.
As the temperature of the coal and solvent mixture is increased any remaining water and a portion of the coal’s volatile matter are released. A pressure-control valve allows these gasses to be vented off while maintaining a constant pressure of 400 psig. These gases are sent to a condenser to knock out the organics and then to a holding tank. The gases are sent to the scrubber for cleaning.
6.2.5 Solids Removal by Centrifugation
The reacted coal is too hot to be dewatered directly. Therefore, the coal slurry is passed through two heat exchangers in series and then sent to a centrate feed and mix tank. From this agitated tank the liquefied coal can be held or sent to the centrifuge. In past runs, the material has been stored for several hours, and then centrifuged. The hope in this extended duration test is that the liquefied coal will spend a relatively short time in the feed tank before being de-ashed.
The solids from the centrifuge fall into a 55-gallon drum. The liquid stream, called the centrate, is sent to a centrate receiver. From there it can be held, pumped back to the centrifuge feed tank for reprocessing or sent on the wiped-film evaporator (WFE).
6.2.6 Thermal Separation into a Product Fraction and a Recycle CTD Fraction
The WFE is a tall, single column that allows the de-ashed, liquefied coal to be separated based on boiling point difference. The unit consists of a heated body and a rotor. The de-ashed, liquefied coal in pumped in above the heated zone and is evenly distributed over the unit's inner surface by the rotor. As the product spirals down the wall, the rotor blades generate highly turbulent flow to promote effective heating and rapid mass transfer. Volatile components are rapidly vaporized. Vapors flow co-currently with the de-volatilized liquid through the WFE and are condensed as our product and stored in a holding tank. The vapors off the tank are cooled to condense additional liquids – non-condensable gases are send to the scrubber for cleaning. The less volatile components are discharged at the WFE bottom directly into large tank and pumped, and pressure equalization, to a 55-gallon drum. A portion of these bottoms could be used as recycle coal tar distillate (RCTD). The unit is operated at vacuum to lower the fluid’s boiling point and allow separation at a lower temperature. Its performance is similar to a single-stage vacuum distillation column without reflux.
The unit is run by selecting a single cut-point temperature. Everything that boils at a lower temperature passes up and is drawn off in the overhead stream, condensed and collected. Everything with a higher than cut-point boiling temperature flows out the bottom. This cut-point temperature is frequently reported as the atmospheric-pressure equivalent temperature. The actual temperature corresponding to this reported temperature depends on the vacuum available.
6.3 Shake-down Testing Conducted at Quantex (Quantex Run #1)
Since the Quantex system had not been operated for a while, several changes were made to the plant before shake-down testing began.
6.3.1 Description of First Run with Leer Coal Company, WV Coal.
In January 2015, a 30-gallon trial was carried out using the following:
During startup, problems with the software used to control the plant delayed operation. The problems were resolved, but the coal continued to recirculate through the colloidal mill in the feed preparation loop for hours (which allowed the coal to be ground and re-ground to a smaller and smaller size and the slurry had to be heated to temperatures greater than the planned 105°C level). The smaller-than planned coal particle size may have led to the clogging and plugging problems that were observed at several points in the system. It was concluded that the output would not be representative of the liquefaction of Leer coal so a complete material balance was not performed. While problems were noted, the equipment was operated long enough to make a first examination of the products that can be obtained by Leer coal liquefaction.
6.3.2 Repeat Run with Leer, WV Coal.
Modifications were made to the Quantex facility, including the use of fume hoods for removal of gas and vapors from the WFE. In addition, the Zenith pump that fed the primary digester was moved upstream in order to avoid gravity feeding from the slurry prep tank to the primary digester. The re-located pump allowed recirculation of liquids until there was demand at the reactor, in which case the 400 psig pump should be sufficient to deliver liquids through the line to the reactor.
As noted above, the colloid mill was used as a way to heat the slurry prior to its entry to the reactor. Operation in this manner allowed the coal to be ground to small particle sizes. This may be beneficiation from a reaction point of view but could also produce a more viscous slurry making it more difficult to pump. As the colloid mill is the primary way to add heat the inlet slurry it was retained – but the internal settings such that the teeth of the mill were adjusted to the largest possible gap, thus reducing the ability of the grinder to create an emulsion, while continuing to rely on the creation of shear as a means to heat the working fluid to desired temperature.
The Quantex team made several other improvements to the liquefaction system including installing cowls to permit takeoff of high temperature liquids from the WFE while containing the vapors. A protective epoxy floor coating was applied around the system to protect the cement floor.
A control run was carried out on 19 February 2015. It was operated without coal in order to ensure that the new pumped section was viable and leak free. The reactor was successfully loaded and unloaded at about 400°C. The system passed this test and was considered ready for trials with coal. However, as this was a major change, the planned Bramhi Ohio coal run was postponed, in order to evaluate the changes using the previously tested Leer coal.
The Leer coal test was initiated. Coal did not flow smoothly. The coarsely ground coal exiting the colloid mill plugged the line. In addition, other technical difficulties were encountered. As the slurry entered the digester, the fluid level sensor started to gyrate, bouncing from an indication of completely full to completely empty. This inability to accurately measure the fluid level may have been due to foaming inside the reactor, creating false liquid level signals.
The run was stopped. The sensor issue was later modified with the assistance of the manufacturer. In addition, because the last run had issues with a difficult-to-flow emulsion being formed, the Quantex team elected to return to the original protocol, which specified 75°C pre-heating of the slurry rather than 105°C. It was recognized, as a result, the heating would be insufficient to boil away moisture prior to the reactor. Moisture could instead be liberated and captured with overhead vent stream from the reactor loop. This was thought to be adequate for coal with moisture content of a few percent.
6.4 Run with Ohio Coal (Quantex Run # 2)
This test was conducted using soybean oil, in an attempt to establish a baseline to compare the Battelle process with. The Ohio (Bramhi) coal flowed more smoothly through the equipment, but it seemed to be more reactive upon entering the reactor, as the liquid level again showed wild gyrations. The varying liquid level can be due to the presence of boiling water. The system was stopped to make further adjustments to the liquid level control system. It was decided to run with WV coal to make sure the control problem was corrected, and then switch to Ohio coal. A sample of the WFE-overhead syncrude from this run was utilized for initial upgrading work at UDRI (see Section 7.4).
6.4.1 Work-up of Syncrude from Quantex Run # 2.
Quantex delivered a 5-gallon pail of the liquefied Ohio coal to Battelle. Inspection showed that the material in the top of the pail was free flowing and had a viscosity similar to vegetable oil while the material in the bottom of the pail was more viscous. This indicates, as also observed in lab-scale testing, that soybean oil does not react with coal, as it is incapable of hydrogen transfer.
The material in the 5-gallon pail was first mixed well and then 6 liters (L) were removed. Approximately 1 L was loaded into a pressure filter and filtered at room temperature. Over a series of tests, a total of 5,545 g was filtered. About 73 wt% of the material passed through the 20 micron filter paper. The once filtered material was re-filtered and 99.6% of this material passed through a 0.22 micron filter.
The filtered material was uniform with no separation in the top or bottom phases. Some physical properties were measured, see Table 17. They showed similar results regardless of whether the test sample was drawn from the top or bottom of the bottle.
6.5 Run with WV and Ohio Coal (Quantex Run # 3)
This test was conducted using Battelle-prepared BS-19A bio-solvent. To minimize digester level fluctuations, the plant setup was returned to a target slurry preheat temperature of 105oC in order to boil off moisture before it could enter the reactor loop. This was successful. The plant was operated one day with Leer coal. The following morning the tanks were drained and an Ohio coal test initiated. This run, designated Quantex Run #3, was successful as the Ohio coal was liquefied and de-ashed. The processed coal was then sent to the WFE. Some problems at the discharge end of the WFE were encountered preventing the production of a representative pitch product. However, the WFE overhead stream was recovered.
Process improvements allowed for processing most of the desired 203 kg run (69.4 kg coal; 96.0 kg CTD; 37.8 kg BS-19A). However, due to a digester wall being at 430°C the digester coked and plugged before processing the last 47.5 kg of mixture. The digester wall temperature will be held lower in future runs in order to avoid coking. The rest of the material was run through a centrifuge and then through the WFE, operating at 290°C atmospheric equivalent. The material through the WFE was found to be 10 wt% <290°C boiling and 90 wt%>290°C boiling. The Flow rates were estimated and materials sampled for analysis.
Discussions with Quantex indicated that there should be little if any insoluble material since the coal had been centrifuged and distilled (in the WFE). They suggested that we hot filter the material.
A sample of the liquefied Ohio coal from the 5-gallon pail, after good mixing, was withdrawn and hot filtered at 80 to 90°C in the same pressure filter. Under these conditions, the following was found:
6.5.1 Work-up of Syncrude from Quantex Run #3.
A sample of the centrate before the WFE was taken and sent to UDRI for hydrotreatment. This centrate had the approximate distillation cuts, shown in Table 19, when distilled in the lab at Battelle.
6.6 Quantex Run #4 with Ohio Coal
Quantex Run #4 was run with bio-solvent BS-27B. Because of processing issues, only 150 lbs coal was processed during this run. The process feed was:
The digester operating conditions for Quantex Run #4 were as follows:
The material exiting the digester was fed to the centrifuge feed tank and maintained at a temperature of about 140°C. Next, the material was run through the centrifuge (at about 110°C). Roughly 123 lbs solids from the centrifuge formed a filter cake and fell into a 55-gallon drum (approximately 50% solids). The liquid stream, called the centrate, was sent the wiped-film evaporator (WFE) feed tank, then pumped to the WFE. The WFE is a tall, single column that allows the de-ashed, liquefied coal to be separated based on boiling point differences. The WFE consists of a heated body and a rotor. The de-ashed, liquefied coal was pumped in above the heated zone and was evenly distributed over the unit's inner surface by the rotor. As the product ran down the wall, the rotor blades generated turbulent flow to promote effective heating and rapid mass transfer. Volatile components in the feed were rapidly vaporized. Vapors flowed co-currently with the de-volatilized liquid through the WFE and were condensed as our product and stored in a holding tank. Vapors off the tank were cooled to condense additional liquids, while non-condensable gases were sent to the scrubber for cleaning. The less volatile components were discharged at the WFE bottom directly into a 55-gallon drum. The unit was operated at a vacuum pressure of about 21 mm Hg and a temperature of 230°C giving an atmospheric equivalent distillation temperature of 380°C. The distillate was labeled “lights” and the WFE bottom was labeled “heavies”.
In Quantex Run #4, approximately 70% of the moisture- and ash-free (MAF) coal fed was solubilized producing about 184 lbs of lights, and 201 lbs of heavies, while the unsolubilized coal and some liquids formed 123 lbs of centrifuge filter cake. These materials were produced from 150 lbs of Ohio Bramhi coal, 303 lbs of Koppers Carbon Black Oil CTD, and 36 lbs of bio-solvent BS 27B. Materials produced during Run #4 were sent to Battelle for further analysis and verification.
A mass balance was conducted on Quantex Run #4 as shown in Figure 21 and Table 20.
Total out/in provides a 97% mass balance for Run #4 at Quantex. Contributions to the 4% mass lost likely came from a small amount of coking on the digesters, residual material left in the system, leaks, and water/light hydrocarbons vented during the evaporation process. Overall, the 97% mass balance provides excellent closure for the entire Run #4.
6.7 Quantex Run #5
Quantex Run #5 was run with bio-solvent BS-32. The operating conditions for Quantex Run #5 were as follows:
Total out/in provides a 95% mass balance for Run #5 at Quantex. Contributions to the 5% mass lost likely came from a small amount of coking on the digesters, residual material left in the system, leaks, and water/light hydrocarbons vented during the evaporation process. Overall, the 95% mass balance provides excellent closure for the entire Run #5.
6.7.1 Discussion on Results from Quantex Runs #4 and #5.
Both Quantex Runs #4 and #5 successfully produced enough material for hydrotreatment/hydrogenation to jet fuel. The Battelle bio-solvents performed well, but minor adjustments are necessary to improve performance. Further optimization is needed for tests using BS-27, due to the coking and plugging issues observed in Run #4. We found that BS-27 would benefit from the addition of a second bio-solvent in order to maintain a slurry that doesn’t swell and plug lines. This is supported by the work in Quantex Run #3 with bio-solvent BS-19A, which has similar structure to BS-27. In addition, maintaining lower slurry temperatures also inhibits early coal swelling. Run #5 with BS-32 ran fairly smoothly and also benefitted from the addition of a second bio-solvent to prevent coal swelling in the slurry stage. When recycling WFE heavies to replace Koppers CTD, it was found that adjustments in the level of BS-32 are needed. Very heavy oil was produced when the WFE heavies were recycled due to hydrogen-starving the system, resulting in only partially liquefied coal. Initially, we used heavies to replace 50% of the CTD without adjusting the bio-solvent, which led to plugging of the centrifuge lines. In order to reduce plugging, the level of bio-solvent or recycled WFE heavies must be adjusted. However, sufficient syncrude quantities have been produced from both runs to allow testing in Task 3 (pre-pilot hydrotreatment/hydrogenation). The syncrude for Quantex Run #5 was actually used in Intertek Run #1 test.
6.8 Run with Ohio Coal (Quantex Run #6A)
This test was conducted using Battelle-prepared BS-40D bio-solvent and the previously-tested Ohio coal from Bramhi Coal Company. To minimize potential issues in the slurry step, the slurry preheater temperature was held below 150ºC. On the first day, we consistently had plugging of Digester 1 recirculation pump. This was due to the slurry not reaching a sufficient temperature and pretreatment time before reaching the gear pump. A decision was made to reverse the recirculation in Digester 1 so that the mixture would gain more pretreatment time at temperature before reaching the gear pump. These changes were successful and were implemented on day two. The slurry temperature on day two ended up reaching 71°C due to manual feeding of the system. The plant was then operated for one day with the Ohio/Bramhi coal. The product was then centrifuged and 15 gallons of centrate were transferred to a drum. The leftover centrate was then sent to the wiped film evaporator (WFE) to fractionate the centrate boiling below 380°C. The WFE ran for one pass to obtain a light (<380°C boiling point) and heavy (>380°C) product.
The improvements allowed for processing of a 234 kg run (66.18 kg coal; 119.36 kg CTD; 31.91 kg BS-19A; 16.55 kg SBO). Fluctuations were noted in the digester temperatures but both indicated a range of 400-415°C. The centrate through the WFE was found to be about 40 wt% light and 60 wt% heavy product. The flowrates were estimated and materials sampled for analysis. The mass balance for this run was about 97.0 wt%; about 18.68 kg were last in piping tanks.
6.9 Run with Ohio Coal (Quantex Run #6B)
This test was conducted using Battelle-prepared BS-40D bio-solvent and recycle solvent from Quantex Run #6A to better simulate commercial-scale operation under steady-state recycle of some solvent. To minimize potential issues in the slurry pretreatment step, the slurry preheater temperature was held below 150ºC. We used the same flow scheme as Run #6A. The plant was operated for a half-day with Ohio/Bramhi coal. The product was then centrifuged and all centrate was transferred to drums for testing at Intertek (Task 2.02.01). This approximately half-day run allowed for processing of 217.1 kg of feed (54.27 kg coal; 68.39 kg CTD; 27.14 kg BS-19A; 67.43 kg Recycle). Fluctuations were noted in the digester temperatures but both indicated a range of 400°C-415°C. This process produced 132.27 kg of centrate, which was placed into drums. The Flow rates were estimated and materials sampled for analysis. The mass balance for this run was about 97.6%. The syncrude from Quantex Run #6B was used for upgrading testing in Intertek Run #2.
6.9.1 Product Analyses
The liquid samples from runs 6A and 6B were submitted for ultimate analysis and can be seen in Table 22. The samples were also submitted for viscosity analysis.
We experienced difficulty in obtaining a representative sample of the centrifuge tails from both runs. This difficulty is due to various layering that takes place when starting and finishing the run through the centrifuge. It was decided that the best option for sampling was to drive a 1 inch pipe through the deepest part of the mixture. The liquid/solid mixture was then worked up by solvent rinsing and drying. The solid was submitted for ultimate analysis. The fraction of solids in the centrifuge tails for Runs #6A and #6B were 38.9 wt% and 42.1 wt%, respectively.
7.0 LAB-SCALE CATALYST TESTING FOR SYNCRUDE HYDROTREATING/HYDROGENATION
7.1 Introduction and Objective
The conversion of liquefied coal (coal converted from a solid to a liquid using bio-derived solvents) into a form suitable as a blending stock for jet turbine fuel requires three broad processes. First, the heteroatom species must be removed; primarily components containing nitrogen, oxygen, and sulfur. Second, the largely polynuclear aromatic structure of the coal derived components must be reduced (hydrogenated/hydrocracked) to predominantly paraffinic components, though some single and two-ring aromatics are permitted within the fuel specification. Third, the boiling range of the final product should be within the fuel specification. The first two processes, heteroatom removal and reduction, are typically accomplished via heterogeneous catalysis. The third may also be accomplished through heterogeneous catalysis, though fractionation may also be used.
The primary objectives of this subtask were to identify commercially available candidate catalysts for heteroatom removal and hydrogenation and the nominal processing conditions.
7.2 Catalytic Upgrading Background
For the upgrading of syncrude from the Battelle CTL process, a two-stage process was conceptualized. The two stages, shown graphically in Figure 23, are designated as Stage-1 and Stage-2, where Stage-1 would perform mainly heteroatom removal and Stage-2 would perform hydrogenation and hydrocracking reactions. A review of the literature shows that the most common commercially available catalysts for heteroatom removal are sulfided forms of CoMo and NiMo [5-11]. Catalysts for hydrogenation and hydrocracking (reduction) are bi-functional catalysts such as PtPd, sulfided NiMo deposed on acidic support such as zeolite Y and alumina [5-8]. Generally, PtPd is preferred, though this catalyst is more expensive than sulfided NiMo and is very sensitive to residual heteroatoms in the feed, principally N and S. NiMo is less expensive and can be sulfided, providing good performance in the presence of heteroatom species.
Literature Review of Catalysts for CTL Syncrude Upgrading. Battelle team member PSU performed a literature review of catalysts for upgrading of coal-derived syncrudes. The review and the associated references are provided in Appendix A.
7.3 Jet Fuel and Diesel Specifications
The primary focus of this project was to produce jet fuel from coal. However, one can also produce diesel if desired. The specification for various distillate products of interest are discussed below.
The specification for conventionally produced (i.e., petroleum-based) jet fuel is ASTM D1655 – 11a. The requirements are presented in Table 23 [12]. Analysis indicates there is no stated ash or nitrogen limit in D1655. A discussion of each is provided below.
There is also a specification for synthetically produced jet fuel, D7566-11a. The D7566 has different Annexes for different synthetic fuel; see Figure 23. However, there is no Annex for jet fuel from coal via direct liquefaction.
Annex A1 is for Fischer-Tropsch (FT) synthetic paraffinic kerosene (SPK) fuel, Annex A2 is for Hydrotreated Esters of Fatty Acids (HEFA) SPK, and Annex A-3 is for the recently approved Direct Sugar to Hydrocarbons process to make Synthesized Iso-Paraffins (SIP). Annex A1 and A2 fuels must be blended with at least 50% petroleum jet fuel; Annex A3 fuel must be blended with at least 90% petroleum jet fuel.
It would be reasonable to expect the direct-coal-liquefaction jet fuel would require its own annex; it would definitely not fall under Annex A1 for FT fuels (even though A1 covers coal gasification to make syngas which is converted by FT technology into jet fuel). The annexes contain, in some cases, limits different than those specified in D1655. For example, for SPK and SIP, which contain only aliphatic hydrocarbons, the upper limits on the amount of aromatics is 0.5%; this is in contrast to D1655 where the limit is 25 volume % aromatics.
There are no stated ash limits, but there are limits on nitrogen, sulfur, and aromatics in D7566-11a. Details are provided below.
The overall approach was to identify and obtain small quantities (100 to 1000 mL) of commercially available catalysts and to screen their performance using model or actual syncrudes using a set of small trickle-bed reactors operating under nominal processing conditions. From this screening process one or more promising candidates were to be selected for more detailed evaluation.
The first task was to adjust and calibrate the reactor systems for these type of reaction “Syncrude Upgrade to jet fuel”. UDRI has several reactor systems that were used for this project. All of these reactor systems are very similar in their overall construction and flow path. A general schematic is shown in Figure 24, and a photograph of reactor #2 is shown in Figure 25. (Note that the reactors may be fitted with different inlet gases and product tanks depending on the specific configuration.) Each system includes separate gas and liquid feeds with heated transfer lines, a heated catalyst reactor with a moveable axial thermocouple, heated product collection tanks, and a gaseous vent that can be monitored with various online and offline analyzers.
Each reactor is a ½” diameter by 24” long 316 stainless steel tube vertically oriented inside a 3-zone furnace (Applied Testing Systems Inc.). The temperature of the 6” central heating zone can be controlled independently through thermocouples that are spot welded to the exterior of the reactor tube at the center of each heated zone and the temperature of the center zone can be monitored by a moveable thermocouple located within a well extending through the center of the reactor tube. The length of the tube within the furnace is covered with a split cylindrical brass tube of 1/8” thickness to assist in evenly distributing the heating loads. Reactor pressure is controlled by a back- pressure regulator downstream of the product receiver. Gases are supplied from regulated cylinders which were also used to pressurize the system. Gas flow rates were precisely controlled with 5850i Brooks mass flow controllers. Liquid feeds can be heated to avoid crystallization in a one liter feed tank and charged into a heated ISCO-500D syringe pump. The liquid and gas feeds combine at the top of the fixed-bed column and mix while flowing through approximately 4” of 54 mesh silicon carbide before contacting the catalyst-containing portion of the bed. Catalysts can be used as the as-received extrudates, crushed and sieved extrudates, or, more commonly, the crushed extrudates diluted with inert, high-conductivity material, such as silicon carbide. (Diluting the catalyst provides superior temperature control in the presence of strongly exo- or endothermic reactions and reduces channeling through the bed.) The lower section of the reactor is similar to the inlet section with a volume of 54 mesh silicon carbide resting on a plug of quartz wool on a 20 μm sintered stainless steel filter at the exit of the reactor tube.
If sulfiding and/or activation procedures are provided by the manufacturer, these procedures are used. In the manufacturer’s SOPs are not provided then procedures developed in-house are used. For example, base metal catalysts are activated after the reactor has been prepared and mounted in the reactor furnace. This is usually conducted at 450°C with hydrogen flowing at an equivalent gas hourly space velocity (GHSV) of 13.7/hr at 100 psi for 4 hours. Catalysts that require sulfiding are treated after they are loaded into the reactor tube, but before the tube is installed in the reactor furnace. Briefly, the reactor is mounted in a horizontal furnace, supplied with a flowing mixture of hydrogen and hydrogen sulfide at atmospheric pressure, and heated gradually over 2 hours to 350°C, held for 6 hours, cooled to room temperature, and then capped for transfer to the reactor furnace.
For each test run the reactors are typically operated with a LHSV of 1/hr based on the dry mass of the catalyst. Hydrogen is supplied at a molar ratio of 10:1 based on the known or estimated mean molecular weight of the liquid feed. To measure H2 consumption N2 (%volume ~2%) can used as internal reference gas. The sulfided reactors and the base metal reactors are typically operated with 800 psi of hydrogen. The sulfided reactors are conditioned at the initial temperature for the test sequence with flowing liquid feed for 1.5 hours. The product collected during this period is then drained from the product tank, and a test volume collected for 1.5 hours. This sample is then drained from the tank and stored for analysis. The temperature is then set for the next test point and the reactor allowed to run until the reactor temperature is steady, typically about 1.5 hours. The product tank is then drained, and the process repeated until the test sequence is completed. The reactor temperature is normally increased from the starting temperature to the final temperature in predetermined steps. However, in some cases the temperature may be decreased from some upper starting temperature to a lower ending temperature. To measure the hydrogen consumption the reactor effluent rates are recorded and averaged, then the effluent gas is analyzed using the online 3000 Micro GC (Inficon).
Reviewing the available reactors, Reactors #2 and #6 were selected for the heteroatom removal since these systems have been used previously with sulfided catalysts. Reactor #1 was selected for the upgrading of the product from the heteroatom removal as this system has been previously used as a hydrotreating reactor and has not been operated with sulfided catalysts.
To quantify the performance of the Stage-1 reactor, the conversion of selected heteroatom components was measured using gas chromatography–mass spectrometry (GC-MS). Furthermore, selected samples were submitted for quantitative elemental analysis. Briefly, the analysis was used to measure the relative concentration of carbazole (N), dibenzothiophene (O), and dibenzofuran (S) between the product and feed. This analysis was relatively quick and could be performed in-house and served as a general-purpose screen tool. The quantitative analysis of selected elements (C, H, N, O, and S) was conducted by an analytical services laboratory (Galbraith Laboratories, Inc., Knoxville, TN) which gave more comprehensive results for the heteroatom conversion, but was more costly in terms of time and resources.
The performance of the Stage-2 reactor was quantified using GCxGC to conduct a hydrocarbon type analysis of the reactor feed and product. This analysis quantifies the major hydrocarbon classes in the sample by type such as the number of aromatic rings, cyclic or linear and branched alkyl aromatics, and cyclic, linear, or branched paraffins. Of particular interest for this program is the conversion of polynuclear aromatic hydrocarbons to single-ring aromatic hydrocarbons and paraffins (linear, branched, and cyclic). From this analysis, the degree of HDA could be determined as well as measuring the concentration of various aromatic components in the product. The latter is of interest as jet fuel specifications allow up to 25% single-ringed aromatics and 3% di-aromatics (alkyl naphthalenes).
7.5 Materials.
The feedstocks used in this program consisted of model fluids composed in-house at UDRI and actual syncrudes provided by Battelle. The model fluids consisted of Aromatic-200 (Exxon Mobile), a complex petroleum distillate of C12–C15 aromatics in the form of alkylnaphthalenes. This was used either as-received or blended with selected heteroatom components. Based on a review of the literature on petroleum refining and the production of ultra-low sulfur diesel fuel, dibenzothiophene (DBT) and carbazole (see Figure 26) were selected as model sulfur and nitrogen containing species, respectively. These were blended with the Aromatic-200 to give a total sulfur content of 1% mole/mole and a nitrogen content of 0.01% mole/mole. After blending, the Aromatic-200 was designated Aromatic-200SN. Note that the molecular structures of these model compounds are very similar to dibenzofuran, which was later used as a marker species for the presence of oxygenates in the actual syncrudes, but not used in the model feedstocks.
The actual syncrudes were initially made available in relatively small volumes and were the products of an evolving process. In total, 6 syncrudes were delivered in two broad versions. The first two syncrudes (SC-1 and 2) were from an early development production batch (Quantex Run #2). These were very viscous and had a significant solid or gel phase that required sonication and/or mild heating (80-90°C) to homogenize. The syncrudes WFE SC-1 through 3 are from the same syncrude batch but filtered on different days as needed; these were from a more refined production process (Quantex Run #3; feed to the wiped film evaporator – WFE). WFE SC-4 is distillation fraction of WFE (Quantex Run #3) below 450°C. These were less viscous than the first set, though they also included a second phase that required sonication and/or mild heating (50-80°C) to homogenize.
GC-MS analysis of the syncrudes showed that their overall composition was very similar. All of the syncrudes were very complex mixtures of polynuclear aromatic hydrocarbons (PAHs) spanning from about naphthalene (2 rings) to benzopyrene (5 rings) with an average density of 1.102 g/mL. Figure 27 includes the GC-MS analysis of the Aromatic-200SN showing that this model feed consists primarily of alkyl naphthalenes plus the added dibenzothiophene and carbazole (average density 0.982 g/mL). The elemental analysis and density of the syncrudes are summarized in Table 24. It shows that there was some variability in the heteroatom content of the various syncrudes, though the density was fairly consistent. Taking 10ppm as a nominal target for the final concentration of N, O, and S, suggested that an average heteroatom removal efficiency on the order of 99.9% was required.
The catalysts used in this subtask were commercial and developmental catalysts. We are required by non-disclosure agreement (NDA) to not disclose the identity of these catalysts. A list of the catalysts tested is provided in Table 25.
7.6 Results and Discussion.
The Stage-1 catalyst evaluation was conducted in four broad phases largely driven by the availability of syncrude and catalysts. The first phase was conducted using the model Aromatic-200SN feed with catalysts available in-house. This verified the operation of the reactors, calibration of the system and established the initial operating conditions for subsequent analyses. The second phase used the first samples of syncrude. These were processed through Stage-1 and provided information on the operating conditions and challenges associated with these feeds and analysis. The third phase focused on screening the remaining Stage-1 catalysts using a standardized set of conditions based on the initial work with the first syncrude material. In the fourth phase the most promising candidate was selected for more detailed evaluation and to produce a finished Stage-1 product for upgrading in Stage-2.
7.6.1 Stage-1, Series 1 - Aromatic-200SN Model Feed.
For this initial test sequence the reactor was prepared and catalyst sulfided as described above using the as-received extrudates that were crushed, sieved 32/60 mesh, weighed (approximately 2.6 g, 3.5 mL), diluted 1:1 with 54 mesh SiC, loaded into the reactor tube, sulfided, and then transferred to the reactor assembly. The catalyst was initially conditioned using the neat Aromatic-200, and then with the Aromatic 200SN. Tests were conducted from 300°C to 400°C at 5°C intervals using a LHSV of 1/hr (nominally 5.2-5.5 L/min) and 800 psi hydrogen flowing at a molar feed ratio of 10:1 (nominally 46-49 mL/min at STP) assuming a mean molecular weight for the feed of 252.54 g/mol. At each temperature, a sample of the product liquid was analyzed by GC-MS. The conversion of DBT and carbazole was measured using the peak areas from the extracted ion chromatograms taking the 184 m/e and 167 m/e mass fragments as being characteristic of DBT and carbazole, respectively. The sulfur and nitrogen concentration was estimated from the measured conversion and the known starting concentrations of DBT and carbazole and assuming that no other organic sulfur or nitrogen compounds were produced as products.
The initial tests were performed using catalyst A (CoMo) and the Aromatic-200SN feed. The overall conversion and heteroatom results are summarized in Figure 28. These results show that the concentration of DBT steadily declined as the temperature increased from 300°C to 400°C, while the carbazole passed through a minimum at approximately 350°C. Conversion of the Aromatic-200 to naphthalenes and tetrahydronaphthalenes was also observed. Small amounts of alkyl benzenes and decalin are also present in the product.
As the tests with Catalyst-1 were concluding the commercial catalysts became available for testing. Catalyst “C” was selected for the first series of tests using the Aromatic-200SN feed as there was interest in comparing this NiMo catalyst with the CoMo catalyst used previously. Catalyst “D” was also selected for evaluation at the manufacturer’s literature indicated that this catalyst may show an activity lower than Catalyst “C”.
The results for the catalysts ‘C” and “D” are summarized in Figures 29 and 30, respectively. These show that the performance of these two catalysts is indeed very similar to each other. Both showed excellent activity towards the conversion of carbazole, reducing it to below the detection limit of the GC-MS by approximately 320°C. They also showed good activity towards the conversion of DBT up to about 350°C, after which the activity slightly declined. Both of these catalysts showed a much higher activity towards the reduction of carbazole as compared to catalyst “A”, while they showed a similar level of activity towards the conversion of DBT throughout most of the temperature range used here.
The overall results from all three of the catalysts described above are summarized in Figure 31 This illustrates that the performance of the sulfided CoMo and sulfided NiMo catalysts was very similar in their ability to reduce the concentration of dibenzothiophene present in the model feed and an optimal temperature range of approximately 340°C to 370°C. The NiMo catalysts showed higher performance in the removal of N with the concentration dropping below the detection limit of approximately 0.1 ppm by 320°C. In contrast, the concentration of N from the CoMo catalyst remained above the detection limit across the temperature range used here, though the performance passed through a maximum at approximately 350°C with less than 1 ppm N remaining in the liquid product.
7.6.2 Stage-1, Series-2 - Syncrudes SC-1 & 2 and WFE SC
Series-2 included the initial tests using the early syncrudes SC-1 and -2 and WFE SC-1. These syncrudes were produced using the baseline solvent, i.e., soybean oil. The SC-1 and -2 were small volumes of a very similar syncrude that had a coarse solid phase at room temperature that required sonication and/or mild heating (80-90°C) to homogenize. In the initial Phase the catalysts were crushed, sieved, and diluted 1:1 with SiC before being sulfided. The goal of dilution of catalyst with SiC is to better control exothermic reaction and prevent hotspots. The runs included an initial conditioning period of 150 hours under a nominal startup condition; typically 360°C with 700 psi of hydrogen and a LHSV of 0.3/hr. After this initial conditioning period the reactor performance was evaluated at various temperatures and pressures at a fixed LHSV. The primary measure of reactor performance was the analysis of the N, O, S marker components (carbazole, dibenzofuran, and dibenzothiophene) as described above as well as the product density and appearance. Selected samples were submitted for elemental analysis (N, O, S). Series-2 ended with the migration to a simpler standardized test sequence using the as-received catalysts without dilution, a LHSV of 0.15/hr, a fixed temperature (380°C) and pressure (950psi), and a fixed run time of 100 hours.
The product from the Series-2, Stage-1 reactors was typically very dark and often contained what appeared to be very fine particulate that slowly settled. The product could be clear when the catalyst was relatively new or when the temperature and pressure were increased. However, the quality of the product as indicated by the GC-MS or elemental analysis did not seem to correlate with the visual appearance of the product.
The GC-MS analysis of the Phase-2, Stage-1 reactor products for WFE SC-1 feed shown in Figure 32 illustrate that there were essentially no differences in the principal organic constituents in the reactor products, though the relative concentrations of the individual components did show some variability. These results also show that the reactor feed was almost universally partially hydrogenated. Specifically, while the feed was composed almost entirely of PAHs, the reactor products show varying degrees of ring saturation. However, there is little evidence for ring opening. It is also interesting to note the presence of dibenzofuran in all of the reactor products, illustrating the limited conversion of this specific heteroatom component. Similarly, pyrene was noted as one of the few PAHs present in both the feed and product, suggesting that this component is exceptionally resistant to hydrogenation.
The overall results summarized in Table 26 suggest that the most promising candidates from the Phase-2 evaluation were the “I” and “F” catalysts based largely on the GC-MS analysis of the marker components for nitrogen (carbazole) and sulfur (dibenzothiophene) as well as the elemental analysis for sulfur. Only modest reduction in the marker component for oxygen (dibenzofuran) was noted from any of the catalysts. This is consistent with literature that suggests that dibenzofuran is exceptionally stable. The decrease in the density of the product relative to the feed indicates a modest degree of hydrogenation consistent with the GC-MS analysis described above. Note that in the absence of significant ring-opening the lightest product that can be produced from these syncrudes is decalin with a density of 0.895 g/mL.
7.6.3 Stage-1, Series-3 - Syncrudes WFE SC-2, 3, and 4
Series 2 showed that relatively aggressive conditions of temperature and pressure would be needed to successfully remove the heteroatom components from the coal-derived syncrudes. It was also desirable to standardize the exposure conditions to make the results more comparable. For these reasons, the exposure conditions were fixed at 380°C and 1250 psi. Furthermore, the Series 3 catalysts were loaded as-received (extrudates) and without dilution. This doubled the amount of catalyst in the reactor, and by leaving the flow rates of feed and hydrogen unchanged the LHSV dropped from 0.3/hr to 0.15/hr without decreasing the production rate of Stage-1 product. It was observed that the hydrotreatment of syncrude is not excessively exothermic and therefore the catalyst can be used catalyst without SiC dilution
Photographs of selected Series-3 products are shown in Figures 33 and 34. The GC-MS total ion chromatograms of selected products are shown in Figure 35. The overall results from the Phase-3 evaluation are summarized in Table 27.
The photographs shown in Figures 33 and 34 illustrate that the product from the Series-3, Stage-1 reactors were relatively clear, though some contained what appeared to be very fine particulate that slowly settled. The fine material could also agglomerate with the water phase, making it somewhat difficult to separate.
The GC-MS analysis of the Series-3, Stage-1 reactor products shown in Figure 38 illustrate that there were essentially no differences in the principal organic constituents in the reactor products after 100 hours of operation with the syncrude, though the relative concentrations of the individual components did show some variability. These results also show that the reactor feed was almost universally partially hydrogenated. Specifically, while the feed was composed almost entirely of PAHs, the reactor products show varying degrees of ring saturation. However, there is little evidence for ring opening after the first few hours of operation with the syncrudes. It is also interesting to note the presence of dibenzofuran in all of the reactor products, illustrating the limited conversion of this specific heteroatom component. Similarly, pyrene was noted as one of the few PAHs present in both the feed and product, suggesting that this component is exceptionally resistant to hydrogenation.
The overall results summarized in Table 27 suggest that the most promising candidates from the Series-3 evaluation were the “E” and “H” catalysts. These showed the best reduction in heteroatom content based on the elemental analysis. The GC-MS analysis of the marker components also showed excellent reduction in nitrogen (carbazole) and sulfur (dibenzothiophene). Only modest reduction in the marker component for oxygen (dibenzofuran) was noted from any of the catalysts after 100 hours of operation with the syncrudes. This is consistent with literature that suggests that dibenzofuran is exceptionally stable. The decrease in the density of the product relative to the feed indicates a modest degree of hydrogenation consistent with the GC-MS analysis described above. Note that the greatest reduction in the product density was also found with the “E” and “H” catalysts. Note that in the absence of significant ring-opening the lightest product that can be produced from these syncrudes in decalin with a density of 0.895 g/mL.
Note that the products from Series-3 were much lighter in color and clean compared to those from Series-2. The key reason for this is that Series-2 syncrude was made using soybean oil, which did not turn out to be a hydrogen donor, so the resulting syncrude was very heavy. On the other hand, Series-3 syncrude was made using a bio-solvent from the Battelle. CTL process.
7.6.4 Stage-1, Series-4 - Production of Stage-1 Product.
Based on the evaluations described above, the “H” sulfided NiMo catalyst was selected to produce Stage1 product for upgrading in Stage-2. To maximize the heteroatom removal from the Stage-1 process the approach was modified by conducting it in two steps referred to as Stage-1a and Stage-1b. The Stage-1a reactor was operated as described above. However, the elemental analysis of the reactor product as shown in Table 27 suggests that significant amounts of N remain in the product and the level of S is not low enough to permit the use of a noble metal catalyst in Stage-2. Therefore, the product from the Stage-1a reactor was sparged with nitrogen and washed twice with water to remove residual NH3 and H2S. (Sparging was conducted with nitrogen for 1 hour. Washing was with 1:1 HPLS water with gentle swirling and decanting in a separatory funnel.) This washed product was then processed through a Stage-1b reactor to complete the heteroatom removal. The catalysts and processing conditions for both reactors were the same; “H” NiMo catalyst at 380°C, 1250 psi, and with a LHSV of 0.15/hr. The syncrude used as the feed to the Stage-1a reactor was WFE SC-4.
Photographs of selected Stage-1a products are shown in Figure 36. This illustrates that the Stage-1a product is quire dark. These contrast with the product from the earlier evaluation of this catalyst using the WFE SC-3 syncrude (see Figure 34) which were relatively clear. As was previously observed, apparently small changes in the feed can make significant changes in the appearance of the product. Photographs of the Stage-1a and Stage-1b products given in Figure 37 shows that the Stage-1b process gives a relatively clear product.
Recall that the primary purpose of the two-step Stage-1 process is to remove products such as NH3 and H2S that could reduce the activity of the catalyst downstream of the initial reaction zone while at the same time preserving the fuel value of the product stream. To monitor the effect of the nitrogen sparge and water wash on the removal of ionic species, the pH of the reactor products was measured as-produced from the reactor, after the nitrogen sparge, and after the water wash. These results, summarized in Table 28, show that the sparge and wash effectively reduced the pH from basic (>9) to near-neutral (~7).
Analysis of the products by GC-MS (see Figure 38) shows that the process of sparging and washing does not have a significant effect on the principal organic components of the product, so the fuel value of the product stream is preserved. This analysis also shows that the Stage-1b process reduces the high molecular weight fraction to a small degree and increases the hydrogenation of the process stream. The extent of hydrogenation was quantified by conducting a hydrocarbon type analysis using comprehensive two-dimensional gas chromatography (GCxGC) as summarized in Table 29. This shows that the overall aromatic content after Stage-1a was approximately 60%, and after Stage-1b 42%. Recalling that the feed is essentially 100% aromatics in the form of PAHs, this data suggests that the that the HDA after Stage-1 was on the order of 40% and after Stage-1b it was on the order of 58%. Furthermore, the GC-MS analysis given in Figure 38 shows that the bulk of the product is heavier than decalin, indicating that the fuel value of the feed is preserved and that the hydrocarbon components of the process stream are well suited for upgrading in Stage-2. Finally, the density of the Stage-1a (0.97) and Stage-1b (0.93) products summarized in Table 30 shows that a reasonable degree of hydrogenation is occurring in the Stage-1 reactors and that the product stream is well suited for upgrading. The H/C ratio increases from 0.89 (feed) to 1.42 (Stage-1, Pass1) and 1.53 (Stage-1, Pass 2)
The choice of the catalyst to be used in Stage-2 will depend on the heteroatom content of the process stream leaving Stage-1. The preferred catalyst in Stage-2 would be a noble metal material such as Pt or Pt/Pd. However, if the heteroatom content of the process stream is not appropriate for a noble metal catalyst, then a hydrogenation catalyst such as a sulfided NiW may be necessary. The elemental analysis of the Stage-1 reactor products summarized in Table 29 and the GC-MS analysis of the heteroatom marker components suggest that the two-step Stage-1 process can indeed produce a product low enough in N and S to permit a noble metal catalyst to be considered for Stage-2.
Characterization of Feed and Product with 1HNMR. The syncrude and products from Stage-1 were characterized with 1HNMR. Figure 39 is 1HNMR profile for SC-1 (blue) and product from Stage-1 (red) on “D” catalyst at 1250 psi, 380°C, hydrogen/syncrude weight ratio = 3000 and LHSV 0.3. There are two regions: aromatic and aliphatic. The hydrotreatment causes shift of aromatic to aliphatic. The aliphatic/aromatic ratio based on hydrogen is 0.41 in the feed (blue) and 2.5 in the product 1(red). Further treatment of product 1 with “J” catalyst under same conditions increases the ratio to 5.4 (green). Figure 40 is 1HNMR of WFE SC-2 and product over catalyst “J” catalyst under the same conditions. Same phenomena are observed: hydrogenation of aromatic and the aliphatic/aromatic ratio increases from 0.49 to 4.1. Figure 41 is 1HNMR from the hydrotreatment of WFE-4 over “H” catalyst at 1250 psi, 380°C, hydrogen/syncrude weight ratio = 3000 and LHSV 0.15 on two passes as described in Phase 4, Stage-1a. After Stage-1b the aliphatic/aromatic ratio increases considerably from 0.35 to 11.2 showing that changing the reaction condition and specifically LHSV improved considerably the hydrogenation of aromatics.
7.6.5 Stage-2 Results
As described above, the function of the Stage-1 reactors is to remove the N, O, and S using a robust, sulfided catalyst to condition the feed for upgrading in Stage-2. This upgrading consists of converting the process stream into a mixture of hydrocarbons suitable for blending with conventional jet turbine fuels at levels up to 50% coal-derived fuel. The general requirements for this product includes that it be within the carbon number range of jet fuel (approximately C8 to C18) and be composed of at least 75% paraffins and no more than 25% aromatics, of which no more that 3% of the aromatics can be di-aromatics (naphthalenes). An overall analysis of the WFE SC-4 feed, the final Stage-1b product, and an example jet turbine fuel (JP-8) shown in Figures 42 and 43 along with the hydrocarbon type analysis summarized in Table 22 shows that the conversion of the coal-derived syncrude to a suitable blending stock has been partially completed during the heteroatom removal in Stage-1. Specifically, the hydrocarbon type analysis summarized in Table 22 shows that the aromatic content has been reduced from approximately 100% in the syncrude feed to 42% in the final Stage-1 product. Furthermore, the carbon distribution summarized in Figure 42 shows that the mass fraction of the syncrude heavier than the jet turbine range was reduced from 61% to 24%. This shows that a modest degree of hydrogenation and cracking is needed in Stage-2, therefore it is desirable to process the heavy fraction above the jet fuel range to bring the composition of the process stream to within the program goals.
7.6.5.1 Stage-2 Testing with Model Compunds
For Stage-2 noble metal and sulfided catalysts can be used. The initial evaluation of the Nobel metal “K” catalyst was done with Aromatic-200 model as feed at 300-400°C, LHSV of 1/hr, 400 psi hydrogen, and a H:feed ratio of 10:1 mol/mol. Sulfided “J” catalyst was also evaluated with same model feed 300-380°C, 0.3 LHSV , 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v. The overall results of these analyses are summarized in the GC-MS data shown in Figures 44 and 45 for the ”K” and “J”, respectively, and the hydrocarbon type analysis summarized in Table 31. Briefly, it was found that the “K” catalyst was very efficient at hydrogenating the model feed, but showed little cracking. In contrast, the “J” showed a very high degree of cracking to the point that at temperature above 320°C only light, largely incondensable products were produced. As shown in Figure 44, even at 320°C the product showed a very high degree of cracking with a modest degree of hydrogenation. Therefore, 300°C was chosen as the basis for comparing these two catalysts. As shown in Table 31, the “K” catalyst gave a very high degree of hydrogenation with an HDA of 96% as compared with 40% for “J”. This suggests that “K” would be preferred as a hydrogenation catalyst whereas “J” may serve as a cracking catalyst. Therefore, a combination of these two may be effective in processing the Stage-1 product as a combination of hydrogenation and cracking is indicated.
7.6.5.2 Initial Stage-2 Testing with Stage-1 Product
The first attempt to upgrade a Stage-1 product was made using the WFE SC-2 syncrude processed through a Stage-1 reactor and the “I” NiMo catalyst. Briefly, all of the product from this reactor was combined to form a composite Stage-1 product. This product was used without any additional cleaning steps described above. This gave a Stage-2 feed with 7400 ppm N and 359 ppm S. The Stage-2 reactor configured with the sulfided “J” catalyst and operated the at 380°C with a LHSV of 0.3/hr, 1250 psi hydrogen, and a H feed ratio of 3000:1 v/v. Figure 46 shows a photograph of the Stage-2 feed and product and the overall results are summarized in the GC-MS data shown in Figure 47 and the hydrocarbon type analysis summarized in Table 32. Briefly, it was found that the “J” catalyst gave a product that was clear and lightly colored. The GC-MS analysis shows that the product was a very complex, partially hydrogenated mixture. The hydrocarbon type analysis also shows a modest degree of hydrogenation with an HDA (based on the Stage-2 reactor feed and product) of 21%.
The second attempt to upgrade a Stage-1 product was made using the WFE SC-1 syncrude processed through a Stage-1 reactor and the “I” NiMo catalyst. Briefly, all of the products from this reactor was combined to form a composite Stage-1 product. This product was then sparged with nitrogen for 1 hour, but it was not washed with water as was later adopted for the Stage-1 product. This gave a Stage-2 feed with <5000 ppm N and 167 ppm S. The Stage-2 reactor was configured with the “K” catalyst and was operated the at 300-360°C with a LHSV of 0.3/hr, 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v. Figure 48 shows a photograph of the Zone-2 feed and product and the overall results are summarized in the GC-MS data shown in Figure 49 and the hydrocarbon type analysis summarized in Table 33. Briefly, it was found that the “K” catalyst gave a product that was surprisingly dark. However, as noted previously, the visual appearance often does not correlate with the product quality. The GC-MS analysis shows that the product was a very complex, partially hydrogenated mixture. The hydrocarbon type analysis also shows a modest degree of hydrogenation with an HDA (based on the Stage-2 reactor feed and product) of 22% was observed at 300°C, though this increased to 45% at 360°C. Recall that the refined Stage-1 product described above had an aromatic content of 42%. Reducing this to 25% to meet the goal for a blending stock would require an HDA on the order of 40%, which compares well with the HDA measured for the “K” catalyst. This suggests that a noble metal catalyst should be capable of upgrading the refined Stage-1 product to a suitable blending stock, though some cracking may be needed to reduce the carbon distribution to within the jet-fuel range.
7.6.5.3 Hydrocracking of Stage-1 Product from Intertek Run #1 Over Proprietary Bi-functional Catalyst
One L of Stage-1 product from Intertek Run #1 (hydrotreated syncrude) was sent to UDRI for further treatment at 1250 psi, LHSV = 0.3 hr-1, Hydrogen to liquid volume ratio = 1000. The test was performed at two temperatures (320ºC and 340ºC) over a proprietary, NiW, bi-functional catalyst, obtained by Battelle. The temperature was increased from 320°C to 340°C at 48.4 hours TOS. The sample taken at 9.4 hrs was clear and lightly colored with some sediment. The sample taken at 19.0 hrs was clear and somewhat darker, with significantly more sediment. The samples taken from 27.0 to 65.6 hrs were clear, without any significant sediment. The samples taken at 70.9 and 77.2 hrs appeared dark and possibly cloudy. Some light, non-condensable product was noted in all of samples.
The volumetric product yield was between 90 wt% and 97 wt%, the balance being light hydrocarbons. Effectively, the online GC analysis (not described in this report) showed formation of C1 through C6 products. The density dropped significantly from 0.95 g/cm3 to a value below 0.86 g/cm3, but it increased with TOS. Increasing the temperature from 320ºC to 340°C dropped the density from 0.85 g/cm3 to 0.82 g/cm3, but again started increasing with TOS. Two explanations are possible:
The GCxGC class fraction analysis (see Table 34) shows that hydrocracking of Stage-1 product significantly reduced the amount of total aromatic from 33 wt% to 15 wt%, which is in the range of commercial jet fuel. The tri- and multi-cycloparaffins were converted to mono- and di-cycloparaffins, which indicated that the catalyst was active in cleavage of the C-C bond.
7.7 Catalysts Selection for Pre-Pilot Testing
Based on the results of laboratory-scale catalyst screening, the following catalysts are recommended for pre-pilot testing:
8.0 PRE-PILOT SCALE: SYNCRUDE HYDROTREATMENT AND HYDROGENATION TO JET FUEL/DIESEL
8.1 Objectives
The objectives of this effort were as follows:
Battelle selected two syncrudes that where produced from Quantex Run #5 and #6B for conducting two tests at Intertek for upgrading these to distillate products. The details of the liquefaction tests were described in Section 6.0. The ultimate analyses of these two feedstocks are provided in Table 35. The key difference between the two is that the bio-solvent for Syncrude 6B had a higher H/C ratio and its density was lower.
8.3 Standard Operating Procedures for Handling and Hydrotreatment of Syncrude
Before shipment of syncrude to Intertek for hydrotreatment, a Standard Operating Procedures (SOPs) critical to the safe hydrogenation operations was discussed with Intertek personnel. Some of the SOPs discussed include Safety, Chemical Hygiene, Spill Response, Fire Safety, First Aid, Lockout/Tagout, PPE, Work Place Air Monitoring, and Emergency Shutdown Procedures. In addition, controls to mitigate potential syncrude hydrogenation safety hazards were identified and added to the Test Plan and Environmental Health and Safety Plan.
8.4 Description of Pre-Pilot-Scale Reactors
Hydrotreatment operations were conducted at Intertek in two pilot plant units: P63 and P67. Each unit is composed of two trickle-bed reactors assembled in series. Figure 50 is photo of the P63 which is similar to P67. Each trickle reactor can hold up to 2.5 L of catalyst and it is composed of 6 heating and quenching zones. The temperature of the reactors is monitored at different zones with several thermocouples (TC). Figure 51 shows the catalyst loading and TC locations for Reactor #1 and Reactor #2.
8.5 Pretreatment of Syncrude 5
Battelle shipped 162 kg of syncrude 5 to Intertek. The syncrude was filtered via 1- micron filter at 65°C to remove any residual material left from its preparation at Quantex. The filtered syncrude was loaded into a Fractioneer distillation unit, but the unit plugged and was not able to achieve the desired 500°C cut point. The bottoms material from the Fractioneer unit was therefore transferred to a smaller, single-stage Sarnia unit to achieve the desired 500°C cut point. Distillate from the Sarnia and the Fractioneer were combined and filtered at 65°C and 0.5 micron. About 136 kg of distillate was recovered after filtration and was then hydrotreated.
8.6 Hydrotreatment of Syncrude 5 (Intertek Run #1)
As discussed in Section 7.0, the upgrading was performed in two stages: hydrotreatment, followed by hydrogenation. The details for each stage are provided below.
8.6.1 Stage-1 Reaction (Run #1)
During the first stage, heteroatoms, specifically sulfur, nitrogen and oxygen, are removed by hydrodesulfurization (HDS), hydrodeoxygenation (HDO) and hydrodenitrogenation (HDN). These reactions take place over a sulfided catalyst at a temperature of 380ºC, Liquid Hour Space Velocity (LHSV) of 0.15 hr-1, a pressure of 1300psi and a hydrogen/syncrude volume ratio of 1000. Table 36 provides the reaction conditions used. Figure 52 displays the temperature profile of the reactors during four TOS (time on stream) periods during hydrotreatment. It can be divided into four sections with respect to time on stream.
The analysis of Stage-1 product showed a substantial (>99.9%) removal of S and N, with the product values being 2 ppm and 7 ppm, respectively.
8.6.2. Stage 2 Reaction (Run #1)
The goals of this Stage-2 (hydrogenation) reaction was to reduce the concentration of aromatics present in syncrude from approximately 100 wt% to less than 20 wt% and to reduce the density from ~0.98 g/cm3 to less than 0.90 g/cm3. This reaction was carried out in unit P63, using a proprietary noble-metal catalyst. This catalyst was tested earlier this year at UDRI with syncrude and with model compounds, as discussed in Section 7.6.5. It has a good hydrogenation activity at temperatures between 200ºC and 360°C. Therefore, we conducted the reaction at low temperatures (240°C) and at a relatively high Liquid Hourly Space Velocity (LHSV) of 0.6 hr-1, as compared to hydrotreatment reaction LHSV of 0.15 hr-1at a pressure of 1300 psi and a hydrogen/syncrude volume ratio of 1,000. Around 32 gallons of syncrude was processed over 2.2 L catalyst volume, filled in Reactor #1 of the P63 unit while bypassing Reactor #2. Table 37 summarizes of reaction conditions. The temperature profile of Reactor #1 is displayed in Figure 53. The target temperature was 240°C, but given the exothermicity of the reaction and short time on stream (TOS) of 90 hrs., the temperature varied between 220°C and 260°C.
8.7. Analysis of Feed, Stage-1 and Stage-2 Products from Intertek Run #1
8.7.1 Feed and Product
Figure 54 shows photos of feed and products from the two stages of upgrading. The hydrotreatment converted a heavy crude to refined liquids. The final product was a yellow, transparent liquid, while the feed was composed of mostly heavy, aromatic compounds.
8.7.2 Ultimate Analysis of Stage-1 and Stage-2 Products
The elemental analyses of the feed and products from Stage-1 (hydrotreatment) and Stage-2 (hydrogenation) are shown in Table 38.
The analyses show the following:
The level of hydrogenation of aromatics in Stage-2 was determined by HPLC. As shown in Figure 55, the aromatics content dropped by 96%. The active-metal catalyst used in Stage-2 is known to be superactive; it completely hydrogenated the double bounds in the aromatics present after Stage-1 reaction.
8.7.4 Simulated Distillation of Feed, Stage-1 and-2 Products (Run #1)
Figure 56 shows the simulated distillation of feed as well as Stage-1 and Stage-2 products. It shows that the hydrotreatment in Stage-1 dropped the boiling point significantly. This result is explained by the cleavage of C- heteroatom bonds. In Stage-2, there is only the hydrogenation of double bonds which explains the only slight change in the simulated distillation curve.
8.7.5 Comparison of Batch and Pre-Pilot Results
The results of laboratory testing were compared with the results from pre-pilot testing, with respect to sulfur and nitrogen removal. As shown in Figure 57, for any given LHSV value, the sulfur and nitrogen removal was better in pre-pilot testing. This is likely because the longer catalyst-bed height in the pre-pilot test improved the gas-solid contact.
8.8 Fractionation of the Stage-2 Product (Run #1)
Around 32 gallons of distillate was produced by hydrogenation of hydrotreated syncrude. It was then divided into three portions:
8.9 Hydrotreatment of Syncrude 6B (Intertek Run #2)
A total of 92.3 kg of Syncrude 6B was hydrotreated in the P67 pre-pilot reactor, for Intertek Run #2. The unit details and operating procedures were reported in section 8.3. The first reactor was loaded with a sulfided catalyst and second reactor was loaded with a proprietary, bi-functional hydrocracking catalyst. During hydrotreatment, NH3 is produced, which may negatively impact the performance of hydrocracking catalyst. Therefore, the upgrading was performed in two stages with intermediate N2 purge at atmospheric pressure to minimize the amount of NH3 and H2S, as described below. In Stage-1, heteroatom removal was achieved using Reactor #1 while Reactor #2 was in bypass mode (under hydrogen atmosphere at room temperature), as indicated in Figure 58A. The reaction conditions are reported in Table 40. The Stage-2 operation was performed using two reactors: Reactor #1 to further reduce the heteroatoms, followed by hydrocracking in Reactor #2 (Figure 58B). The general reaction conditions are reported in Table 41.
The Stage-1 reaction was carried out in Reactor #1 during the first 112 hrs. The temperature profile can be divided in two sections: first 30 hrs. heat up period and the period between 30 and 180 hrs where the reactor reached the target temperature (380ºC). The syncrude was injected at temperature around 150ºC. At a total on-stream time (TOS) between 40 and 48 hrs, a hydrogen leak occurred in the lab, which required shut down of the system and resumption of the reaction after controlling the leak. Therefore, the initial, off-spec material was reprocessed. The temperature profile is shown in Figure 59.
The liquid from Stage-1 was processed in Reactor #1 (380°C) and Reactor #2 (320°C) at TOS between 180 hrs. and 300 hrs. (blue zone in Figure 11). The target temperature for Stage-2 was 320°C. However, given the limited amount of feed and the high space velocity (0.4 hr-1) we were not able to adjust the temperature, so we decided to run it at a slightly higher temperature (360°C).
8.10 Analysis of Syncrude and Distillate Products for Run #2
Table 42 summarizes the ultimate analysis of the feed and products from Stages-1 and -2. As shown, the H/C mole ratio increased by 50% in Stage-1 and by an additional 20% in Stage-2. Both sulfur and nitrogen dropped by more than 98% in Stage-1 and by more than 90% of the remaining portion in Stage-2. The density decreased by 8% in Stage-1 and by an additional 8% in Stage-2. The liquid wt% in Table 42 represents the mass recovery from the process.
Table 43 shows the aromatic contents of the syncrude, Stage-1 product, and Stage-2 product. The di-, tri-, and total aromatic concentration decreased with upgrading in Stage-1 and Stage-2. The tri-aromatics and di-aromatic were converted to mono-aromatics. A large portion of the total aromatics was converted to cycloparaffins.
Table 44 summarizes the GCxGC-MS of Stage-1 and Stage-2 products. The Stage-2 catalyst dropped aromatics concentration by at least 60%, increased the cycloparaffins by 100%, and reduced the paraffins by 14%. It looks like the hydrocracking catalyst has good activity in reducing double bonds but limited activity for the cleavage of C-C bonds.
Table 43. Aromatic Composition of Feed (Syncrude) and Stage-1 and Stage-2 Products for Run #1
Figure 60 provides the Simulated-Distillation curves of feed, Stage-1 product, and Stage-2 product. The amount of jet fuel (fraction below 300oC) in the feed is 20%, but increases to 50% and to 70% after Stage-1 and Stage-2, respectively.
8.11 Fractionation of the Distillate from Run #2
A total of 27.3 kg of Stage-2 product was batch distilled to obtain cuts representative of jet fuel products. The material was first cut at 205°C at atmospheric pressure to prevent loss of the light fraction. Afterwards we dropped the temperature to 100ºC and then the unit was brought down to 100 mm Hg absolute pressure. Then, 1-liter volume cuts were taken until the overhead vapor temperature reached 217°C at 100 mm Hg, which is equivalent to 295ºC at atmospheric pressure. A total of 14 (1-liter) cuts were taken in this way. Table 45 summarizes the performed cuts. Figure 61 is a photo of jet fuel fraction and the still pot liquid fraction (residue) with boiling point above 295ºC.
To prepare a jet-fuel fraction that meets flash point specifications, we removed by distillation the light fraction that has boiling point below 160°C. This light fraction corresponds to 4.5 percent volume of the liquid product. A synthetic jet fuel blend was thus prepared using 65% volume from the upper boiling range from cut #1 along with 100% each from cut #2 through cut #7. A detailed characterization of this “neat” jet-fuel product along with a blend with a commercially-used jet fuel is provided in Section 9.1.
The synthetic diesel fraction was prepared by removing the 35% volume from the lower-boiling range from cut #1 along with 100% each of all other cuts. A detailed characterization of this “neat” diesel product is provided in Sec 9.2.
8.12 Comparison of Distillates from Runs #1 and #2
Due to differences in the syncrude properties as well as due to differences in the Stage-2 catalysts, the two upgrading runs produced distillates that had significantly different properties. The results for jet fuel fractions for Run #1 and Run #2 are shown in Table 46. As shown, the jet fuel produced in Run #2 has better properties than jet fuel produced in Run #1: lower freezing point; lower viscosity; lower pressure drop cross the filler; and slightly lower density. It has significantly higher aromatic and n-paraffin content with a concurrent reduction in tri-cycloparaffinic composition. Based on the GCxGC analysis (Table 47), it appears that there is less severe hydrogenation of the Stage-2 product. Although the total aromatic content of Run #2 jet fuel (~17%) is similar to conventional jet fuels, there is an appreciable quantity of 2- and 3-ring aromatic compounds, which is higher than in typical aviation fuels. The distillation step for the Run #2 jet fuel resulted in an improved distillation profile and carbon number distribution as compared to the Run #1 jet fuel sample (Figure 62 and Table 48). The higher concentration of lower molecular weight compounds (e.g., < C9) in Run #2 helped reduce the flash point to 38.9°C from 61.1°C in Run #1.
A comparison of diesel fractions from Runs #1 and #2 is provided in Table 49. A more detail analysis of the diesel product from Run #2, which was significantly better than from Run #1, is provided in Section 9.2.
9.0 CHARACTERIZATION OF DISTILLATE – PRODUCT FRACTIONS AS JET FUEL OR DIESEL
The distillate product from Intertek Run #2 was fractionated into a jet-fuel cut as well as to a diesel cut. The jet-fuel cut was analyzed extensively because jet fuel was the primary focus of this project. However, standard specification testing also was done to analyze the diesel fraction.
9.1 Evaluation of Battelle-CTL-Derived Neat Synthetic Fuel and Synthetic Fuel Blend for Use in Aviation Applications
9.1.1 Introduction
In recent years, significant efforts have been made to develop, evaluate and certify synthetic (e.g., non-petroleum derived) jet fuels for use in commercial and military aircraft. Initial focus was related to the approval of synthetic formulations which could be blended with conventional fuels for use. These efforts resulted in the certification and approval of several types of synthetic fuels as blending feedstocks in commercial and military aviation fuels. Commercial Jet A (ASTM D7566) and military JP-8 (MIL-DTL-83133J) fuel specifications detail requirements for both the currently approved synthetic blending feedstocks and the resulting fuel blends [13-14]. The properties of the synthetic fuel blends must conform to those required for typical petroleum-derived fuels. In addition, each synthetic blending feedstock has specific property requirements included in appendices of the fuel specifications. These latter requirements provide confidence that the fuel blend will conform to all operational and safety needs of current aircraft fuel systems and engines and insure process quality control during production.
Guidance on recommended protocols and methodologies for the evaluation and certification of synthetic aviation fuels and fuel additives was formalized and documented in ASTM D4054-09 and MIL-HDBK-510A. These protocols were developed to facilitate the approval process in a time- and cost-effective manner. The overall process is divided into ‘Tiers’; the proposed test program from ASTM D4054 is shown in Figure 63 [15]. Initial tiers focus on evaluation of chemical and physical properties of the fuel candidate to determine potential suitability prior to larger-scale turbine hot section and component/system-level testing. Fuel specification testing (referred to as Tier 1) is initially performed to determine conformity of the synthetic fuel candidate to physical property specification requirements. Upon determination of acceptable property conformance, more detailed testing (referred to as Tier 2) is performed to evaluate select “Fit-for-Purpose” (FFP) properties. FFP refers to a property required for safe operation which is not directly controlled by the respective fuel specification; a petroleum-derived fuel which meets all fuel specification requirements will inherently satisfy all FFP property needs. The FFP evaluations include several temperature-dependent properties (e.g., density, specific heat, thermal conductivity). Evaluation of all fuel specification and FFP properties outlined in ASTM D4054 and MIL-HDBK-510 requires several gallons of the neat synthetic fuel and blending feedstock [16].
Battelle produced a synthetic aviation fuel candidate from its CTL process based on the use of biomass-derived solvents. The ‘final’ synthetic jet fuel produced by Battelle was provided to the UDRI for preliminary evaluation of the suitability for use as either a neat ‘drop-in’ jet fuel or synthetic bending component. The evaluation approach was based on guidance in the aforementioned certification protocols; however, it was not possible to complete all required fuel specification and FFP property evaluations for the neat synthetic fuel and blend due to insufficient total available volume. Therefore, fuel specification and select FFP properties were evaluated to provide a quantitative basis for preliminary evaluation of the potential suitability of the submitted synthetic fuel for use in aircraft systems. The following sections discuss the analyses and results performed on the neat synthetic fuel and fuel blend.
9.1.2 Evaluation of Neat Synthetic Fuel Formulation
The ‘final’ neat synthetic fuel in the DOE program was produced at Intertek Laboratories in a pilot scale unit using Battelle catalysts and processing conditions and was given the internal designation “54486-38-22”. It was the composite of distillation fractions of cut #1 (65% heavy fraction) trhough cut #7 (see Table 45). The synthetic fuel was analyzed for selected specification (and additional elemental analysis) properties using ASTM test methods by Intertek and analyzed for Hydrocarbon-type (HC-type) composition and carbon number distribution by UDRI. HC-type and carbon number analyses were performed using Two-Dimensional Gas Chromatography (GCxGC) with simultaneous Mass Spectrometry (MS) for species identification and Flame Ionization Detection (FID) for quantitation.
The specific ASTM tests performed were selected to provide guidance regarding suitability of the neat synthetic fuel as a direct ‘drop-in’ aviation fuel (i.e., blending with petroleum-derived fuel not necessary). The results from these ASTM tests are shown in Table 50. The HC-type composition is shown in Table 51 and the carbon number distribution is shown in Table 52. The analyses of the neat synthetic fuel indicated that it was not feasible to use this formulation directly as a ‘drop-in’ synthetic fuel; several properties did not conform to the required Jet A/A-1 and JP-8 property requirements. Specifically, the density (0.885 g/mL) was greater than the maximum specification requirement for aviation fuels (0.840 g/mL), the smoke point (17 mm) did not satisfy the sooting requirement (25 mm), and the hydrogen content (12.48% by mass) was below the military fuel requirement (13.4%).
With respect to other properties, the low temperature viscosity (7.3 cSt) was near the specification requirement (8.0 cSt), but the fuel had an acceptable measured freeze point (< -60 C). Nitrogen and trace elemental analyses showed the fuel had minimal residual inorganics from the production and upgrading processes employed; additional sulfur and nitrogen analyses were performed using non-ASTM test methods (reported in Section 4.2.2/Table 9) and indicated very low heteroatomic content in the synthetic fuel. These results are notable based on the elemental composition of the feedstocks used to produce this synthetic fuel. The distillation profile was similar to conventional aviation fuels; with a T90-T10 of 69.2 C. This range is significantly higher than the current synthetic fuel blending feedstock minimum limit of 22 C, which was implemented to insure a blending feedstock would not result in a significant discontinuity in the fuel distillation profile. The flash point was higher than typical Jet A/A-1 fuels due to the lower concentration of low molecular weight species. However, as discussed in Section 8.11, this can be brought down closer to typical Jet A/A-1 fuel values by not removing some of the <160°C fraction.
The thermal-oxidative stability of the neat synthetic fuel was evaluated via testing with the Jet Fuel Thermal Oxidation Tester (JFTOT) with a test temperature of 325 C. This test condition is very aggressive and has been used for evaluation of previously approved synthetic aviation fuel blending feedstocks; a satisfactory result is an indication that the synthetic fuel should not detrimentally affect the thermal stability characteristics of the fuel blend (per ASTM D7566). The coal-derived synthetic fuel did not pass the JFTOT requirements under this test condition (details regarding potential causes for this behavior will be discussed in Section 9.1.3.1.5). However, this result does not preclude the possible use of this synthetic fuel as a blending feedstock as the specification test temperature for a synthetic fuel blend is 260 C.
The HC-type analysis shown in Table 51 indicted the fuel had a total concentration of aromatics (~18.1% by mass) similar to typical aviation fuels; however, there was a higher quantity of 2- and 3-ring cycloaromatics (i.e., partially hydrogenated) than in typical aviation fuels. Petroleum-derived aviation fuels typical contain appreciable concentrations of iso- (~30-40%) and n-paraffin (~20-25%); however, the neat synthetic fuel had very low concentrations of branched and linear paraffins (approximately 0.7 and 2.2%, respectively). The synthetic fuel was primarily comprised of cycloparaffins (79% by mass), with a very high concentration of di- and tri-cycloparaffins (51.1% and 15.2%, respectively). Petroleum-derived aviation fuels typically contain ~30% cycloparaffins, which are primarily monocycloparaffins. The larger molecular weight cycloparaffins in the synthetic fuel were most likely produced directly from the coal feedstock via fragmentation and hydrogenation of the high molecular weight coal moieties. Likewise, the cycloaromatics in the neat synthetic fuel were produced via incomplete hydrogenation of the upgraded feedstocks. The high concentrations of the di-/tri-cycloparaffins and cycloaromatics are the primary cause for the high fuel density and most likely the reduced smoke point value (e.g., increased sooting tendency).
9.1.2.1 Estimation of Blending Ratio for Synthetic Fuel
Based on properties and compositional analysis of the neat synthetic fuel, it was determined that the viable approach for use in aviation applications was as a blending feedstock with petroleum-derived aviation fuel. Therefore, determination of optimal and maximum blending ratios is necessary to allow detailed evaluation of the resulting specification and selected FFP properties of the synthetic fuel blend. The target blend ratio must be specified such that the synthetic fuel blend conforms to all specification requirements identified for commercial (ASTM D7566/D1655) and military (MIL-DTL-83133J) aviation fuels. The maximum allowable blend concentration is dependent on the specific chemistry and properties of the synthetic fuel. For example, Synthetic Paraffinic Kerosenes (SPKs) derived from either Fischer-Tropsch (FT) Synthesis or Hydoprocessed Esters and Fatty Acids (HEFAs), which are primarily comprised of iso- and n-paraffins, can be blended at concentrations up to 50% by volume provided all specification requirements are satisfied. Depending on the properties of the fuel to which these are blended, the density or aromatic content (or both) of the refined fuel may limit the amount of SPK that can be added to the final blend to less than 50% (per D7566). On the contrary, iso-paraffinic compounds produced via oligomerization of iso-butanol (termer Alcohol-to-Jet Synthetic Paraffinic Kerosene [ATJ-SPK]) can only be blended to a maximum concentration of 30% by volume for use.
Based on the specification and compositional properties shown in Tables 1 and 2, the high density of the neat synthetic fuel (0.885 g/mL) is the primary property limiting the maximum blending ratio for application. Specifically, the synthetic fuel blend must have a final density ≤ 0.840 g/mL for use. Improved thermal stability and smoke point are also expected upon blending (due to dilution). The density (@ +15 C) of typical aviation fuels range from approximately 0.780 to 0.830 g/mL, with an average of approximately 0.803 g/mL. The resulting density of a synthetic and petroleum-derived aviation fuel blend is expected to be linear with the volumetric blending ratio as the fluids are expected to behave as ideal fluids; this has previously been shown during blending of FT-derived SPK with aviation fuels [17]. Calculations were performed to investigate the effect of blending ratio and petroleum-derived fuel density on the resulting synthetic fuel blend density. Figure 64 shows the estimated density values of the fuel blend as a function of blending ratio with aviation fuels with densities of 0.780, 0.803 and 0.830 g/mL, respectively. As shown, the primary limiting factor occurs when the petroleum-derived fuel has a high density (e.g., maximum blend percentage of ~18% by volume when petroleum fuel has a density of 0.830 g/mL). However, fuels with lower density values (which comprise the majority of aviation fuels) allow higher blend percentages with the neat synthetic fuel. Based on the density trends shown in Figure 2, the recommendation was made to evaluate the corresponding specification and FFP properties for a 20% blend by volume of the synthetic fuel with an ‘average’ (or ‘nominal’) jet fuel. This provides guidance on the expected performance when using the synthetic fuel as blending feedstock and will most likely mitigate potential issues related to the concentration of certain compound classes which are higher than typically present in aviation fuels (e.g., di-/tri-cycloparaffins).
9.1.3 Evaluation of Synthetic Fuel Blend
Evaluation of the specification and selected Fit-For-Purpose properties of a 20% blend by volume of the synthetic fuel with a petroleum-derived aviation fuel was performed to provide preliminary guidance regarding the potential suitability for use in aviation applications. Subsequent testing and evaluation as outlined in ASTM D4054 and MIL-HDBK-510 is necessary to provide sufficient data to determine suitability for pursuing certification and approval of the synthetic fuel as a blending feedstock. Blending was performed using a Jet A fuel provided by the Fuels & Energy Branch of the Air Force Research Laboratory. This specific Jet A (with internal identification POSF 10325) has been termed an ‘average’ (or ‘nominal’) jet fuel as the specification properties are very close to historical averages for aviation fuels. This specific Jet A has been used as the ‘average’ fuel in the Federal Aviation Administration (FAA) National Jet Fuel Combustion Program (NJFCP). Results obtained for the synthetic fuel blend will be compared to those for the neat Jet A and are discussed in the following sections.
Due to the limited quantity of batch 54486-38-22, Battelle prepared a “second batch” of neat synthetic fuel from the same process streams; this batch of neat synthetic fuel was given the internal designation “54486-39-16”. The composition of 54486-39-16 was compared to the initial batch (54486-38-22) to determine suitability for testing. HC-type composition and carbon number distribution analyses (shown in Section 4.1) indicated the compositions of the two batches were sufficiently similar to proceed with blend preparation and preliminary evaluation for this program.
Battelle had a limited overall volume of the neat synthetic fuel 54486-39-16 available for blend preparation and testing, which precluded completion of all recommended Tier 1 and 2 testing and evaluations in this effort. Therefore, the decision was made to evaluate all fuel property specification requirements; specific FFP evaluations were selected which would provide detailed insight regarding the suitability of the blend for use. The 20% blend by volume of the neat synthetic fuel (54486-39-16) with the ‘nominal’ jet fuel (POSF 10325) was given the internal designation “54486-39-26”. Results from these tests and pertinent discussion are included in the following sections.
9.1.3.1 Chemical Composition and Specification Properties
The chemical composition and aviation fuel ASTM specification fuel properties were evaluated for the 20% by volume synthetic fuel blend. HC-type analysis obtained using GCxGC analysis and ASTM D2425/6379 are shown in Tables 53 and 54 (a more detailed summary of HC-type analysis is included in appendix of report), carbon number distributions are shown in Table 55, and ASTM specification properties are presented in Table 56. Results for the Jet A used for blending and the neat synthetic fuel(s) are included for comparison.
9.1.3.1.1 Composition
Blending of the neat synthetic fuel at 20% by volume with petroleum-derived Jet A resulted in an overall chemical composition similar to typical aviation fuels, with the exception of the cycloaromatic and di-/tri-cycloparaffins content. The concentration of these latter compound classes is higher than typically observed in aviation fuels. This result was expected based on the coal feedstock used for production of the synthetic blend feedstock. Differences were observed between hydrocarbon type analyses performed on the neat synthetic fuel and blend using GCxGC and ASTM D2425/D6379. However, there was good agreement between the techniques for the ‘nominal’ Jet A fuel. The differences most likely occur since the ASTM techniques were developed for petroleum-derived aviation fuel; therefore, the GCxGC results provide improved accuracy of the overall fuel composition. A comparison of the carbon number distribution data from Table 55 is shown in Figure 65. The synthetic fuel had a carbon number distribution slightly narrower than typical aviation fuels, but with an acceptable range of molecular weights. The higher concentration of C10-C14 compounds in the synthetic fuel and blend is due to high concentrations of multi-ring cycloparaffins and cycloaromatics (e.g., decalin, tetralin, dodecahydro-acenaphthylene, perhydrophenalene). The impact of increased concentrations of these higher molecular weight cycloaromatics and cycloparaffins will be more apparent when comparing combustion and FFP characteristics of the synthetic fuel blend. If necessary, subsequent processing could be performed to reduce the concentration of these compound classes in the neat synthetic fuel via either further distillation or hydrocracking.
The compositional characteristics determined by ASTM specification testing shown in Table 56 indicated the synthetic fuel blend conformed to the primary compositional properties required for aviation fuels containing synthesized hydrocarbons. More specifically, the acid number, total and mercaptan sulfur and hydrogen content all met current specification requirements for synthetic fuel blends. The total aromatic content for the synthetic fuel blend (15.6%) was lower than expected based on the composition of the neat synthetic fuel and Jet A; a repeat analysis via ASTM D1319 resulted in a total of 18% aromatics by volume. These quantified aromatic values are within the measurement uncertainty of the ASTM technique employed.
9.1.3.1.2 Volatility
The volatility characteristics of the synthetic fuel blend satisfied all current fuel specifications and were similar to a ‘nominal’ aviation fuel. Although the flash point of the neat synthetic fuel (62.5 C) was near the high end of typical Jet A fuels, the synthetic blend had a flash point similar to the Jet A used for blending. This behavior is a result of the flash point being primarily dependent on the most volatile components in the fuel. The distillation profile of the blend was very similar to the nominal Jet A, which is a favorable characteristic during application. The distillation profiles for the synthetic blend and ‘nominal’ Jet A are shown in Figure 66. Included are distillation profiles for two additional jet fuels, a JP-8 (POSF 10264) and JP-5 (POSF 10289), which have also been used in the FAA NJFCP. The JP-8 has been termed a ‘best case’ jet fuel (e.g., low viscosity, aromatics and flash point with high H-content) while the JP-5 is a ‘worst case’ jet fuel (e.g., high viscosity and flash point with low H-content). As shown in Figure 4, the volatility profile of the synthetic fuel blend is within the bounds of the two reference fuels. The blend density (0.820 g/mL) was within the specification limits and is identical to the value calculated based on the blending ratio of the two fuels. A favorable characteristic of the synthetic aviation fuel produced in this program is that blending will increase the density relative to the neat petroleum-derived fuel.
9.1.3.1.3 Fluidity
The fluidity characteristics of the synthetic fuel blend satisfied the aviation fuel specification requirements for fuel freeze point and low temperature viscosity (at -20 and -40 C). The freeze point behavior was expected due to the low value for the neat synthetic fuel. Blending of the synthetic fuel with the Jet A only slightly increased the viscosity relative to the neat petroleum-derived fuel (from 4.4 to 4.9 cSt at -20 C and 9.1 to 10.3 cSt at -40 C). This is a favorable result as the resulting viscosity was within specification requirement. The low temperature viscosity characteristics of the synthetic fuel blend will be further discussed below (See Section 9.1.3.2.1).
9.1.3.1.4 Combustion
The primary combustion characteristics in the aviation fuel specification requirement pertain to the energy content (i.e., Heat of Combustion) and sooting propensity. The synthetic fuel blend heat of combustion (43.1 MJ/kg) satisfied the specification requirement, as expected based on the high energy content for the neat synthetic fuel (45.1 MJ/kg). As discussed in Section 3.0, the high concentrations of high molecular weight cycloparaffins and cycloaromatics were most likely the cause for the low smoke point of the neat synthetic fuel (17 mm). However, blending with the petroleum-derived fuel resulted in an improvement to the smoke point (20.9 mm). This increased value with the corresponding concentration of naphthalenes (1.2% by volume) satisfy the fuel specification requirement. Further evaluation of the impact of the higher molecular weight cyclic compounds on combustion performance/stability and sooting propensity, as recommended in the ASTM D4054 and MIL-HDBK-510A processes, may be necessary to insure suitability of the fuel for use.
9.1.3.1.5 Thermal Stability
The thermal-oxidative stability of aviation fuels is evaluated using the JFTOT; the specification requirement for a synthetic fuel blend is a passing rating at 260 C. The synthetic fuel blend passed the JFTOT specification at this test temperature with a Visual Tube Rating (VTR) of < 1 and Filter pressure drop of 14 mm Hg. The VTR result demonstrated negligible tube deposition during stressing. However, the filter pressure drop of the blend was higher than observed for the neat Jet A. As discussed in Section 9.1.2, the neat synthetic fuel did not obtain a passing rating at 325 C, which is the test condition previously used for qualifying synthetic fuel blending feedstocks. However, passing at 325 C may not be necessary for eventual certification and use as the specific requirements for a synthetic fuel blending feedstock are determined based on the feedstock and production processes employed. Further evaluation, including determination of the JFTOT breakpoint (highest test temperature at which there is a passing result), is necessary to provide improved insight into the overall thermal stability characteristics of this fuel. Potential causes for the observed behavior for the neat synthetic fuel and the blend may be related to the presence of high molecular weight polar compounds (e.g., oxygenated multi-ring cycloparaffins and cycloaromatics). Further evaluation of fuel blend thermal-oxidative stability characteristics was performed using a Quartz Crystal Microbalance, and will be discussed in Section 9.1.3.2.3.
9.1.3.1.6 Contaminants/Corrosion/Wear
Several aviation fuel specification tests are performed to verify the fuel has acceptable compatibility and performance characteristics with minimal contamination. The synthetic fuel blend passed the existent gum, microseparator (MSEP; measure of impact on water coalescing/removal performance) and copper corrosion requirements. As expected, the blend had zero conductivity due to the low heteroatomic composition. If the conductivity of the blend would need to be increased, this can be addressed via use of static dissipater additive. The lubricity of the fuel blend (wear scar of 0.64 mm) is below the maximum allowable value of 0.85 mm.
9.1.3.2 Fit-For-Purpose Properties
The synthetic fuel blend met all current specification requirements for an aviation fuel which contains synthesized hydrocarbons. However, further evaluation of the performance and compatibility of the synthetic fuel blend beyond those characteristics evaluated by the specification properties is necessary to determine suitability for implementation and use on aviation platforms. These are referred to as “Fit-for-Purpose” (FFP) properties. FFP properties refer to a property required for safe operation which are not directly controlled by the respective fuel specification; a petroleum-derived fuel which meets the fuel specification will inherently satisfy all FFP property requirements. ASTM D4054 and MIL-HDBK-510 provide guidance on pertinent FFP tests to be performed when evaluating synthetic fuels and additive for use in aviation applications. Specific areas of recommended tests include those shown in Figure 1. Additional recommended testing may arise depending on results from the specification and compositional property results. FFP properties do not have well defined limits; rather the effect of the proposed fuel or additive on the corresponding FFP property must fall within the scope of experience of the engine and airframe manufacturers. The results from the FFP testing provide the basis for the FAA, DOD and aircraft engine/airframe OEMs to evaluate the potential suitability of the proposed fuel/additive for use and determine if further testing and evaluation is warranted. Completion of all recommended FFP tests requires several gallons the synthetic fuel blend for completion.
There was insufficient total volume of the synthetic fuel blend to complete all recommended FFP tests. Therefore, specific FFP tests were selected based on the available volume and composition/specification test results which would provide guidance regarding the potential suitability of the synthetic fuel blend for use in aviation applications. The selected FFP property tests performed in this effort are shown in Table 57. Subsequent evaluation of other FFP characteristics as outlined in the recommended approval guidelines must be completed to provide the basis for determination if the synthetic fuel blend is suitable for further evaluation and use. The following section discuss results from the specific FFP testing performed in this effort.
9.1.3.2.1 Low Temperature Viscosity
The synthetic fuel blend satisfied the specification requirements for viscosity at -20 and - 40 C; however, additional guidance regarding the corresponding fluid behavior within the low temperature regime is beneficial. Therefore, the low temperature dynamic viscosity characteristics of the neat Jet A and synthetic fuel blend were measured from - 20 to < -56 C using a Brookfield Scanning Viscometer with a cooling rate of -5 C/hr. The dynamic viscosity is converted to kinematic by normalizing to the corresponding temperature-dependent density. Results from these measurements are shown in Figure 67. Both the Jet A and synthetic fuel blend exhibit characteristics consistent with typical aviation fuels; the viscosity increases with decreasing temperature until there is a rapid increase near the phase transition (e.g., crystallization) temperature. For aviation fuels, this is typically due to crystallization of long chain n-paraffins in the fuel. The addition of the synthetic fuel suppressed this transition by approximately 2 C. The addition of the synthetic fuel slightly increased (~1-2 cSt) the kinematic viscosity relative to the baseline Jet A; however, this increase is minor and not expected to result in a significant performance impact.
9.1.3.2.2 Analysis of Polars and S/N Content
Analysis of the heteroatomic and polar composition was performed to provide insight into the quantitative levels of these classes of species in the synthetic fuel blend. Heteroatomic and polar species are known to affect thermal-oxidative stability of aviation fuels and can impact material and storage compatibility. The sulfur and nitrogen content (total and polar) of the neat synthetic fuel, Jet A and neat synthetic blend were quantified using Gas Chromatography with Sulfur and Nitrogen Chemiluminescence Detection; results are shown in Table 58. The neat synthetic fuel had very low sulfur/nitrogen content, demonstrating that the processes and catalysts used in the fuel production were effective at removing the heteroatomic species.
The polar composition of the fuels was investigated via extraction using silica gel solid phase extraction. Retained polar compounds were eluted with methanol and analyzed using GCxGC-FID/MS to quantify the respective classes of polar compounds in the fuels; results are shown in Table 59. The Jet A fuel had ‘typical’ levels of polar compounds, primarily phenols and lower molecular weight compounds. Although the neat synthetic fuel had a similar total polar content, the respective species were of higher molecular weight (primarily 2-ring). The presence of these types of species is due to the composition of the neat synthetic fuel (e.g., tetralone related to high concentration of tetralin). The oxygenated species are either residual in the fuel following synthesis or formed via oxidation during storage. The presence of these types of species may impact the thermal-oxidative stability and seal swell propensity during application (discussed below). The polar content of the synthetic fuel blend was a result of blending of the neat synthetic and Jet A fuels.
9.1.3.2.3 Thermal-Oxidative Stability via Quartz Crystal Microbalance
The thermal-oxidative characteristics of the neat synthetic fuel and blend were previously analyzed via JFTOT testing. The synthetic fuel blend passed the specification criteria at a test temperature of 260 C; however, the neat synthetic fuel failed at an elevated test temperature of 325 C. Improved insight into potential causes for these results would be useful regarding the suitability of the synthetic fuel for use in aviation applications. Therefore, the thermal-oxidative stability of the synthetic fuel blend was further evaluated using a Quartz Crystal Microbalance (QCM). QCM analysis has primarily been used to provide insight regarding oxidation and deposition tendencies of jet fuels due to differences in trace chemical composition (e.g., heteroatomic species and dissolved metals content).
The thermal-oxidative stability characteristics of the Jet A and synthetic fuel blend were assessed using the QCM operated under typical experimental conditions (i.e., 140°C, air saturated fuel, 15 hours). These experimental conditions are sufficient to identify inherent differences in oxidation rate and level of deposit accumulation of aviation fuels. Results from this testing are shown in Figure 68. The Jet A fuel is a slow oxidizer; deposit accumulation occurs as oxidized fuel species are produced and undergo further molecular growth reactions. The oxidation and deposition characteristics of the synthetic fuel blend were qualitatively similar to the Jet A fuel; however, the rate of oxidation and level of deposition were increased relative to the petroleum-derived fuel. Typical aviation turbine fuels produce about 1 to 6 g/cm2 of deposition under these test conditions; the synthetic fuel blend is near the high end of this range.
The impact of the addition of the synthetic fuel on the corresponding thermal-oxidative stability characteristics may be related to the polar composition and the primary species formed during oxidation of the synthetic fuel. As discussed in Sections 4.1 and 4.2.2, the synthetic fuel had a higher concentration of 2 and 3-ring cycloparaffins and cycloaromatics (and higher molecular weight polar compounds) than typically observed in aviation fuels. These species may be more prone to molecular growth via oligomerization reactions during oxidative stressing; similar behavior had been observed during a previous study investigating the impact of aromatic addition to SPK fuels [18]. Pretreatment of the synthetic fuel prior to blending (e.g., extraction of polars via silica or clay filtration) and addition of an antioxidant could possibly improve the corresponding thermal stability behavior. Although the neat synthetic fuel may be inherently more prone to deposit formation than a typical aviation fuel, the corresponding behavior is within ranges of typical aviation fuels when blended at reasonable concentrations (i.e., ≤ 20% by volume). Further evaluation of the thermal stability characteristics, such as JFTOT breakpoint or flowing system testing, would provide improved insight regarding the impact of the synthetic fuel during use. This could assist in determining if processing modifications or post-production treatments, such as filtration or addition of antioxidant, could further improve the thermal stability characteristics of the synthetic fuel blend.
9.1.3.2.4 Ignition Quality Test (Derived Cetane Number)
The Derived Cetane Number (DCN) of a fuel is primarily relevant to use in diesel engines. DCN is a measure of the combustion speed during compression needed for ignition. This is an important property for use of military aviation fuels in ground support systems and vehicles. The DCN can be measured using an Ignition Quality Tester (IQT) per ASTM D6890. An acceptable DCN range of 40-65 has been recommended during qualification of synthetic aviation fuels (per MIL-HDBK-510A). The Jet A (POSF 10325) used in the blend formulation had a measured DCN of 48.3, while the synthetic fuel blend had a measured value of 45.5. The synthetic fuel blend value is within the recommended range for DCN and would be expected to perform satisfactorily during use.
9.1.3.2.5 Initial Material Compatibility (Seal Swell)
Compatibility of candidate synthetic fuels and additives with fuel system materials is critical to insure suitability for use without adverse effects. Materials of interest include metallic/non-metallic components and elastomers. The commercial and military fuel certification processes describe recommended approaches for evaluating the physical properties of materials following extended exposure to the test fuel. This includes aging the materials in the test fluids for extended durations and comparing the impacts to those observed with ‘typical’ petroleum-derived fuels. Compatibility testing for all fuel system materials requires very large volumes of fuel for completion. However, initial material compatibility evaluations can be performed via investigation of seal swell characteristics of elastomers. Elastomer seal swell has been shown to be critical during the certification of synthetic fuel blends to insure there is sufficient swell to prevent leakage in aircraft fuel systems. Therefore, the volume swell characteristics of selected O-rings exposed to the synthetic fuel blend were evaluated and results were compared to a historical reference data sets for normal volume swell behavior in aviation turbine fuels. This testing has previously been used to provide detailed information regarding material compatibility of potential synthetic fuel formulations while requiring very small fuel volumes for testing.
The volume swell of the synthetic fuel blend was evaluated using three different types of elastomers which are the most prevalent fuel wetted o-ring materials in conventional aircraft. Specifically, nitrile rubber (Identification Number N0602), fluorosilicone (L1120), and fluorocarbon (V0747) O-rings manufactured by Parker Hannifin were used in this evaluation. The plasticizer was removed from samples of the N0602 nitrile rubber using acetone solvent extraction and designated N0602e. For each analysis, two size -001 O-rings were cut in half with one section from each O-ring being used for this analysis. Volume swell characteristics were measured by performing optical dilatometry at room temperature. UDRI has used optical dilatometry for evaluating the impact of synthetic fuels on the seal swell of various elastomeric materials, including the initial evaluation of synthetic SPK fuel blends during certification for use in B-52 and C-17 aircraft. This technique requires a very small volume of fuel for evaluation and significantly shorter test durations than conventional ASTM soak tests. Briefly, two samples were placed in a reservoir with 5 mL of the test fuel. Starting at 1 minute after being immersed in the fuel, the samples were digitally photographed every 20 seconds for the next 3 minutes. At 5 minutes total elapsed time, the samples were photographed every 30 minutes for the next 40 hours. After the aging period was completed, the cross-sectional area was extracted from the digital images and taken as a characteristic dimension proportional to the volume. The results reported below are the average values obtained from the two samples. At the completion of the aging period, the absorbed fuel was extracted and the composition was analyzed by GC-MS. By comparing the GC-MS analysis of the fuel absorbed by the O-ring with that of the neat fuel, the relative solubility of the major classes of fuel components were summarized in terms of their respective polymer-fuel partition coefficients (Kpf). The major class fractions were taken as the normal-, and iso-alkanes, normal- and iso-alkyl benzenes, naphthalene, and alkyl naphthalenes. These were isolated from the GC-MS total ion chromatograms using ions 57 (n,i-alkanes), 105 (principally the C3 substituted alkyl benzenes), 128 (naphthalene), and 141 (C1 and C2 substituted naphthalenes). Furthermore, cycloalkanes and tetrahydronaphthalene (tetralin) were analyzed using ions 83 and 104, respectively.
The basis for comparison was the average values and 90% prediction intervals for the volume swell and partition coefficients obtained from 12 reference JP-8s with aromatic contents from 10.9% to 23.6% by volume (v/v) (prior studies have shown that the volume swell behavior of JP-8 is similar to Jet-A). It should be noted that the prediction interval is a statistical estimate for the range of values that would be exhibited by 90% of all individual JP-8s, and therefore reflects the estimated range of behavior for ‘normal’ JP-8s with 10-25% aromatics. The results summarized in Table 60 and Figure 69 show that the volume swell behavior of the synthetic fuel blend and the Jet A used for blend preparation were well within the normal range typically observed for JP-8 for the nitrile rubber and fluorosilicone materials. The volume swell of the fluorocarbon was somewhat lower than average, likely due to the lack of Fuel System Icing Inhibitor additive (Di-Ethylene Glycol Monomethyl Ether) which is required in military aviation fuels. However, the absolute difference is very small and does not indicate a potential issue with either the Jet A or synthetic fuel blend.
The overall absorption of fuel by the sample O-rings is summarized in Table 61 and Figure 70. These results show that the average solubility of all major class fractions examined are within the range typically observed for JP-8s. Note that the solubility of the cycloalkanes in the test fuels with nitrile rubber is about 10% higher than exhibited by linear and branched alkanes, which is consistent with prior work. The solubility of tetralin, a significant component of the synthetic fuel, was modestly higher than the C3 alkyl benzenes, but lower than the alkyl naphthalenes. This suggests that the presence of the synthetic fuel components slightly elevates the overall solubility in the elastomer which increased the volume swell character of this fuel blend as compared to the neat Jet A.
Overall, the volume swell character of the synthetic fuel blend is within the range typically observed for aviation turbine fuel and should therefore be compatible with the O-ring materials studied. Further evaluation of the compatibility characteristics of the synthetic fuel with other fuel system and engine materials may be warranted.
Figure 70. Example GC-MS chromatograms of similar volumes (approximately 1 L)
of synthetic fuel blend and fuel components absorbed by elastomers (see Table 11
for Parker Identification Number). Primary compounds are identified.
chromatograms scaled for comparison. Note: Peak assignments tentative based on
NIST mass spectral library.
Advertisement
The 10 largest coal producers and exporters in Indonesia:
Battelle has demonstrated a novel and potentially breakthrough technology for a direct coal-to-liquids (CTL) process for producing jet fuel using biomass-derived coal solvents (bio-solvents). The Battelle process offers a significant reduction in capital and operating costs and a substantial reduction in greenhouse gas (GHG) emissions, without requiring carbon capture and storage (CCS). The results of the project are the advancement of three steps of the hybrid coal/biomass-to-jet fuel process to the technology readiness level (TRL) of 5. The project objectives were achieved over two phases. In Phase 1, all three major process steps were explored and refined at bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a synthetic crude (syncrude); and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, all three subsystems of the CTL process were scaled up to a pre-pilot scale, and an economic analysis was carried out.
A total of over 40 bio-solvents were identified and prepared. The most unique attribute of Battelle’s bio-solvents is their ability to provide much-needed hydrogen to liquefy coal and thus increase its hydrogen content so much that the resulting syncrude is liquid at room temperature. Based on the laboratory-scale testing with bituminous coals from Ohio and West Virginia, a total of 12 novel bio-solvent met the goal of greater than 80% coal solubility, with 8 bio-solvents being as good as or better than a well-known but expensive hydrogen-donor solvent, tetralin.
The Battelle CTL process was then scaled up to 1 ton/day (1TPD) at a pre-pilot facility operated in Morgantown, WV. These tests were conducted, in part, to produce enough material for syncrude-upgrading testing.
To convert the Battelle-CTL syncrude into a form suitable as a blending stock for jet turbine fuel, a two-step catalytic upgrading process was developed at laboratory scale and then demonstrated at pre-pilot scale facility in Pittsburg, PA. Several drums of distillate products were produced, which were then distilled into unblended (neat) synthetic jet fuel and diesel products for a detailed characterization. Based on a detailed characterization of the synthetic jet fuel, a 20% synthetic, 80% commercial jet fuel blend was prepared, which met all specifications. An analysis of the synthetic diesel product showed that it has the promise of being a drop-in fuel as super-low (less than 15 ppm)-sulfur diesel fuel.
A detailed economic analysis showed that the Battelle liquefaction process is economical at between 1000 metric tons/day (MT/day) and 2000 MT/day. The unit capital cost for Battelle CTL process for making jet fuel is $50K/daily bbl compared to $151K/daily bbl for indirect CTL, based on 2011 dollars. The jet-fuel selling cost at the refinery, including a 12% capital cost factor (which included profit), for the Battelle CTL process is $61/bbl ($1.45/gallon). This is competitive with crude oil price of $48/bbl. At the same time, the GHG emissions of 3.56 MT CO2/MT fuel were lower than the GHG emissions of 3.79 MT CO2/MTfuel for petroleum-based fuels and 7.77 MT CO2/MT fuel for indirect CTL. Thus, the use of bio-solvents completely eliminates the need for carbon capture in the case of Battelle CTL process. The superior economics and low GHG emissions for the Battelle CTL process has thus sparked worldwide interest and some potential commercialization opportunities are emerging.
1.0 EXECUTIVE SUMMARY
Battelle has demonstrated a novel and potentially breakthrough technology for a direct coal-to-liquids (CTL) process for producing jet fuel using biomass-derived coal solvents (bio-solvents). The Battelle process offers a significant reduction in capital and operating costs and a substantial reduction in greenhouse gas (GHG) emissions, without requiring carbon capture and storage (CCS). The results of the project are the advancement of three steps of the hybrid coal/biomass-to-jet fuel process to the technology readiness level (TRL) of 5. The project objectives were achieved over two phases. In Phase 1, all major process steps were explored and refined at bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a synthetic crude (syncrude); and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, all three subsystems of the CTL process were scaled up to a pre-pilot scale.
Biomass-Derived Bio-solvents. While the goal of Phase 1 was to identify and prepare 6 hydrogen-donor bio-solvents, we actually prepared a total of over 40 bio-solvents. The raw materials for these, mostly non-edible bio-solvents, are believed to be readily available around the world. In some cases, the biomass feedstocks are commercially available from specialty chemical companies though none have previously been used for CTL processing. In many cases, commercially-available feedstocks had to be modified by Battelle to provide the desired solvation and other physical properties. The most unique attribute of Battelle’s bio-solvents is their ability to provide much-needed hydrogen to liquefy coal and thus increase its hydrogen content so much that the resulting syncrude is liquid at room temperature.
Direct Coal Liquefaction. Coal liquefaction tests were done in several different reactor systems. The majority of parametric testing, especially to down-select preferred bio-solvents, was done at Battelle, using a 0.5L autoclave system. The resulting products were analyzed to determine the coal solubility, defined as the yield of THF-soluble fraction. The viscosity of the THF-free syncrude was also measured at 50°C to assess the degree of hydrogen transferred from the bio-solvent to coal-derived liquids. The initial testing was done with a bituminous coal from West Virginia, followed by more extensive testing on an Ohio bituminous coal. The various bio-solvents were compared to tetralin, which is a well-known hydrogen-donor solvent, as well as soybean oil, which was referenced in prior art. The Battelle tests showed that a total of 12 novel bio-solvent met the goal of greater than 80% coal solubility, with 8 bio-solvents being as good as or better than tetralin. On the other hand, soybean oil gave solubility below 70% and the product was very viscous. The viscosity of the product with preferred bio-solvents was an order-of-magnitude lower than with soybean oil. The solubility goal was also met with a sub-bituminous coal from the Powder River Basin. The Battelle parametric testing was supported by microcatalytic-reactor testing by Pennsylvania State University (PSU).
The Battelle CTL process was scaled up to 1 ton/day (1TPD) at a pre-pilot facility operated in Morgantown, WV by Quantex. These tests were conducted, in part, to produce enough material for syncrude-upgrading testing. The Quantex plant required several changes to adequately carry out liquefaction. The preliminary data show that the solubility and syncrude viscosity are comparable to those attained during autoclave testing at Battelle.
Upgrading the CTL Syncrude to Distillates. To convert the Battelle-CTL syncrude into a form suitable as a blending stock for jet turbine fuel, a two-step catalytic upgrading process was developed at laboratory scale and then demonstrated at pre-pilot scale facility operated by Intertek. For the first step (Stage-1), a number of commercially-available hydrotreatment catalysts were tested for removal of heteroatoms, most notably sulfur, nitrogen, and oxygen. A proprietary, sulfided catalyst was selected for demonstration. For Stage-2, two proprietary catalysts for cracking and hydrogenation of Stage-1 product were successfully demonstrated at pre-pilot scale. Several drums of distillate products were produced, which were then distilled into unblended (neat) synthetic jet fuel and diesel products for a detailed characterization.
Characterization of Distillate Fraction as Jet Fuel or Diesel. A synthetic jet-fuel produced from the Battelle CTL process was evaluated by UDRI to determine potential suitability for use in aviation applications. Efforts focused on testing of the specification and limited Fit-For-Purpose (FFP) properties of the neat synthetic fuel and a blend with petroleum-derived aviation fuel, and followed recommended protocols for certification of synthetic fuels for commercial and military applications. Analyses of the neat synthetic fuel indicated it was not feasible to use the current formulation as a direct “drop-in” fuel as a couple of the properties did not conform to required Jet A/JP-8 specification requirements. Overall, the results indicate that the synthetic fuel has the potential for use as a synthetic blending feedstock for aviation applications. Based on the analyses and testing, it appears feasible to make slight modifications to the syncrude upgrading process to better tailor the final synthetic fuel to aviation applications. An analysis of the synthetic diesel product showed that it has the promise of being a drop-in fuel as super-low (less than 15 ppm)-sulfur diesel fuel.
Economic Analysis. A detailed economic analysis was carried out to show that the liquefaction process is economical at between 1000 metric tons/day (MTPD) and 2000 MTPD, which is an order of magnitude smaller than commercially-available indirect CTL processes. The elimination of the need for gaseous hydrogen and a catalyst in the Battelle liquefaction process leads to process simplifications that greatly reduce capital and operating costs. A plant design using 4 Battelle CTL plants and a single syncrude upgrading plant producing 32,000 barrels (bbl) per day of jet fuel (and/or diesel) was compared to a 19,000 MTPD FT-technology based indirect CTL plant producing 50,000 bbl/day (BPD) of jet fuel plus diesel plus naphtha. The unit capital cost for Battelle CTL process is $50K/daily bbl compared to $151K/daily bbl for indirect CTL, using 2011 costing basis required by DOE. The jet-fuel selling cost at the refinery, including a 12% capital cost factor (which included profit), for the Battelle CTL process is $58/bbl ($1.38/gallon). This is competitive with crude oil price of $46/bbl. The selling price for the indirect CTL process was much higher at $95/bbl. An analysis also showed that the unrefined syncrude from the Battelle CTL process could also be sold to petroleum refineries at $32/bbl. No premiums were placed on either the syncrude or the jet fuel or diesel from the Battelle CTL process for having an ~40% bio-content. This however was a major factor in meeting the GHG reduction goals.
Greenhouse Gas (GHG) Analysis. A GHG emissions analysis was performed by Prof Bhavik Bakshi of The Ohio State University (OSU). The total GHG emissions from well (for petroleum) or coal mine/biomass-feedstocks source-to-wheels (e.g., well-to-wheels [WTW] for petroleum-based jet fuel) were estimated. While the baseline WTW value for petroleum-based jet is 3.79 MT CO2/MT fuel, the Battelle-CTL jet fuel GHG emissions were somewhat lower at 3.56 MT CO2/MT fuel. On the other hand, the GHG emissions on the same basis for FT-based jet fuel was much higher at 7.77 MT CO2/MT fuel. For the FT process to meet the Section 526 of EISA 2007 goal of being no worse than petroleum-to-jet baseline, about 90% of the pre-combustion GHG emissions from CTL will need to be controlled by CCS. However, no CCS will be required for the Battelle CTL process.
The superior economics and low GHG emissions for the Battelle CTL process has thus sparked worldwide interest and some potential commercialization opportunities are emerging. Thus, the project goal of demonstrating a fast, straight forward path to commercialization has also been achieved through this project.
2.0 PROJECT OBJECTIVES
Battelle has invented a potentially breakthrough direct coal-to-liquids (CTL) technology using biomass-derived solvents. The objectives of this project were as follows:
- Advance the Battelle CTL process technology to technology readiness level (TRL) 5, which involved pre-pilot-plant scale testing.
- Demonstrate that the process is applicable to a variety of coals, achieving at least 80% coal conversion to synthetic crude (syncrude).
- Demonstrate that the syncrude from the Battelle CTL process can be upgraded to jet fuel and, if desired, diesel.
- Demonstrate that the process can substantially reduce capital and operating costs of coal-to-jet fuel, making it competitive at today’s crude-oil prices.
- Demonstrate that the process can achieve substantial reduction in greenhouse gas (GHG) emissions without using carbon capture and storage (CCS).
The project objectives were accomplished over a 3-year, 5-task R&D effort to advance the hybrid, direct CTL process for jet fuel to a TRL of 5. The three major Subsystems of the process – biomass to bio-solvent conversion, coal dissolution and demineralization to produce a syncrude, and hydrotreatment/ hydrogenation of the syncrude to jet fuel – were developed and tested in batch/lab-scale, bench-scale, and then at pre-pilot scale. The project objectives were achieved over two phases. In Phase 1, all major process steps were explored and refined at continuous bench-scale, including: (1) biomass conversion to high hydrogen-donor bio-solvent; (2) coal dissolution in biomass-derived bio-solvent, without requiring molecular H2, to produce a syncrude; and (3) two-stage catalytic hydrotreating/hydrogenation of syncrude to jet fuel and other distillates. In Phase 2, these same process steps were scaled-up to continuous, pre-pilot scale, allowing realistic estimates of process economics and GHG emissions reduction, thus defining the path for widespread process commercialization in a short time period. The process meets the requirements of Section 526 of Energy Independence and Security Act of 2007 (EISA 2007) without requiring CCS, and it should help reduce the dependence on imported petroleum crude for jet fuel production. More information on each task is provided below.
Phase 1
Task 1 - Lab/Bench-Scale Coal Liquefaction and Syncrude Hydrotreating/Hydrogenation. Several combinations of coals and bio-solvents were tested at laboratory- and bench-scale to determine preferred operating conditions for the scale-up of the coal liquefaction Subsystem to 1 ton per day (1 TPD) pre-pilot scale in Task 2. Additionally, a two-stage catalytic system was tested for upgrading the syncrude to jet fuel. Several catalysts were screened at laboratory-scale to determine the preferred conditions for scale-up to one barrel per day (BPD) pre-pilot scale in Task 3. This task was supported by four Battelle subcontractors: Pennsylvania State University (PSU), University of Dayton Research Institute (UDRI), Advanced Research Associates (ARA), and Quantex.
Task 2 – Pre-Pilot-Scale Coal Liquefaction and Syncrude Production. A 1 TPD coal liquefaction pre-pilot plant was tested at the Quantex facility in Morgantown, WV. Several hundred gallons of syncrude were produced for upgrading to jet fuel in Task
Task 3 – Pre-Pilot-Scale Syncrude Hydrotreating /Hydrogenation to Jet Fuel. The syncrude from Task 2 was upgraded to a distillate product, which was fractionated into jet fuel and diesel fractions at the Intertek facility near Pittsburgh, PA, employing catalysts and operating conditions determined in Task 1. The jet fuel was analyzed against the commercial Jet-A specifications. Some “Fuel-fit-for-use” testing was also performed, with the testing conducted by UDRI. A diesel product was also characterized.
Task 4 – Conceptual Plant Design and Process Economics. Battelle completed a comprehensive conceptual plant design and utilized the design for a techno-economic analysis (TEA), following DOE/NETL guidelines.
Task 5 – Greenhouse Gas (GHG) Emission Analysis. A GHG life-cycle emissions analysis was performed, using DOE/NETL and USAF guidelines, to demonstrate progress towards meeting requirement of Section 526 of EISA 2007.
This is a Final Report on the 2-phase project.
4.0 DIRECT CTL BACKGROUND AND DESCRIPTION OF THE BATTELLE CTL PROCESS
4.1 Direct CTL Background
The U.S. DOE-supported direct CTL programs in the 1970s and 1980s included solvent-refined coal (SRC), H-Coal, Exxon Donor Solvent (EDS), and other two-stage liquefaction processes. The DOE’s focus was on direct CTL since it had a significantly higher thermal efficiency than indirect CTL using the coal gasification plus Fischer-Tropsch (FT) route. Most R&D and demonstration efforts were stopped in the late 1980s due to escalating cost estimates for CTL and very low petroleum crude prices. During the last decade, CTL has received renewed interest, though the focus of recent R&D has been indirect CTL. In many respects, direct CTL is more appealing than indirect CTL relative to the priority objectives of this project. First, it is possible with direct CTL, by carefully dissolving, depolymerizing, and hydrotreating coal, to produce acceptable jet fuel without blending in petroleum-based jet fuel. Indirect CTL, e.g., via FT route, converts a highly aromatic coal structure to a linear, paraffinic structure, which is unacceptable as JP-8 or Jet-A, as those fuels require a minimum of 8% aromatics, and hence must be blended with petroleum-based jet fuel, per Military (Mil-DTL-83133H w/Amendment 1–Tables A-II & B-II) and Commercial (ASTM D7566-12A–Table 1, Part 2) specifications. Second, a carefully controlled direct CTL process, such as the one Battelle is developing, is thermodynamically more efficient in terms of yield than the indirect CTL approach of converting coal to synthesis gas and then recombining to make condensable liquids. The higher expected thermal efficiency of direct CTL drives a reduction in greenhouse gas (GHG) emissions per gallon of jet fuel produced.
4.2 Battelle’s Novel Direct CTL Process
Battelle has developed a hybrid CTL process that uses a significant amount of biomass for direct CTL in its innovative approach. The hybrid CTL process includes three basic steps: (a) biomass conversion to a high hydrogen-donor (H-donor) bio-solvent; (b) coal dissolution in novel biomass-derived solvent without molecular H2; and (c) syncrude hydrotreating/hydrogenation to jet fuel.
State-of-the-art direct CTL processes first quickly depolymerizes/dissolves the coal, typically in a coal-derived solvent, and then these slowly hydrotreat the solution to break up large molecules, remove heteroatoms (e.g., S, N, O), and increase the H/C atomic ratio. The resulting syncrude can be further refined by hydrotreating to various distillate fuels, a portion of which may be jet fuel. Battelle has investigated ways to overcome several disadvantages of current direct CTL. First, the GHG emissions for current CTL jet fuel are about twice that for petroleum jet fuel, so it would be necessary to capture 90% CO2 at the CTL plant to meet the CTL GHG emissions reduction goal. Second, a straightforward process for dissolving coal and biomass in a solvent has not been practical, partly because of the high moisture content of biomass. Third, the H/C atomic ratio in a typical bituminous coal is about 0.80, while it is about 1.90 for jet fuel, so a large amount of H2 must be added to coal, which contributes to high GHG emissions, and drives up capital and operating costs for current direct CTL processes. In particular, with processes that both dissolve coal and hydrotreat in one step, the high temperature (~450°C) and pressure (~2500 psig) requirements make the reactor costs uneconomically high, cause equipment erosion, and utilize H2 inefficiently due to excessive production of lighter hydrocarbons. Finally, the yield of jet fuel vs. diesel is significantly less in these current direct CTL processes than for FT-based jet fuel production.
To address these concerns, Battelle has developed a hybrid CTL process that uses a significant amount of biomass for direct CTL in an innovative approach, as shown in Figure 1.
Figure 1. Battelle’s novel, direct CTL process proposed by Battelle.
As currently conceived, the Battelle Direct CTL process has three major Subsystems including:
(1) biomass conversion to high hydrogen-donor solvent; (2) coal dissolution in biomass-derived solvent without molecular H2; and (3) 2-stage catalytic hydrotreating/hydrogenation to jet fuel and other distillates. In Subsystem 1, biomass, derived primarily from non-food sources, is converted to a bio-solvent in processing plants that are economical at smaller scale (≤100 TPD). The resulting bio-solvents, with a H/C atomic ratio above about 1.40, are delivered to a larger coal dissolution/demineralization facility (≥1,000 TPD) in Subsystem 2. Based on data on solvent refining of coal published by Longanbach and Chauhan of Battelle in late 1970s, and more recently confirmed by West Virginia University as well as by Battelle project team member Quantex, coal can be dissolved quickly (in ˂10 minutes) at mild conditions (~400°C and 500-800 psig) with addition of only 0.3–0.5% hydrogen, by weight of coal, which increases the H/C molar ratio from about 0.80 to 0.86 [1-4]. The bio-solvents can be engineered to alter the nature and quantities of the cyclic/aromatic and linear species with desired hydrogen-donor properties.
Based on the low hydrogen-addition requirements to dissolve coal, as little as 10% of bio-solvents based on weight of coal is sufficient without requiring any gaseous H2, and thus minimizing CO2 emissions at the CTL plant (Subsystem 2). However, larger amounts of bio-solvent input will not only help meet the GHG reduction goal without CO2 capture, but also reduce the viscosity of the synthetic crude (syncrude), improving separation of ash and unconverted coal. Further, the larger quantity of bio-solvent helps produce a jet fuel that has a more manageable balance of aromatic and non-aromatic species, thus overcoming a significant limitation of current direct CTL. The solvent/coal weight ratio is expected to be about 2.5, so a portion of the solvent, including some coal-derived liquids, will need to be recycled after solid/liquid separation.
In Subsystem 3, the syncrude from the liquefaction plant can be catalytically hydrotreated/hydrogenated at either a petroleum refinery or a facility dedicated to maximizing jet fuel production. This is the location where all the molecular H2 will be added. The overall, gaseous H2 requirements for the Battelle CTL process are expected to be substantially less than for the H-Coal or EDS processes. We also expect the yield of jet fuel to be much higher than from H-Coal or EDS processes.
Battelle’s approach is to use novel, biomass-derived solvents with high hydrogen-donor capacity. These bio-solvents can be engineered to have significant amounts of cyclic/aromatic compounds (>20%) and a controllable H/C ratio, with good hydrogen-donor capabilities. The objective was to achieve as good hydrogen donor performance as with the well-known solvent tetralin, used in the EDS process as well as tested for direct CTL for the last 75+ years, but with solvents that are biomass-derived and cheaper than tetralin.
4.3 Potential Benefits of the Battelle Direct CTL Process
Battelle’s hybrid CTL process offers these specific advantages: (a) straightforward system integration of proven process steps; (b) improved process reliability due to mild liquefaction operating conditions (less than 800 vs. 2500 psig); (c) elimination of CCS at coal liquefaction site as well at the syncrude refining site; (d) significant reduction in molecular H2 requirement for syncrude refining; (e) increased aromatic content and density of jet fuel close to that of JP-8; (f) short time period to commercialization; (g) significant reduction in the capital and operating costs; and (h) substantial reduction in GHG emissions.
The Battelle CTL technology helps achieve the GHG emissions reduction goal of this project, unlike state-of-the-art CTL technologies, without requiring CCS. The key reasons for this are as follows:
- A major portion of the coal is replaced by biomass, which will significantly reduce the GHG footprint of this hybrid CTL process.
- The novel biomass-based solvents (bio-solvent) are high in H/C ratio compared to coal. The resulting calculated syncrude H/C ratio, at commercial-scale where some solvent is recycled, is ~1.20, compared to ~0.80 for coal and ~1.60 for crude oil. The single-pass syncrude H/C ratio using no recycle is typically ~1.00. Thus, the H2 requirements for hydrogenating syncrude to jet fuel (H/C ~1.90) are lowered by as much as 40%, which is a key determinant of GHG emissions during upgrading of syncrude to jet fuel.
- Unlike indirect coal-biomass to liquid (CBTL), where the coal and biomass contributions to jet fuel are only additive or proportionate, Battelle’s hybrid, Direct CTL process brings considerable synergy since the bio-solvent carries a significant hydrogen-donor capability. As a result, up to 90% of the organic fraction of coal can be dissolved without requiring molecular H2. This means that the GHG emissions in Subsystems 1 and 2 will be less than the GHG credit applicable to the biomass-derived content of the syncrude.
- The ability to operate the coal liquefaction step at mild conditions (~400°C, ~500-800 psig, no gaseous H2, no catalyst) allows for lower plant-energy requirements leading to further GHG emissions reduction.
- The absence of the vast majority of the mineral matter in syncrude increases catalyst life of the first stage of two-stage hydrotreatment/hydrogenation, which reduces catalyst regeneration requirements, as well as minimizes the wastage of H2 in producing the lighter, non-jet-fuel fractions.
- The biomass is converted to bio-solvent in small, distributed plants (~100 TPD) near the sources of biomass so the energy and cost required for biomass transport is greatly reduced. Additionally, bio-solvent is easily pumpable compared to cellulosic and other plant biomass. Similarly, smaller coal liquefaction plants (1,000-2,000 TPD) are economical due to use of non-catalytic, mild conditions, so coal transportation energy and cost is reduced as well.
5.0 LAB/BENCH-SCALE COAL LIQUEFACTION
The objectives of this effort were as follows:
- Select various feedstocks for coal liquefaction for exploratory testing
- Demonstrate that at least 80% of coal could be dissolved in biomass-derived solvents
- Perform parametric testing to help identify the preferred operating conditions for scale-up of Subsystems 1 and 2.
The Battelle CTL process combines coal, coal-derived recycle solvent, and an additional bio-solvent, and then heats the agitated mixture at elevated temperatures and pressures. After the mixture is cooled, mineral matter and undissolved organic matter in coal are removed in a solid-liquid separation step. Finally, the de-ashed, liquefied coal is fractionated in a distillation column to separate the liquefied coal or synthetic crude (syncrude) from the recycle solvent and heavy oil. The heavy oil may be sold as a by-product, or coked to recover more liquefied coal and the coke sold. If desired, the heavy oil may also be hydrocracked with the rest of the syncrude. Other potential, high-value, products from the heavy oil are: (a) binder pitch; (b) anode-grade coke; (c) needle coke; or (d) polyols or other chemicals. The centrifuge cake may be sold as an asphalt additive, burned to generate heat or power, or gasified to generate syngas.
Presented below is a brief discussion of the selection of the three major feedstocks.
5.1.1 Coal
Three coals were selected: (1) a West Virginia (WV) high volatile A, bituminous coking coal1 from Leer Mine; (2) an Ohio high volatile A, bituminous, coal from Waterloo Coal Company (sample obtained from Bramhi Coal Company), and (3) a Powder River Basin sub-bituminous coal from the Black Thunder Mine. One ton each of the first two coals were acquired and a portion ground to -25 mesh size at Quantex. A drum of the Black Thunder subbituminous coal also was obtained and ground to -25 mesh size at Quantex.
WV Bituminous Coal. During the previous West Virginia University (WVU) coal-liquefaction program, they successfully processed a Lower Kittanning seam, high-volatile A, coking, bituminous coal. Unfortunately, the coal mine in West Virginia where this coal was obtained from has now closed. In reviewing options, Quantex identified a coal from a Lower Kittanning seam about 30 miles from the original site. This coal is from the Leer Mining Complex, located in the town of Grafton in Jackson County, WV; a map showing the location of the mine is presented in Figure 2. The mine is owned by Arch Coal Inc. One ton of this coal was acquired. Four 5-gallon pails of the coal were initially ground by Quantex to a size smaller than 25 mesh (designated -25 mesh) to support chemical analysis and small scale liquefaction tests. More of this coal was ground to -25 mesh to support the Quantex 1 ton per day (TPD) unit testing.
1 A high volatile A bituminous coal has a fixed carbon content, on a moisture and ash (MAF) basis, of less than 69 wt%, and volatile matter content, on an MAF basis, of greater than 31 wt%, and a higher heating value (HHV) equal to greater than 14,000 Btu/lb on an MAF basis.
Figure 2. Location of Leer Mine
Ohio Bituminous Coal. There are three major seams that underlie southeastern OH, western PA and northern OH; see Figure 3. They include the Pittsburgh, Upper Freeport, and Lower Kittanning seams. The Lower Kittanning seam is the largest and is the same seam as was found successful in the WVU liquefaction tests that employed hydrogenated soybean oil [3, 4].
Figure 3. Location of the Lower Kittanning coal seam in PA, OH, and WV
Therefore, the team looked for a Lower Kittanning seam, high-volatile A, coking, bituminous coal in OH and identified existing mines with coal preparation plant in the area. Coal from the Waterloo Coal Plant, was identified. The location of the preparation plant is shown in Figure 4. Leer Coal does not sell directly to the public, but a vendor (Bramhi Coal Company) was identified. Four pails of the Ohio coal were initially ground by Quantex to -25 mesh for analysis and testing. More of this coal was ground to -25 mesh to support the Quantex 1 TPD liquefaction unit testing.
Figure 4. Location of the Waterloo Coal Preparation Plant
Wyoming Subbituminous Coal. WVU also successfully treated lower rank, low-ash coals in their liquefaction process. One of the subbituminous coals processed was Black Thunder coal, surface mined in Wright, WY. The location of the mine is shown in Figure 5 (this figure also shows the mine is located in the Powder River Basin covering southeast Montana and northeast Wyoming).
Figure 5. Location of the Black Thunder Coal Company
Fortunately, this coal is still commercially mined by The Black Thunder Coal Company (owned by Arch Coal, Inc.). A 55-gallon drum of a fresh sample of this coal was obtained.
The Leer (WV) and Waterloo (OH) coals were analyzed for ultimate and proximate analysis, sulfur forms, and Free Swelling Index. The Geisler plasticity was also determined to provide a measure of how easily the coal softens and flows upon heating. Results are shown in Table 1. The analyses are reported on a dry basis, except for moisture content and hydrogen, for which the as received (AR) value is also reported. Typical Black Thunder analyses (obtained from Arch Coal, Inc.) are also included in this table.
Table 1. Coal Analyses
Coal parameters expressed on a dry or MAF basis are often used to characterize coal. The coal values on these bases are noted in Table 2.
Table 2. Coal Analyses on a Moisture- and Ash-Free (MAF) Basis
5.1.2 Coal-Derived Recycle Solvent
Coal tar distillate (CTD) is used to start the liquefaction process. After a period of time sufficient coal liquids will be generated in the process so they can displace the CTD and very little (≤ 10% based on wt of coal) additional CTD will be required. For the initial liquefaction tests at Battelle, PSU, and Quantex, a Koppers CTD was used. The typical Koppers CTD, used to make carbon black, has a boiling range of 230ºC to 360ºC. However, the Koppers CTD supplied to Quantex a couple of years ago has about 50% boiling above 400ºC, as indicated by the simulated distillation (SimDis) test conducted by PSU (see Figure 6). These two start-up solvents were compared so we could select the correct one for further testing. In any case, this CTD is expected to contain significant quantities of cresol, naphthalene, naphthol, and anthracene.
Figure 6. Simulated distillation data for Koppers CTD supplied by Quantex
5.1.3 Bio-solvents
In the initial work at West Virginia University (WVU), a combination of coal, CTD (recycle solvent), and hydrogenated soybean oil was used to liquefy the coal [3, 4]. It is believed by Quantex that the soybean oil facilitated coal depolymerization and the resulting coal dissolution. On the other hand, in Battelle’s direct coal liquefaction process, we use a bio-solvent that helps dissolve coal as well as serves as a hydrogen donor to species generated during coal depolymerization/ dissolution. In this fashion, we expect to increase the H/C atomic ratio of coal-derived syncrude without using any molecular hydrogen (H2). One example of a well-known hydrogen donor-solvent is tetralin, which is expensive and typically not bio-based.
A major effort on this project was the selection, procurement, modification, and liquefaction screening of a number of bio-solvents. Over 40 bio-solvents were prepared and screened. The majority of these bio-solvents were prepared utilizing non-edible biomass, employing proprietary treatment methods. Originally, about 6 bio-solvents were targeted as the testing began with the West Virginia coal. Next, the Ohio coal was tested, during which time many more bio-solvents were identified or became available. As such, most of the bio-solvents were tested on Ohio coal. A very limited amount of testing, with a preferred bio-solvent, was carried out on the Wyoming coal. Some of the bio-solvents were a mixture of two bio-solvents to achieve better hydrogen transfer as well as to facilitate coal depolymerization.
5.2 Laboratory-Scale Coal Liquefaction Testing
5.2.1 Test Objectives
The goal for this effort was to demonstrate the feasibility of Battelle’s hybrid, direct CTL process for producing a syncrude using novel bio-solvents without using any molecular H2. The primary liquefaction testing was performed in Battelle’s 0.5 L high temperature, high pressure autoclave in order to achieve a coal solubilization of 80% or greater. Microreactor liquefaction tests were also run at PSU in parallel to Battelle’s work to better understand the kinetics of the liquefaction process.
5.2.2 Lab-scale Coal Liquefaction Test Procedures.
The test procedure for the autoclave tests is shown schematically in Figure 7.
Figure 7. Typical coal-liquefaction processing conditions for 0.5 L autoclave system.
Shown in Figure 8, below, is a photo of the autoclave as well as the 0.5-L cup (i.e., the base of the autoclave).
Figure 8. Photograph of 0.5 L autoclave system at Battelle.
After a run, water was passed through a Hastelloy-C coil located within the reactor. This cooled the liquids and allowed us to reduce the pressure to atmospheric level. The retaining bolts were unscrewed and the autoclave head that held the mixing assembly and cooling coil was removed. A low-boiling inert liquid (tetrahydrofuran, THF) was sprayed onto the head, mixing shaft and propeller, and the coil to recover all remnants. Additional THF was added to the cup (the more viscous the product, the more solvent was used).
The contents of the cup were also transferred to a 1 L Erlenmeyer flask. The solution was heated to boiling while being stirred to separate the liquid fraction from the solid particles. This step also speeded up filtration. Once boiling was achieved, it was poured into the top of a pressure-filter. In some cases the Syncrude produced after liquefaction was low viscosity, water-like liquid and was easily transferred. In other cases the product was a thicker, more viscous fluid. In these cases, THF alone could not reduce viscosity enough for effective filtration. For this reason, hot (150°C) dimethylformamide (DMF) was utilized. The recovered liquid and undissolved coal and mineral matter were pressure filtered at up to 40 psig (easy to filter products required less pressure to filter). More inert fluid, THF or DMF, was used to try to recover as much coal material as possible from the filter. The filter cake was washed again with THF and the cake placed in a vacuum oven to remove the wash solvent. The filtrate was placed in an evaporating flask and put in a specialized vacuum heated apparatus called a Rotovap operated at 50°C and 20 amHg (20 Torr) for 1 hour. If DMF was used, a specialized Rotovap, called a Kugelrohr, was used at 80°C and ~1 mm Hg pressure. The solvent-free filtrate and filter cake were then weighed.
Provided below is an example run. The following was weighed out into the autoclave cup:
- 88.39 g of as-received Ohio coal (with 8.20 g, 9.28 wt% moisture and 6.80 g, 7.69 wt% mineral matter) ground to smaller than 20 mesh (designated -20 mesh)
- 15.1 g of bio-solvent (i.e., 17 lb/100 lb coal)
- 210.65 g of coal tar distillate (CTD).
- Weigh components into autoclave cup, place autoclave cup onto reactor assembly. Inter head with stirred and cooling coil. Tighten vessel retainer and purge with N2 to remove air.
- Begin stirring at 360 RPM and heat to 400ºC.
- Once temperature is reached (approximately 47 minutes), maintain temperature for 30 minutes.
- Once 30 minutes has passed, cool reactor contents to 80ºC by passing water through the internal cooling coil and vent off gas.
- Lift autoclave head and rinse with solvent into separate beaker.
- Combine all material (autoclave cup and rinse fluid) into an Erlenmeyer flask.
- Heat to boiling with manual stirring and then pour contents into pressure filter containing a10-micron nylon membrane filter; close lid and apply up to 40-psig N2 pressure; capture filtrate into jar. (With a good bio-solvent, this pressure may be less than 5 psig, indicating low viscosity and ease of filtering).
- Rinse filter cake with boiling THF solvent into round bottom flask.
- Evaporate off wash solvent using a Rotovap.
- Place filter cake into vacuum oven at 60°C and 20-mm Hg vacuum overnight to remove the wash solvent.
Solubility of coal is defined as 100 minus (the “quantity of filter cake minus coal ash fed” divided by the quantity of moisture, ash free (MAF) coal fed) times 100, or
The filter cake was ashed at 500°C for 240 minutes. The ash content was determined to be 36.9 wt%. We compared this measured ash value to a theoretical solubility versus ash content curve, see Table 3 and Figure 9. The mass balance for this example was 96.4%.
Table 3. Solubility versus Predicted Filter Cake Ash Content
Figure 9. Theoretical solubility versus product (syncrude) ash content.
This relationship was calculated by assuming (1) the mineral matter in the coal equals its ash content, (2) 100 % of the ash in the coal ends up in the filter cake, (3) all of the non-solubilized coal ends up in the filter cake, and (4) no residual coal liquids are retained with the filter cake. So, for 0 wt% solubility, all of the coal would end up in the filter cake and the predicted ash content would be the moisture-free ash content of the coal (8.48 wt%). At 100% solubility, only the ash would be retained in the filter cake, and the ash content of the filter cake would be 100 wt%.
For the measured 36.9 wt% ash content, the predicted solubility was between 84% and 85 %, which is in reasonably good agreement with the calculated value of 87.12 %.
5.2.3 Results for West Virginia (WV) Coal
The reactions were performed following the autoclave procedure described above. The coal was minus 25 mesh with no further treatment. We initially began running reactions at 400ºC for 30 minutes with WV coal and Koppers CTD but found that the resulting syncrude was very viscous and only one candidate bio-solvent performed reasonably well. Because of the higher viscosity of filtrate, we decided that a higher temperature was needed to help depolymerize the coal to a greater degree. We next ran at a temperature of 425ºC. Tetralin performed well at this temperature yielding 86.8% dissolution of coal. However, after running the best bio-solvent at that point (BS-3) at times of 30 min, 60 min, and 90 min, we found that we were getting some coking. This was supported by previous observations at Quantex indicating they had seen coking at temperatures greater than 420ºC. This led us to testing at 410ºC. With Leer coal and Koppers CTD, the best bio-solvent yielded a dissolution value of 86.7%, which is same as for tetralin. At 415ºC the same material gave a dissolution of 85.5%; essentially the same value as at 410ºC. Under these same conditions Tetralin was run twice and gave dissolution values of 84.3% and 86.6% at 415ºC and 410ºC, respectively. From this point onwards, we selected 410ºC as the standard operating temperature for feedstock screening. The liquefaction yields for the several bio-solvents, using WV coal, are compared with soybean oil and tetralin in Table 4. The yields for other bio-solvents were below 50%.
Table 4. Batch Autoclave Coal Liquefaction Screening Results on WV Coal
The data on WV coal showed that the percent solubility of coal, as defined by solubility in hot THF, as well as the viscosity and filtration behavior of the product are highly dependent on the solvent used. For example, the product based on use of tetralin was of a lower viscosity and the product work-up, including dissolution in hot THF and filtration were easy. At the same time, the solubility with tetralin was quite high at 86.6%. Soybean oil base case gave a relatively good dissolution yield at 76.5%, but a stronger solvent (DMF) had to be used and the product was very viscous. After re-extracting the filtrate with hot THF, we found that 23.0% of the filtrate was THF insoluble. Thus, the corrected THF solubility with soybean oil was only 58.9%. On the other hand, three bio-solvents, BS-3, BS-9, and BS-15, provided as high dissolution yields as with tetralin and the products were of lower viscosity and relatively easy to dissolve in hot THF. This result was expected since soybean oil is not an H-donor.
Based on the initial results from 31 autoclave tests, BS-3 (Autoclave Test No. WV-13) and BS-15 (Autoclave Test No. WV-34) appeared to be the best bio-solvents. One of these tests, WV-13, was selected for further analysis of the filter residue and syncrude, which included a portion that will normally be returned to the front-end of the process as the CTD recycle solvent. The ultimate analysis of the filter residue and syncrude, are compared with that of coal and CTD solvent in Table 5.
Table 5. Coal, CTD Solvent, Syncrude, and Filter Residue Analyses for Autoclave Test No. WV-13
Table 6. Estimated Composition of Syncrude, Free of CTD Solvent, Compared to Coal, on Moisture-Free (MF) Basis
5.2.4 Results for Ohio Coal.
After completing the tests with WV coal, we began testing the Ohio coal. A total of 151 batch tests were carried out. The primary reason for more testing on Ohio coal was that many more bio-solvents became available during Ohio coal testing.
The preferred bio-solvents were compared with tetralin and soybean oil, as shown in Table 7.
Table 7. Batch Autoclave Coal Liquefaction Results on Ohio Coal
As shown in Table 7, seven (7) bio-solvents performed as well as or better than tetralin. A total of 11 bio-solvents gave over 80% coal solubility, compared with only 66.9% for soybean oil, which served as the base case. The BS-19A and BS-32 bio-solvents were actually used for larger-scale testing at Quantex. The liquid product from the BS-19A test, called Syncrude 2, was evaluated for lab-scale hydrotreating/hydrogenation. Based on cost considerations, two bio-solvents, BS-27B and BS-32, were down-selected for scale-up in Task 1.03.
5.2.5 Results for Wyoming Coal.
Black Thunder coal was found to be 6% ash and 22.2% water. We ran four tests through our dissolution process. After correcting for moisture content, we ended up using a liquid to coal ratios of 2.5-2.8 and ran at temperatures of 400°C or 410°C for 30 minutes. The bio-solvents used were soybean oil, BS-19A, BS-27, and BS-32. The soybean oil run could not be worked up due to a very thick tar product plugging the filter. We tried using hot THF and hot DMF but the material would still not filter. The reaction with BS-19A went well and produced a low viscosity oil. But, even with THF this material would not filter due to the solid material being more like a tar/thick oil than a solid as seen with other coals. It is believed this reaction would do well with a centrifuge, as used in continuous tests at Quantex. The solid gummed so well that the liquid solvent mixture was decanted. The solid portion was rinsed with hot solvent and continued to show a tar-like consistency. This test with BS-19A calculated to 91.9% solubilized coal. The filtrate was of very low viscosity and this would be expected due to the fact that we did not remove the water from the coal, making our real liquid to coal ratio equal to 2.6 instead of 2.0.
5.2.6 Parametric Testing Results
Effect of Time and Temperature. During early studies on WV coal, temperature was varied from 400°C to 425°C. We began our work at 400°C, but found that we could only obtain good data for one candidate, the BS-3 solvent. The soybean oil and other candidates produced too thick of tars even without removal of the extraction solvents. Because of these issues, we then switched and ran at 425°C. This temperature gave lower viscosity oils, but also gave lower yields. When time was increased from 30 to 90 minutes on the BS-3, the solubilization dropped by 40.0 percent. This meant that we were coking our mixtures at 425°C. This was supported by WVU work which saw coking above 420°C. On the other hand, at 410°C, the same BS-3 solvent solubilized 86.7% of coal. When this temperature was increased to 415°C the solubilization was 85.5%. Therefore, at that time, we settled on a reaction temperature of 410°C. Later on, while working with the Ohio coal, we decided to evaluate lower temperatures due to the belief that no coking would take place at 400°C and below. Based on PSU work (see Section 5.2.7), we also knew that solubility was better at lower temperatures, but typically at longer times. Due to lack of the normal coal tar distillate at the time, we had to use a light coal tar distillate which was known to perform less favorably, but was believed to be acceptable for providing the trends relative to time and temperature. The results are in Table 8 below.
Table 8. Effect of Reaction Time and Temperature for WV Coal Using BS-27 at 29% BS/MF Coal and Lighter CTD
As seen above, there was little difference in the solubility between 390°C at 30 minutes and 400°C at 10 and 30 minutes. However, lower solubility was observed at 390°C and 10 minutes. This suggested that time is more critical when working at lower temperatures. Once we had obtained more of the desired coal tar distillate (CTD), we repeated two tests shown above. The results are shown in Table 9 below.
Table 9. Effect of Temperature on Solubility for Ohio Coal Using Heavier CTD
As expected, the solubilities at 400°C for 30 minutes were quite good with the new, heavier CTD. The solubility was good even at 390°C when time was extended to 45 minutes. However, the reaction is likely not complete as the viscosity was still high. If lower temperature (e.g., 390°C) is desired, a time of 60 minutes or greater is likely needed.
Effect of Bio-solvent. The various bio-solvents have varying degrees of performance based on their potential to dehydrogenate, and thus serve as an H-donor. As seen in Table 7, BS-15 and BS-19 seem to have the greater potential to reduce the final viscosity of the syncrude. Also, based on the molecular structures, BS-9, BS-15, BS-26, BS -27, and BS-32 have the highest potential for dehydrogenation, so it was expected that they would perform well, and which they did. Another surprising matter is that the physical state of the bio-solvent before the reaction does not necessarily effect the final syncrude state. For example, BS-24, BS-27, and BS-19 are solids at room temperature. They were blended with soybean oil in order to maintain them as liquids. However, the non-blended samples produced lower viscosities than after blending with soybean oil. This further confirms that soybean oil, which was tested by WVU and therefore served as a base case, doesn’t transfer hydrogen to depolymerized coal unlike BS-19, BS-24, and BS-27. Some selected comparisons of the novel bio-solvents with soybean oil relative to coal solubility and viscosity of syncrude are shown in Table 10.
Table 10. Effect of Bio-Solvent Type on Coal Solubility and Syncrude Viscosity for Ohio Coal.Solvent Recycle. We evaluated replacing Koppers CTD with our own syncrude five different times. When we took our syncrude from Run OH-43 and replaced 100% Koppers CTD in run with BS-19A, the coal solubility dropped from 84.5% to 51.0%. Similarly, when we took syncrude from Run OH-60 with BS-24 and replaced 50% Koppers CTD in run with BS-24, the coal solubility dropped from 76.6% to 48.3%. However, when we took syncrude heavies from Quantex Run #3 with BS-19A and replaced 50% Koppers CTD in run with BS-19A, the coal solubility increased from 82.0% to 89.7%. Similarly, when we took syncrude from Quantex Run #3 with BS-19A and replaced 100% Koppers CTD in run with BS-19A, the coal solubility increased from 82.0% to 84.7%. The reason the first two failed when compared to those run with Quantex Run #3 heavies is likely due to the low-boiling fractions being included in the first two. The Quantex Run #3 heavies is the material that passed through the wiped film evaporator (WFE) to remove everything boiling below ~290°C. This means that the material contains everything >290°C boiling point. It is known that heavier coal tar distillate typically solubilizes coal better than lighter fractions. By removing the lighter fractions, we made a better solvent out of our syncrude. Replacement of 10% of the BS-24 with hydrotreated syncrude from Quantex Run #3 improved the coal solubilization from 76.6% to 80.1%, which is due to production of tetralin type of molecules during hydrotreatment. Finally, in Run OH-140, we replaced 50% of the Coppers CTD with centrate from Quantex Run #6A and used Stage-1 product from syncrude hydrotreatment as a H-donor solvent. This reaction resulted in 85.2% coal solubility, indicating that the bio-solvent can be regenerated via syncrude hydrotreatment.
5.2.7 Microcatalytic Reactor Testing at PSU
Test Equipment and Procedure. A series of coal dissolution experiments were performed in vertical tubing reactors (microreactors). Having an internal volume of approximately 25 milliliters (mL), these microreactors are fitted with a pressure gauge, adjustable pressure relief valve and isolation valve. A photograph of a microreactor is shown in Figure 10.
A blend of coal, bio-oil, and coal-derived solvent, having a total weight of approximately 7.50 grams, was loaded into the reactor’s body for each test. The reactor was then connected to a gas manifold, pressurized with nitrogen, and tested for leaks. The reactor was vented and again pressurized. This step was repeated two times to purge all air from the reactor. Finally, the reactor was pressurized to the reaction pressure and the pressure relief valve adjusted as necessary. The reactor was then plunged into a preheated sand bath held at the desired reaction temperature. The bath temperature was measured by a thermocouple placed in close proximity to the reactors. Each reactor and its contents were agitated at approximately 200 cycles per minute to promote mixing between the coal, bio-solvent, and coal-derived solvent throughout the test.
After the reaction time had been reached, the reactors were removed from the sand bath and quenched in cold water. Product work-up was similar to typical coal liquefaction product work-up [5]. Any gas products were vented and the solid or liquid by-products were flushed from the reactor with THF. The solution was sonicated for 20 minutes to promote dissolving of the coal dissolution products. THF insolubles were then separated from solution by filtering the THF through glass fiber filter paper having particle retention of 2.2 µm. The filter paper and solids were allowed to dry for 24 hours in a fume hood before being transferred into a glass fiber thimble. The thimble had particle retention of approximately 0.8 µm. Solvent extraction was performed for 24 hours using the THF collected during the previous filtration. The thimble and its contents were dried for 24 hours in a fume hood and then dried under vacuum for at least 1 hour at 110°C. The final weight of the THF insolubles was determined and used in Equation 1 to calculate the percent conversion of the coal on a dry, ash-free basis.
The THF solvent was removed from the THF solubles by rotary evaporation combined with final evaporation under vacuum for 1 hour at 70°C.
Figure 10. Microreactor used for coal dissolution experiments.
Coal Dissolution Results. A total of 38 coal dissolution experiments were performed in the microreactors. These experiments were intended as a screening tool to evaluate the effect of different bio-oil solvents, coal-derived solvents, chemical solvents, and operating conditions (reaction temperature and reaction time) in promoting coal conversion. The operating conditions and coal conversion (on a dry-ash free basis) for each microreactor experiment is provided in Table 11. A slight correction (0.1 to 0.2 weight %) for coal conversions reported in the previous quarterly report was made to adjust for coal analysis subsequently determined at PSU. Only the Leer (WV) coal was used in the microreactor testing. The Bramhi (Ohio) coal was used in the large lab reactor testing discussed in Section 5.3.
Table 11. Operating Conditions and Coal Conversions for the Microreactor Experiments on WV (Leer) Coal
Table 11. Operating Conditions and Coal Conversions for the Microreactor Experiments on WV (Leer) Coal (continued)
The first microreactor test was performed using soybean oil blended with QRS coal-derived solvent. Attempts to separate the solid and liquid products from this test using dichloromethane were not successful, as also observed in Battelle’s testing. Therefore, a decision was made to perform the next series of tests using BS-2 followed by recovery with THF. With the exception of PSU-4, the first five tests in this next series were performed using BS-2 blended with the QRS coal-derived solvent. These initial tests were used to develop the protocol for recovering the coal dissolution products from the microreactors and separating the unreacted coal and mineral matter from the liquid products. The highest level of coal conversion using BS-2, 50.5 weight % MAF), was achieved at 385°C with a reaction time of 60 minutes.
Tetralin, a hydroaromatic solvent capable of donating hydrogen to cap free radicals, was used in PSU-4 to compare with the coal conversions achieved with the bio-oils. This test resulted in the highest coal conversion (78.1 wt% MAF) achieved in the microreactor experiments.
Although the microreactors were agitated during each test, a concern was raised regarding the effect of reactor orientation on mixing and coal conversion. Previous coal dissolution studies have used both vertical (see Figure 10) and horizontal reactor bodies. To evaluate the effect of reactor orientation, an experiment (PSU-5) was performed in a horizontal reactor body. The coal conversion of PSU-5 (conversion – 48.2 wt% MAF was compared with PSU-3, the same test performed in a vertical reactor body (conversion – 50.5 wt% MAF). The difference between these tests was 2.3 wt% with the vertical reactor achieving the higher coal conversion. Because the vertical reactor orientation achieved the higher conversion, it was determined that this design did not limit internal mixing and coal conversion. Therefore, continued dissolution testing was performed in the vertical reactor design. Also, an analysis of the liquid products by GC SimDis confirmed that the reactor’s orientation did not significantly affect the liquid product’s boiling point distribution.
Testing performed at Battelle determined that the coal conversions achieved using BS-3 were greater than those with BS-2. Therefore, Battelle requested that all further testing be performed using BS-3. Several series of tests were then performed varying the reaction temperature, reaction time, and coal-derived solvents in combination with BS-3. Figure 11 shows a plot comparing the percent coal conversion versus temperature for different coal-derived/bio-oil solvent combinations. These tests were all performed at a reaction time of 30 minutes. The data indicate that the best temperature to run reactions using either coal-derived solvent (Koppers or QRS) with BS-3 is 400°C for a 30-minute reaction time. Lower coal conversions were achieved at 385°C, while no significant increase in coal conversion was gained by running at 415°C. While no advantage in coal conversion was gained by operating at 415°C, an analysis of the liquid product’s boiling point distribution also showed no benefit at the higher temperature.
Two series of experiments were also performed at a 30-minute reaction time replacing BS-3 with tetralin. The purpose of these tests was to evaluate the benefit of introducing a strong hydrogen donor solvent. Only a slight improvement in coal conversion was observed at 385°C. However, a greater increase in coal conversion was seen with tetralin at higher reaction temperatures (400°C and 415°C). This increased conversion can likely be attributed to the donation of additional hydrogen from the tetralin to free radicals, thus limiting coking of the coal and bio-oil mixture.
Figures 12 and 13 plot the percent coal conversion versus reaction temperature for each reaction time studied. Figure 12 compares tests performed using a solvent blend of Koppers and BS-3, while Figure 13 compares tests performed using a solvent blend of QRS and BS-3. These show that the coal-derived solvent from Quantex (QRS) performs better than the coal-derived solvent provided by Koppers. It also appears that a reaction time of 30 minutes is insufficient to achieve a maximum coal conversion at 385°C. While the shorter reaction time requires higher reaction temperatures, the coal conversion at longer reaction times (60 or 90 minutes) decreases at higher temperatures. This observation can possibly be attributed to repolymerization of the liquid products or thermal degrading of the bio-oil.
The best reaction condition appears to be when using tetralin with QRS (70.4 wt%), but the bio-solvent used thus far was getting decent results (~60-65 wt%). An additional bio-solvent (BS-19A) was tested in the microreactors at a solvent-to-coal ratio of 2:1 and higher. Although the coal conversions achieved with BS-19A were not as high as those observed with BS-3, the higher solvent-to-coal ratio did yield increased conversions. The higher solvent-to-coal ratio of 3:1 was also tested with tetralin (66.4 wt%). This test yielded similar increases in coal conversion to those using BS-19A.
The final series of tests performed in the microreactors was a determination of repeatability for this type of experiment. The repeatability was determined by performing an experiment in triplicate (PSU-13-1, PSU-13-2, and PSU-13-3) using the Koppers coal-derived solvent and BS-3 at a solvent-to-coal ratio of 2:1. The mean coal conversion for these three tests was 61.2 wt% with a repeatability of ± 0.6 wt%. This value can be used when comparing the coal conversion of a dissolution test with any other dissolution test.
Figure 11. Percent coal conversion as a function of temperature for different coal-derived + bio-solvent combinations (solvent-to-coal ratio: 2:1, reaction time: 30 minutes).
Figure 12. Reaction temperature versus reaction time for solvent blend of Koppers plus BS-3 reacted with Leer coal (solvent-to-coal ratio: 2:1).
Figure 13. Reaction temperature versus reaction time for solvent blend of QRS plus BS-3 reacted with Leer coal (solvent-to-coal ratio: 2:1).
5.3 Bench-Scale Coal Liquefaction Testing at PSU
5.3.1 Test Equipment and Procedures.
A bench-scale coal dissolution system was constructed by PSU for this task. The system, shown in Figures 14 and 15, contains three one-liter continuously stirred tank reactors (CSTR(s)) operated in parallel to increase the processing volume. These larger reactors permit greater quantities of coal dissolution liquids to be produced for further analysis or upgrading. Each CSTR was equipped with a magnetic stirrer that had two impellers mounted on the internal shaft. The mixer’s speed was continuously monitored by a tachometer. The reactors were housed within an electrical heater that provided the energy to heat the reactor and coal/coal tar distillate/bio-oil slurry up to the reaction temperature. The internal reaction temperature was monitored throughout the test by a type-K thermocouple inserted into a thermowell located inside the reactor.
A test was performed in the large reactors by individually filling one of the CSTRs with approximately 800 grams of coal/coal tar distillate/bio-oil slurry. This is equivalent to roughly 750 mL in volume. The slurries were manually prepared and loaded into each CSTR. The reactors were then closed, the headspace purged with nitrogen, and the reactor pressurized with additional nitrogen up to 600 psig. The reactor was heated to the reaction temperature and held there for the required residence time. After the residence time had been reached, the dissolution products were drained into a product/settling tank through heated transfer lines (see Figure 15). Light ends were allowed to pass from the product tank, through a heat exchanger, and were collected in a separate condensate tank. Air loaded backpressure regulators were initially used to maintain pressure within each CSTR prior to venting to the atmosphere. The backpressure regulators were subsequently replaced by manual metering valves for much of the large reactor testing. The filter, shown in Figure 16, allows pressure filtration of the liquid products to remove unreacted coal and mineral matter greater than 10 µm in size. Product from either shakedown runs or actual tests performed at a solvent-to-coal ratio of 2:1 were too viscous to be filter. Therefore, the dissolution products were collected from the product and condensate tanks prior to filtration. Typically, a 15-gram sample was collected as the dissolution products flowed from the product tank. This sample was then filtered and extracted following the same procedure used in processing the microreactor samples.
Figure 14. One-liter CSTRs used at PSU.
Figure 15. Product/settling and condensate tanks used at PSU.
Figure 16. Pressure filter for the bench-scale coal liquefaction system at PSU.
5.3.2 Bench-Scale Test Results
Several initial shakedown runs were performed in the CSTRs using the Ohio (Bramhi) coal, Koppers coal tar distillate, and BS-19A at the same weight percentages used in the microreactor testing. Performed at a reaction temperature of 410°C and a reaction time of 30 minutes, these tests produced a tarry product that had a consistency of a petroleum resid at room temperature. Attempts to reheat this material for filtering were not successful. We have learned that getting the material to flow in a uniform manner after allowing the reaction products to cool down does not work well, even when heating the material to 125-150°C. Three options for lowering the viscosity to permit filtering were discussed. They included: 1) increasing the solvent-to-coal ratio to 2.5:1, 2) mixing THF with the products in a 50:50 blend, 3) ship the product to Battelle for filtering. For time’s sake, a decision was made to send the product from the first set of runs to Battelle for processing.
The next series of runs in the large reactor system were conducted using different bio-oil solvents. These tests are listed in Table 12. Not until test PSU-LR5 was a sample of the dissolution products processed and the coal conversion determined at PSU. Battelle also analyzed a portion of the dissolution product and we determined a significantly greater coal conversion (75-85 wt%). The reason for this difference is not apparent, but remains a point of interest.
Table 12. Dissolution Tests Performed in CSTRs (Bench-Scale Liquefaction Reactors)
5.3.3 Reaction Differences Between Microreactors and Bench-Scale Reactor
The purpose of microreactor testing was to evaluate the relative effects of different bio-solvents and operating conditions, such as reaction temperature and reaction time, and coal solubility. There are several operating differences between the microreactors and the larger CSTRs. First, the microreactors are agitated at approximately 200 cycles per minute, while internal impellers mix the CSTRs. It’s assumed that the impellers provide better mixing of the slurry (and improved mixing may be better with higher solvent to coal ratios), but the importance of this difference hasn’t been investigated. The second difference lies in the heating rate. It’s predicted that the contents of the microreactors reach the reaction temperature in ~3 minutes after the reactor is plunged into the preheated sand bath. However, the heavy-walled CSTRs take approximately 75 minutes to reach the reaction temperature (400°C). These differences may make it difficult to directly compare results between reactor types. However, each type of reactor should provide a relative ranking for experiments performed in them.
Assuming that the extended time interval required for the CSTRs to reach the reaction temperature was allowing free radicals to repolymerize, yielding heavier products, an additional series of tests (PSU-LR7, PSU-LR8, and PSU-LR9) were performed using shorter reaction times of 0, 15, and 30 minutes. As shown in Table 3, these experiments indicate that a reaction time of 0 or 15 minutes is insufficient to maximize coal conversion. In fact, reaction times greater than 30 minutes may be required when performing reactions in the CSTRs.
5.3.4 Reaction Differences in PSU CSTRs and Battelle Batch Autoclave
There were also differences between CSTR reactor conditions at PSU and batch autoclave at Battelle. As discussed in Section 2.2, PSU tried reactions at lower residence times to see if PSU was reacting the materials took with the longer heat up times. However, PSU found that these experiments indicate that a reaction time of 0 or 15 minutes is insufficient to maximize coal conversion. In fact, reaction times greater than 30 minutes may be required when performing reactions in the CSTRs. Another issue is PSU operated reactors at lower pressures than Battelle; PSU used controllers to keep pressures at ~500 psig (as suggested by Battelle), while Battelle allowed the internal pressure to reach the maximum pressure, ~600-1200 psig. PSU also found that Battelle’s batch autoclave has a faster heat up rate and an internal cooling loop, so heating and cooling could take place relatively quickly. PSU could not increase the heating rate, but PSU was able to remove the heater from the reactor at the end of the residence time and drop liquids out quickly in order to mimic Battelle’s cooling profile. As discussed in the previous section, the differences may make it difficult to directly compare results between reactor types. However, each type of reactor should provide a relative ranking for experiments performed in them. Future work may require increased reaction pressures.
5.4 Down-Selection of Feedstocks and Operating Conditions for Continuous Coal Liquefaction Testing
The objective of this subtask was to document the rationale used in down-selection of the pre-pilot scale operating conditions for conversion of coal to jet fuel. The Program Plan outlines two series of pre-pilot runs. The first involves the preparation of Syncrude using the Quantex 1ton/day (1 TPD) coal liquefaction facility located in Morgantown, WV. The second run, to convert the syncrude to distillation fuels, will be conducted at the Intertek 1 BPD hydrotreating system. A discussion of the rationale for bio-solvent(s) and conditions used for the Quantex pre-pilot unit is provided below.
Battelle conducted over 100 batch liquefaction tests, PSU conducted over 30 microreactor semi-continuous tests, and Quantex conducted three continuous runs to gather the data needed to down-select the preferred set of operating conditions. As noted earlier, we set a minimum of 80% solubility as our acceptance criteria. Subsequently, we added viscosity, bio-solvent availability, bio-solvent usage and raw-material cost as additional discriminators.
5.4.1 Initial Liquefaction Testing
Based on our work in the 1970s and 1980s, along with input from Quantex, we selected the initial range of operating conditions as follows:
- Temperature: 390 to 420°C
- Pressure: 400 to 500 psig (in some cases higher pressures were observed)
- Liquid to coal weight ratio: Initially at 2.0, and then increased to ~2.5 to accommodate the Quantex 1 TPD unit
- Residence time at temperature: 10 to 30 minutes.
- Bio-solvent: Soybean oil (SBO) was our baseline solvent.
In our testing we found that two types of bio-solvents were useful. One that could provide hydrogen to the coal matrix, such as a hydrogen-donor solvent, and one that could promote depolymerization. In order to maximize the donor-solvent capabilities, it was found that an optional pretreatment could, in some cases, be used to optimize performance. The process is shown schematically in Figure 17.
Figure 17. Production of a biomass-derived solvent for coal liquefaction
We conducted a series of tests, covering the range of liquefaction operating parameters. Some parametric testing was also conducted on high-priority bio-solvents. After review of the solubility data, we noticed that a test might produce a similar solubility, but the liquefied coal could appear dramatically different. Some products were a thick sludge and some were low-viscosity syrup-like liquids. Our experience indicated that lower viscosity product was superior, so viscosity determined at 50°C was added to our down-select criteria.
We initially tested a promising bio-solvent along with a depolymerization solvent. If successful, we progressively reduced the content of the depolymerization solvent and even dropped the bio-solvent-to-coal ratio to find the lowest, but still effective, bio-solvent-to-coal ratio.
The final criterion was bio-solvent availability and cost. In some cases we could purchase the bio-solvent from an established company, or we could easily convert a commercially available chemical. Cost was considered, but not in a rigorous manner. Some of the very best bio-solvents were also the most expensive. But, some relatively expensive bio-solvents could be used at relatively low proportions and still achieve good performance.
5.4.2 Initial Testing at Quantex.
We initially conducted three full-scale exploratory tests at the Quantex facility; one with Leer coal and two with Ohio coal. In these tests we fed -20 mesh Ohio (supplied by the Bramhi Coal Company), operated at approximately 400°C, 400 psig, and maintained a liquid to coal ratio of 1.9 to 2.0.
The coal was screw fed into a mixing pot that was co-fed with CTD (from a 250 gallon, room temperature tote), and bio-solvent (from a room temperature drum). The mixture was mixed in a special pump and then transferred to a preheater where most of the moisture was driven off. Next, the slurry was sent to a pump to bring the pressure up to the desired operating pressure. The coal was liquefied in two digesters. Excess pressure, caused by the vaporization of trapped water in the coal or the production of light hydrocarbons from thermal devolatilization, was purged to maintain the desired pressure. After liquefaction, the slurry was cooled to 100°C and then centrifuged to remove the bulk of the unreacted coal and residual mineral matter. The solids stream (the centrifuge “cake”) was sent to a drum. The centrate (liquid fraction from the centrifuge) was fed to a wiped-film evaporator (WFE) which served as a single-stage, vacuum-distillation column with a single cut point. Material boiled above or below this cut point. The below cut-point material exited the top of the WFE where it was cooled and stored in drums. The greater-than cut point material exited the bottom of the WFE and discharged directly to a drum.
Feed pump and feed line plugging was observed when using a 2:1 liquids to coal ratio. Coking in the two reactors (called digesters) was observed when operated at temperatures of 410°C. Plumbing modifications were instituted to reduce plugging. At Quantex’s request, it was decided to operate the multi-ton run using a 2.5:1 liquids to coal ratio to further reduce the chance of plugging. The temperature used to control the digester temperature was the bulk fluid temperature (measured via a thermocouple inserted in a well extending into the digesters). If a layer of material built up on the digester walls, reducing heat transfer, the wall temperature was automatically raised to achieve the desired bulk temperature. Thus, it was possible to have a 410°C or even 420°C wall temperature, and that led to coking in the earlier runs. To overcome this, the temperature set point was lowered in future runs, as discussed in Section 6. This may result in reduced time at temperature. If in the longer-time tests it is found that the centrate viscosity is too high, we may need to reduce the feed rate, or raise the liquid level in the digesters, in order to increase residence time.
In the planned longer-time Quantex tests, we hoped to simulate commercial operation as closely as possible. One major difference between what we do in the lab and what we expect to do in a commercial plant relates to the use of CTD. In a commercial plant we would use CTD only for startup, using WFE bottoms instead of CTD. Tests in the batch autoclave where the equivalent of WFE bottoms was used to displace CTD resulted (in some cases) in comparable, or even superior solubility rates. So our plan is to start with a high proportion of CTD, but in subsequent runs to displace it with recycled CTD (RCTD) prepared from a mixture of WFE overhead and bottoms.
5.4.3 Top Biomass-Derived Solvents.
The top ten biomass derived solvents, along with their solubility of Ohio coal, viscosity, and availability are noted in Table 13.
Table 13. Top 10 Bio-Solvents
To accommodate the existing feeding system at the Quantex site, a 50:50 mixture of bio-solvent 27B and CTD was prepared at Battelle and then shipped to Quantex in four 55-gallon drums for Quantex Run #4. The contents of these drums, when heated to ~50°C, will have a viscosity of 3,650 cP (at 70°C, 359 cP, and at 90°C, 89 cP), so it can be fed in a manner similar to SBO that has been used in prior tests. Findings for BS-27B are noted in Table 14.
Table 14. Batch Autoclave Liquefaction for Bio-solvent 27 Using Carbon-Black Oil CTD
We chose BS-27B for the main test series at Quantex because of its excellent performance and excellent availability (it can be purchased from a commercial vendor). To support those tests, a series of parametric tests were conducted at two temperatures (390 and 400°C) and two residence times (10 and 30 minutes). Unfortunately, we had used up all of our high quality (Carbon Black Oil) CTD, and were forced to use an inferior “light” CTD. This resulted in an average drop in solubility of 15 to 17%. The parametric test results are shown in Table 15. As can be seen, at 390°C, even with longer reaction times, we obtained a difficult to filter product. Solubility was higher with the longer reaction time. In contrast at 400°C, even at 10 minutes reaction time, we produced an easy to filter product. At this higher temperature, solubilities were unchanged with longer reaction times.
Table 15. Parametric Study Results for BS-27 Using Light CTD at 29% BS/MF Coal
The down-selected operating conditions for the Quantex pre-pilot scale run are as follows:
- Bio-solvent: 27B
- Temperature: 375°C in digester 1 and 400°C in digester 2
- Pressure: 400 psig
- Liquid to coal weight ratio: 2.5 to 1
- Bio-solvent to coal ratio: 24 lb/100 lb coal (mixed 50:50 with CTD)
- Additional CTD to coal ratio: initially 677/300 lb coal; reduced progressively to 72/300 lb coal
- Residence time at temperature: ~15 minutes in digester 1 and 15 minutest in digester 2, approximately 15 to 20 minutes at 400°C
- WFE temperature set point: 160°C at 29 mm Hg vacuum (atmospheric-pressure temperature equivalent of 290°C); more about this parameter below.
5.4.4 Planned Quantex Run.
A typical Quantex batch size is 300 lb of as-received (AR) coal, enough for about 6 hours of operation. Quantex does not have a sufficient number of trained staff to run 24 hours a day, so operation between 3 and 6 hours (1 to 2 batches) a day was planned.
If we assume the coal has 8.7 wt% ash (dry basis), and the centrifuge will capture 98% of the ash/mineral matter, the coal solubility will be 87%, and the centrifuge cake ash level will be 36.9%, we can make a material balance. The assumptions are shown in Table 16 and the results are shown graphically in Figure 18.
Table 16. Quantex Run Assumptions
Figure 18. Projected material balance around Digester/Centrifuge and WFE in the Quantex plant.
As noted in Figure 18 above, the process requires 605 lb/hr of RCTD. The relative split between the WFE overhead and bottoms is controlled by the selected set point temperature. The WFE is operated at a modestly-high vacuum 15-in. water (28 mm Hg) so that has been factored into the WFE set point operating parameter calculations. Based on batch single-stage distillation tests conducted under vacuum, we believe that set point such 320°F (160°C) at 15-in. water vacuum (28 mm Hg) is equivalent to 554°F (290°C) atmospheric pressure temperature and would meet these requirements.
In this configuration, all of the WFE overhead (92 lb) goes to the product stream. About 66% of the WFE bottoms (605 lb) are recycled as RCTD, leaving 223 lbs of bottoms to go to product. Total product per 300 lb coal lot is 315 lb or about 38 gallons.
6.0 PRE-PILOT-SCALE COAL LIQUEFACTION AND SYNCRUDE PRODUCTION
6.1 Introduction
Our team partner Quantex Energy was selected to produce large quantities of liquefied coal (Syncrude) in their pre-pilot plant scale coal liquefaction facility. To obtain information on scale up of the batch liquefaction tests conducted by Battelle and PSU, we elected to perform some tests to make enough material for syncrude upgrading in Subtask 1.04, described earlier in Section 5.4.
6.2 Description of the Quantex Pre-Pilot-Scale Coal Liquefaction Plant
A schematic of the Quantex liquefaction system is shown in Figure 19. A photograph of the system is shown in Figure 20.
Figure 19. Process flow diagram for the 1 TPD Quantex facility
Figure 20. Photograph of the Quantex 1 TPD pre-pilot plant facility.
The five components that were used in the testing to date are noted below:
- Coal preparation
- Co-feeding of coal, bio-solvent, and coal tar distillate
- Digestion
- Solids removal by centrifugation
- Thermal separation into a product fraction and a recycle CTD fraction.
6.2.1 Coal Preparation
Coal is obtained from the coal company in either 1-ton synthetic-fiber “Super Sacks” of in metal 55-gallon drums and transported to the Quantex site. In a special coal-grinding building, coal is conveyed up from the feed bin to a hammermill fitted with a metal screen. The mill produces coal with a size 90% smaller than 20 mesh (0.0331 inches, 0.841 mm); after it is ground it falls into 55-gallon drums. There is no drier including with the Quantex setup, but a small level of moisture removal is achieved during the handling and grinding process. Enough coal for the next day’s processing is typically ground and stored and be ready for liquefaction.
Representative samples of the Leer, West Virginal coal and Bramhi, Ohio coal were obtained and were analyzed for proximate and ultimate analysis along sulfur forms; results were presented earlier, in Section 5.1.1.
6.2.2 Preparation of the Battelle Bio-solvent and Coal Tar Distillate
The bio-solvents were prepared at Battelle using proprietary processes. Coal tar distillated (CTD) is a liquid obtained through pyrolysis of coal. It is distillation into different fractions before sale. Samples of three different CTD fractions were obtained from Koppers. Their relative effectiveness was evaluated in batch autoclave tests at Battelle and “carbon black oil” was determined to be the best.
6.2.3 Co-feeding of Coal, Bio-solvent, and Coal Tar Distillate
The pre-ground coal was screw-fed at a controlled rate of 100 lb/hr using a volumetric feeder. There it is mixed with CTD, from a 250-gal tote, and bio-solvent fed from a 55-gallon drum via a metering pump. There they were mixed using a special feed-mixing pump.
Depending on the set up of the feed-mixing pump, there can be limited or extensive grinding of the coal during its operation. Due to the grinding, the coal is preheated from ambient temperatures (80°F, 27°C) to over 150°F (66°C). The output of the mixing pump is sent to a slurry-preparation vessel. A pump at the bottom of the slurry-prep vessel sends the pre-heated coal/solvent/CTD mix to a valve that allows a portion to be sent back to the top of the slurry-prep vessel and a portion sent to the dewatering vessel. A similar mixing action is achieved as the coal temperature is raised about the boiling point of water and a portion of the water is removed. These gases are cooled and condensed and the liquids are sent to an oil/water separator. The gases are sent to a water scrubber for cleaning and removal of an organic mist.
After slurry prep and dewatering, the slurry is cooled and passed to a digester feed pump. Here the pressure is raised to the desired operating pressure.
6.2.4 Liquefaction of Coal
The output of the pressurization is sent to the first of two 10-gallon digesters. The digesters have level- control sensors so that when the volume exceeds 50%, a valve is automatically opened to allow a portion to be removed. A pump at the bottom of Digester 1 sends the fluid to a splitter. It allows a portion to be returned to the top of the first unit while the other portion is sent to Digester 2. The digesters do not have agitators, so this internal recirculation provides mixing. The output of Digester 2 output flows by pressure difference to a downstream heat exchanger/cooler. By adjustment of the level set points, different residence times can be achieved.
As the temperature of the coal and solvent mixture is increased any remaining water and a portion of the coal’s volatile matter are released. A pressure-control valve allows these gasses to be vented off while maintaining a constant pressure of 400 psig. These gases are sent to a condenser to knock out the organics and then to a holding tank. The gases are sent to the scrubber for cleaning.
6.2.5 Solids Removal by Centrifugation
The reacted coal is too hot to be dewatered directly. Therefore, the coal slurry is passed through two heat exchangers in series and then sent to a centrate feed and mix tank. From this agitated tank the liquefied coal can be held or sent to the centrifuge. In past runs, the material has been stored for several hours, and then centrifuged. The hope in this extended duration test is that the liquefied coal will spend a relatively short time in the feed tank before being de-ashed.
The solids from the centrifuge fall into a 55-gallon drum. The liquid stream, called the centrate, is sent to a centrate receiver. From there it can be held, pumped back to the centrifuge feed tank for reprocessing or sent on the wiped-film evaporator (WFE).
6.2.6 Thermal Separation into a Product Fraction and a Recycle CTD Fraction
The WFE is a tall, single column that allows the de-ashed, liquefied coal to be separated based on boiling point difference. The unit consists of a heated body and a rotor. The de-ashed, liquefied coal in pumped in above the heated zone and is evenly distributed over the unit's inner surface by the rotor. As the product spirals down the wall, the rotor blades generate highly turbulent flow to promote effective heating and rapid mass transfer. Volatile components are rapidly vaporized. Vapors flow co-currently with the de-volatilized liquid through the WFE and are condensed as our product and stored in a holding tank. The vapors off the tank are cooled to condense additional liquids – non-condensable gases are send to the scrubber for cleaning. The less volatile components are discharged at the WFE bottom directly into large tank and pumped, and pressure equalization, to a 55-gallon drum. A portion of these bottoms could be used as recycle coal tar distillate (RCTD). The unit is operated at vacuum to lower the fluid’s boiling point and allow separation at a lower temperature. Its performance is similar to a single-stage vacuum distillation column without reflux.
The unit is run by selecting a single cut-point temperature. Everything that boils at a lower temperature passes up and is drawn off in the overhead stream, condensed and collected. Everything with a higher than cut-point boiling temperature flows out the bottom. This cut-point temperature is frequently reported as the atmospheric-pressure equivalent temperature. The actual temperature corresponding to this reported temperature depends on the vacuum available.
6.3 Shake-down Testing Conducted at Quantex (Quantex Run #1)
Since the Quantex system had not been operated for a while, several changes were made to the plant before shake-down testing began.
6.3.1 Description of First Run with Leer Coal Company, WV Coal.
In January 2015, a 30-gallon trial was carried out using the following:
- 10 gallons of Leer WV Lower Kittanning coal ground to -25 mesh
- 20 gallons of solvent
- 14 gallons CTD.
- 6 gallons soybean oil (SBO).
During startup, problems with the software used to control the plant delayed operation. The problems were resolved, but the coal continued to recirculate through the colloidal mill in the feed preparation loop for hours (which allowed the coal to be ground and re-ground to a smaller and smaller size and the slurry had to be heated to temperatures greater than the planned 105°C level). The smaller-than planned coal particle size may have led to the clogging and plugging problems that were observed at several points in the system. It was concluded that the output would not be representative of the liquefaction of Leer coal so a complete material balance was not performed. While problems were noted, the equipment was operated long enough to make a first examination of the products that can be obtained by Leer coal liquefaction.
6.3.2 Repeat Run with Leer, WV Coal.
Modifications were made to the Quantex facility, including the use of fume hoods for removal of gas and vapors from the WFE. In addition, the Zenith pump that fed the primary digester was moved upstream in order to avoid gravity feeding from the slurry prep tank to the primary digester. The re-located pump allowed recirculation of liquids until there was demand at the reactor, in which case the 400 psig pump should be sufficient to deliver liquids through the line to the reactor.
As noted above, the colloid mill was used as a way to heat the slurry prior to its entry to the reactor. Operation in this manner allowed the coal to be ground to small particle sizes. This may be beneficiation from a reaction point of view but could also produce a more viscous slurry making it more difficult to pump. As the colloid mill is the primary way to add heat the inlet slurry it was retained – but the internal settings such that the teeth of the mill were adjusted to the largest possible gap, thus reducing the ability of the grinder to create an emulsion, while continuing to rely on the creation of shear as a means to heat the working fluid to desired temperature.
The Quantex team made several other improvements to the liquefaction system including installing cowls to permit takeoff of high temperature liquids from the WFE while containing the vapors. A protective epoxy floor coating was applied around the system to protect the cement floor.
A control run was carried out on 19 February 2015. It was operated without coal in order to ensure that the new pumped section was viable and leak free. The reactor was successfully loaded and unloaded at about 400°C. The system passed this test and was considered ready for trials with coal. However, as this was a major change, the planned Bramhi Ohio coal run was postponed, in order to evaluate the changes using the previously tested Leer coal.
The Leer coal test was initiated. Coal did not flow smoothly. The coarsely ground coal exiting the colloid mill plugged the line. In addition, other technical difficulties were encountered. As the slurry entered the digester, the fluid level sensor started to gyrate, bouncing from an indication of completely full to completely empty. This inability to accurately measure the fluid level may have been due to foaming inside the reactor, creating false liquid level signals.
The run was stopped. The sensor issue was later modified with the assistance of the manufacturer. In addition, because the last run had issues with a difficult-to-flow emulsion being formed, the Quantex team elected to return to the original protocol, which specified 75°C pre-heating of the slurry rather than 105°C. It was recognized, as a result, the heating would be insufficient to boil away moisture prior to the reactor. Moisture could instead be liberated and captured with overhead vent stream from the reactor loop. This was thought to be adequate for coal with moisture content of a few percent.
6.4 Run with Ohio Coal (Quantex Run # 2)
This test was conducted using soybean oil, in an attempt to establish a baseline to compare the Battelle process with. The Ohio (Bramhi) coal flowed more smoothly through the equipment, but it seemed to be more reactive upon entering the reactor, as the liquid level again showed wild gyrations. The varying liquid level can be due to the presence of boiling water. The system was stopped to make further adjustments to the liquid level control system. It was decided to run with WV coal to make sure the control problem was corrected, and then switch to Ohio coal. A sample of the WFE-overhead syncrude from this run was utilized for initial upgrading work at UDRI (see Section 7.4).
6.4.1 Work-up of Syncrude from Quantex Run # 2.
Quantex delivered a 5-gallon pail of the liquefied Ohio coal to Battelle. Inspection showed that the material in the top of the pail was free flowing and had a viscosity similar to vegetable oil while the material in the bottom of the pail was more viscous. This indicates, as also observed in lab-scale testing, that soybean oil does not react with coal, as it is incapable of hydrogen transfer.
The material in the 5-gallon pail was first mixed well and then 6 liters (L) were removed. Approximately 1 L was loaded into a pressure filter and filtered at room temperature. Over a series of tests, a total of 5,545 g was filtered. About 73 wt% of the material passed through the 20 micron filter paper. The once filtered material was re-filtered and 99.6% of this material passed through a 0.22 micron filter.
The filtered material was uniform with no separation in the top or bottom phases. Some physical properties were measured, see Table 17. They showed similar results regardless of whether the test sample was drawn from the top or bottom of the bottle.
6.5 Run with WV and Ohio Coal (Quantex Run # 3)
This test was conducted using Battelle-prepared BS-19A bio-solvent. To minimize digester level fluctuations, the plant setup was returned to a target slurry preheat temperature of 105oC in order to boil off moisture before it could enter the reactor loop. This was successful. The plant was operated one day with Leer coal. The following morning the tanks were drained and an Ohio coal test initiated. This run, designated Quantex Run #3, was successful as the Ohio coal was liquefied and de-ashed. The processed coal was then sent to the WFE. Some problems at the discharge end of the WFE were encountered preventing the production of a representative pitch product. However, the WFE overhead stream was recovered.
Process improvements allowed for processing most of the desired 203 kg run (69.4 kg coal; 96.0 kg CTD; 37.8 kg BS-19A). However, due to a digester wall being at 430°C the digester coked and plugged before processing the last 47.5 kg of mixture. The digester wall temperature will be held lower in future runs in order to avoid coking. The rest of the material was run through a centrifuge and then through the WFE, operating at 290°C atmospheric equivalent. The material through the WFE was found to be 10 wt% <290°C boiling and 90 wt%>290°C boiling. The Flow rates were estimated and materials sampled for analysis.
Table 17. Properties of Room-Temperature-Filtered Liquefied Coal from Quantex Run #2
Discussions with Quantex indicated that there should be little if any insoluble material since the coal had been centrifuged and distilled (in the WFE). They suggested that we hot filter the material.
A sample of the liquefied Ohio coal from the 5-gallon pail, after good mixing, was withdrawn and hot filtered at 80 to 90°C in the same pressure filter. Under these conditions, the following was found:
- A total of 6,758 g was filtered; 99.5 wt% of the WFE material passed through 20-micron filter paper.
- Actual filtration time (after heating a liter of feed to temperature and preparing the apparatus) of each 1-L lot took less than 5 minutes.
- The once-filtered -20 micron material was re-filtered using a 0.22-micron paper. 98.6 wt% of the once-filtered material passed through the finer filter. The filtration of 1 L took 10 to 15 minutes.
- Overall, 98% of the raw WFE material was found to be smaller than 0.22 micron. It could be at higher temperatures, even more could be filtered.
- The filtrate the following day, at room temperature, was a sludge like material with liquid on top and a second semi-solid phase at the bottom. But, the measured densities of both phases were very similar, so the bottom phase was not actually a “heavy” phase. It compressed easily between your fingers, so it should be easy to pump.
Table 18. Properties of Hot Filtered Liquefied Coal (Quantex Run #2)
6.5.1 Work-up of Syncrude from Quantex Run #3.
A sample of the centrate before the WFE was taken and sent to UDRI for hydrotreatment. This centrate had the approximate distillation cuts, shown in Table 19, when distilled in the lab at Battelle.
Table 19. Material Sent to UDRI for Hydrotreatment Evaluation (Quantex Run #3)
6.6 Quantex Run #4 with Ohio Coal
Quantex Run #4 was run with bio-solvent BS-27B. Because of processing issues, only 150 lbs coal was processed during this run. The process feed was:
- Coal: 150 lbs of ground, as received (AR) coal.
- CTD: 303 lbs of Koppers Carbon Black Oil CTD obtained from their Clairton, PA plant.
- Bio-solvent BS-27B: 36 lbs bio-solvent 27 premixed with 36 lbs of Koppers Carbon Black Oil CTD for a total feed of 72 lbs.
The digester operating conditions for Quantex Run #4 were as follows:
- Temperature:
- Digester 1: 375°C
- Digester 2: 400-450°C (there was a thermocouple issue leading to over-heating)
- Pressure: 390 psig
- Residence time: 35 minutes
- Liquid to coal ratio: 2.5:1 (weight basis)
The material exiting the digester was fed to the centrifuge feed tank and maintained at a temperature of about 140°C. Next, the material was run through the centrifuge (at about 110°C). Roughly 123 lbs solids from the centrifuge formed a filter cake and fell into a 55-gallon drum (approximately 50% solids). The liquid stream, called the centrate, was sent the wiped-film evaporator (WFE) feed tank, then pumped to the WFE. The WFE is a tall, single column that allows the de-ashed, liquefied coal to be separated based on boiling point differences. The WFE consists of a heated body and a rotor. The de-ashed, liquefied coal was pumped in above the heated zone and was evenly distributed over the unit's inner surface by the rotor. As the product ran down the wall, the rotor blades generated turbulent flow to promote effective heating and rapid mass transfer. Volatile components in the feed were rapidly vaporized. Vapors flowed co-currently with the de-volatilized liquid through the WFE and were condensed as our product and stored in a holding tank. Vapors off the tank were cooled to condense additional liquids, while non-condensable gases were sent to the scrubber for cleaning. The less volatile components were discharged at the WFE bottom directly into a 55-gallon drum. The unit was operated at a vacuum pressure of about 21 mm Hg and a temperature of 230°C giving an atmospheric equivalent distillation temperature of 380°C. The distillate was labeled “lights” and the WFE bottom was labeled “heavies”.
In Quantex Run #4, approximately 70% of the moisture- and ash-free (MAF) coal fed was solubilized producing about 184 lbs of lights, and 201 lbs of heavies, while the unsolubilized coal and some liquids formed 123 lbs of centrifuge filter cake. These materials were produced from 150 lbs of Ohio Bramhi coal, 303 lbs of Koppers Carbon Black Oil CTD, and 36 lbs of bio-solvent BS 27B. Materials produced during Run #4 were sent to Battelle for further analysis and verification.
A mass balance was conducted on Quantex Run #4 as shown in Figure 21 and Table 20.
Figure 21. Quantex Test #4 process conditions, inputs, and outputs.
Table 20. Quantex Run #4 Overall Material Balance
Total out/in provides a 97% mass balance for Run #4 at Quantex. Contributions to the 4% mass lost likely came from a small amount of coking on the digesters, residual material left in the system, leaks, and water/light hydrocarbons vented during the evaporation process. Overall, the 97% mass balance provides excellent closure for the entire Run #4.
6.7 Quantex Run #5
Quantex Run #5 was run with bio-solvent BS-32. The operating conditions for Quantex Run #5 were as follows:
- Material
- Coal: 160 lbs of ground, as received (AR) coal.
- Koppers Carbon Black Oil CTD: 368 lbs CTD obtained from their Clairton, PA plant.
- Bio-solvent BS-32: 64 lbs.
- Temperature:
- Digester 1: 401°C
- Digester 2: 404°C
- Pressure: 370 psig
- Residence time: 30-34 minutes
- Liquid to coal ratio: 2.7:1 (weight basis)
- WFE Light: 228 lbs
- WFE Heavy: 206 lbs
- Filter Cake: 131 lbs
Figure 22. Quantex Test #5 process conditions, inputs, and outputs.
Table 21. Quantex Test #5 Overall Material Balance
Total out/in provides a 95% mass balance for Run #5 at Quantex. Contributions to the 5% mass lost likely came from a small amount of coking on the digesters, residual material left in the system, leaks, and water/light hydrocarbons vented during the evaporation process. Overall, the 95% mass balance provides excellent closure for the entire Run #5.
6.7.1 Discussion on Results from Quantex Runs #4 and #5.
Both Quantex Runs #4 and #5 successfully produced enough material for hydrotreatment/hydrogenation to jet fuel. The Battelle bio-solvents performed well, but minor adjustments are necessary to improve performance. Further optimization is needed for tests using BS-27, due to the coking and plugging issues observed in Run #4. We found that BS-27 would benefit from the addition of a second bio-solvent in order to maintain a slurry that doesn’t swell and plug lines. This is supported by the work in Quantex Run #3 with bio-solvent BS-19A, which has similar structure to BS-27. In addition, maintaining lower slurry temperatures also inhibits early coal swelling. Run #5 with BS-32 ran fairly smoothly and also benefitted from the addition of a second bio-solvent to prevent coal swelling in the slurry stage. When recycling WFE heavies to replace Koppers CTD, it was found that adjustments in the level of BS-32 are needed. Very heavy oil was produced when the WFE heavies were recycled due to hydrogen-starving the system, resulting in only partially liquefied coal. Initially, we used heavies to replace 50% of the CTD without adjusting the bio-solvent, which led to plugging of the centrifuge lines. In order to reduce plugging, the level of bio-solvent or recycled WFE heavies must be adjusted. However, sufficient syncrude quantities have been produced from both runs to allow testing in Task 3 (pre-pilot hydrotreatment/hydrogenation). The syncrude for Quantex Run #5 was actually used in Intertek Run #1 test.
6.8 Run with Ohio Coal (Quantex Run #6A)
This test was conducted using Battelle-prepared BS-40D bio-solvent and the previously-tested Ohio coal from Bramhi Coal Company. To minimize potential issues in the slurry step, the slurry preheater temperature was held below 150ºC. On the first day, we consistently had plugging of Digester 1 recirculation pump. This was due to the slurry not reaching a sufficient temperature and pretreatment time before reaching the gear pump. A decision was made to reverse the recirculation in Digester 1 so that the mixture would gain more pretreatment time at temperature before reaching the gear pump. These changes were successful and were implemented on day two. The slurry temperature on day two ended up reaching 71°C due to manual feeding of the system. The plant was then operated for one day with the Ohio/Bramhi coal. The product was then centrifuged and 15 gallons of centrate were transferred to a drum. The leftover centrate was then sent to the wiped film evaporator (WFE) to fractionate the centrate boiling below 380°C. The WFE ran for one pass to obtain a light (<380°C boiling point) and heavy (>380°C) product.
The improvements allowed for processing of a 234 kg run (66.18 kg coal; 119.36 kg CTD; 31.91 kg BS-19A; 16.55 kg SBO). Fluctuations were noted in the digester temperatures but both indicated a range of 400-415°C. The centrate through the WFE was found to be about 40 wt% light and 60 wt% heavy product. The flowrates were estimated and materials sampled for analysis. The mass balance for this run was about 97.0 wt%; about 18.68 kg were last in piping tanks.
6.9 Run with Ohio Coal (Quantex Run #6B)
This test was conducted using Battelle-prepared BS-40D bio-solvent and recycle solvent from Quantex Run #6A to better simulate commercial-scale operation under steady-state recycle of some solvent. To minimize potential issues in the slurry pretreatment step, the slurry preheater temperature was held below 150ºC. We used the same flow scheme as Run #6A. The plant was operated for a half-day with Ohio/Bramhi coal. The product was then centrifuged and all centrate was transferred to drums for testing at Intertek (Task 2.02.01). This approximately half-day run allowed for processing of 217.1 kg of feed (54.27 kg coal; 68.39 kg CTD; 27.14 kg BS-19A; 67.43 kg Recycle). Fluctuations were noted in the digester temperatures but both indicated a range of 400°C-415°C. This process produced 132.27 kg of centrate, which was placed into drums. The Flow rates were estimated and materials sampled for analysis. The mass balance for this run was about 97.6%. The syncrude from Quantex Run #6B was used for upgrading testing in Intertek Run #2.
6.9.1 Product Analyses
The liquid samples from runs 6A and 6B were submitted for ultimate analysis and can be seen in Table 22. The samples were also submitted for viscosity analysis.
Table 22. Ultimate Analyses of Syncrude for Quantex Runs #6A and #6B
We experienced difficulty in obtaining a representative sample of the centrifuge tails from both runs. This difficulty is due to various layering that takes place when starting and finishing the run through the centrifuge. It was decided that the best option for sampling was to drive a 1 inch pipe through the deepest part of the mixture. The liquid/solid mixture was then worked up by solvent rinsing and drying. The solid was submitted for ultimate analysis. The fraction of solids in the centrifuge tails for Runs #6A and #6B were 38.9 wt% and 42.1 wt%, respectively.
7.0 LAB-SCALE CATALYST TESTING FOR SYNCRUDE HYDROTREATING/HYDROGENATION
7.1 Introduction and Objective
The conversion of liquefied coal (coal converted from a solid to a liquid using bio-derived solvents) into a form suitable as a blending stock for jet turbine fuel requires three broad processes. First, the heteroatom species must be removed; primarily components containing nitrogen, oxygen, and sulfur. Second, the largely polynuclear aromatic structure of the coal derived components must be reduced (hydrogenated/hydrocracked) to predominantly paraffinic components, though some single and two-ring aromatics are permitted within the fuel specification. Third, the boiling range of the final product should be within the fuel specification. The first two processes, heteroatom removal and reduction, are typically accomplished via heterogeneous catalysis. The third may also be accomplished through heterogeneous catalysis, though fractionation may also be used.
The primary objectives of this subtask were to identify commercially available candidate catalysts for heteroatom removal and hydrogenation and the nominal processing conditions.
7.2 Catalytic Upgrading Background
For the upgrading of syncrude from the Battelle CTL process, a two-stage process was conceptualized. The two stages, shown graphically in Figure 23, are designated as Stage-1 and Stage-2, where Stage-1 would perform mainly heteroatom removal and Stage-2 would perform hydrogenation and hydrocracking reactions. A review of the literature shows that the most common commercially available catalysts for heteroatom removal are sulfided forms of CoMo and NiMo [5-11]. Catalysts for hydrogenation and hydrocracking (reduction) are bi-functional catalysts such as PtPd, sulfided NiMo deposed on acidic support such as zeolite Y and alumina [5-8]. Generally, PtPd is preferred, though this catalyst is more expensive than sulfided NiMo and is very sensitive to residual heteroatoms in the feed, principally N and S. NiMo is less expensive and can be sulfided, providing good performance in the presence of heteroatom species.
Figure 23. Two-stage hydrotreatment/hydrogenation of CTL syncrude.
Literature Review of Catalysts for CTL Syncrude Upgrading. Battelle team member PSU performed a literature review of catalysts for upgrading of coal-derived syncrudes. The review and the associated references are provided in Appendix A.
7.3 Jet Fuel and Diesel Specifications
The primary focus of this project was to produce jet fuel from coal. However, one can also produce diesel if desired. The specification for various distillate products of interest are discussed below.
The specification for conventionally produced (i.e., petroleum-based) jet fuel is ASTM D1655 – 11a. The requirements are presented in Table 23 [12]. Analysis indicates there is no stated ash or nitrogen limit in D1655. A discussion of each is provided below.
- Ash: No limit stated in D1655. However, there are practical limits. The limit for diesel fuel is 0.01 wt. % ash; so the practical limit for jet fuel is approximately the same.
- Nitrogen: D1655 states “Conventionally refined jet fuel contains trace levels of materials that are not hydrocarbons, including oxygenates, organosulfur, and nitrogenous compounds.”
- Sulfur:
- Sulfur, mercaptan, 0.003 mass % max by D3227
- Sulfur, total 0.30 mass % max by D1266, D2622, D4294, or D5453.
- Aromatics
- Aromatics, 25 volume % max by D1319; or 26.5 volume % max of by D6379.
- Naphthalene (a C10H8, double cyclic aromatic hydrocarbon), 3 volume percent max.
- Further breakdown of the types and amount of allowable aromatics are not listed. However they are indirectly restricted due to limitations on smoke point. The aromatic content and type affect combustion characteristics and smoke-forming tendencies.
- In general, paraffin hydrocarbons offer the most desirable combustion cleanliness characteristics for jet fuels. Naphthalenes are the next most desirable hydrocarbons for this use. Although olefins generally have good combustion characteristics, their poor gum stability usually limits their use in aircraft turbine fuels to about 1% or less. Aromatics generally have the least desirable combustion characteristics for aircraft turbine fuel. In aircraft turbines they tend to burn with a smoky flame and release a greater proportion of their chemical energy as undesirable thermal radiation than the other hydrocarbons. Naphthalenes or bicyclic aromatics produce more soot, smoke, and thermal radiation than monocyclic aromatics and are, therefore, the least desirable hydrocarbon class for aircraft jet fuel use. All of the following measurements are influenced by the hydrocarbon composition of the fuel and, therefore, pertain to combustion quality: smoke point, percent naphthalenes, and percent aromatics.
Table 23. ASTM D1655 Standard Specification for Aviation Turbine Fuels
There is also a specification for synthetically produced jet fuel, D7566-11a. The D7566 has different Annexes for different synthetic fuel; see Figure 23. However, there is no Annex for jet fuel from coal via direct liquefaction.
Figure 23. Annexes for synthetic aviation fuels
(From: https://www1.eere.energy.gov/bioenergy/pdfs/10_brown_roundtable.pdf )
Annex A1 is for Fischer-Tropsch (FT) synthetic paraffinic kerosene (SPK) fuel, Annex A2 is for Hydrotreated Esters of Fatty Acids (HEFA) SPK, and Annex A-3 is for the recently approved Direct Sugar to Hydrocarbons process to make Synthesized Iso-Paraffins (SIP). Annex A1 and A2 fuels must be blended with at least 50% petroleum jet fuel; Annex A3 fuel must be blended with at least 90% petroleum jet fuel.
It would be reasonable to expect the direct-coal-liquefaction jet fuel would require its own annex; it would definitely not fall under Annex A1 for FT fuels (even though A1 covers coal gasification to make syngas which is converted by FT technology into jet fuel). The annexes contain, in some cases, limits different than those specified in D1655. For example, for SPK and SIP, which contain only aliphatic hydrocarbons, the upper limits on the amount of aromatics is 0.5%; this is in contrast to D1655 where the limit is 25 volume % aromatics.
There are no stated ash limits, but there are limits on nitrogen, sulfur, and aromatics in D7566-11a. Details are provided below.
- Ash: No limit stated in D7566, however the practical limit is 0.01 wt. % established for diesel fuel.
- Nitrogen: In D7566 for all three annexes the nitrogen limit is 2 ppm; from: http://mycommittees.api.org/rasa/jfm/Shared%20Documents/Resource%20Materials/AS TM%20on%20Bio-Derived%20Fuels%20-%20D02.J0%20(10-02)%20item%205.pdf - see section 2.2.1.2.1 “Table A1.2 of ASTM D7566.”
- Sulfur: In D7566 for all three annexes, the limit is 0.0015 mass percent (15 ppm) – the same as ultra-low sulfur diesel.
- Aromatics limits are set for SPK and SIP (paraffinic compounds): In D7566 for all three annexes, the limit is 0.5 %. It would be reasonable to expect that for direct CTL jet fuel, a higher aromatics content would be permitted up to at least the D1655 maximum of 25 volume %.
- Grade No. 1-D S15: A special-purpose, light middle distillate fuel for use in diesel engine applications requiring a fuel with 15 ppm sulfur (maximum) and higher volatility than that provided by Grade No. 2-D S15 fuel.
- Grade No. 2-D S15: A general purpose, middle distillate fuel for use in diesel engine applications requiring a fuel with 15 ppm sulfur (maximum). It is especially suitable for use in applications with conditions of varying speed and load. S15 grades were not in the previous grade system and are commonly referred to as “Ultra-Low Sulfur” grades or ULSD. This is the more likely the target grade for the Battelle CTL process.
The overall approach was to identify and obtain small quantities (100 to 1000 mL) of commercially available catalysts and to screen their performance using model or actual syncrudes using a set of small trickle-bed reactors operating under nominal processing conditions. From this screening process one or more promising candidates were to be selected for more detailed evaluation.
The first task was to adjust and calibrate the reactor systems for these type of reaction “Syncrude Upgrade to jet fuel”. UDRI has several reactor systems that were used for this project. All of these reactor systems are very similar in their overall construction and flow path. A general schematic is shown in Figure 24, and a photograph of reactor #2 is shown in Figure 25. (Note that the reactors may be fitted with different inlet gases and product tanks depending on the specific configuration.) Each system includes separate gas and liquid feeds with heated transfer lines, a heated catalyst reactor with a moveable axial thermocouple, heated product collection tanks, and a gaseous vent that can be monitored with various online and offline analyzers.
Figure 24. General schematic of the trickle-bed catalyst reactors at UDRI.
Figure 25. Photograph of Reactor #2 configured for heteroatom removal.
Each reactor is a ½” diameter by 24” long 316 stainless steel tube vertically oriented inside a 3-zone furnace (Applied Testing Systems Inc.). The temperature of the 6” central heating zone can be controlled independently through thermocouples that are spot welded to the exterior of the reactor tube at the center of each heated zone and the temperature of the center zone can be monitored by a moveable thermocouple located within a well extending through the center of the reactor tube. The length of the tube within the furnace is covered with a split cylindrical brass tube of 1/8” thickness to assist in evenly distributing the heating loads. Reactor pressure is controlled by a back- pressure regulator downstream of the product receiver. Gases are supplied from regulated cylinders which were also used to pressurize the system. Gas flow rates were precisely controlled with 5850i Brooks mass flow controllers. Liquid feeds can be heated to avoid crystallization in a one liter feed tank and charged into a heated ISCO-500D syringe pump. The liquid and gas feeds combine at the top of the fixed-bed column and mix while flowing through approximately 4” of 54 mesh silicon carbide before contacting the catalyst-containing portion of the bed. Catalysts can be used as the as-received extrudates, crushed and sieved extrudates, or, more commonly, the crushed extrudates diluted with inert, high-conductivity material, such as silicon carbide. (Diluting the catalyst provides superior temperature control in the presence of strongly exo- or endothermic reactions and reduces channeling through the bed.) The lower section of the reactor is similar to the inlet section with a volume of 54 mesh silicon carbide resting on a plug of quartz wool on a 20 μm sintered stainless steel filter at the exit of the reactor tube.
If sulfiding and/or activation procedures are provided by the manufacturer, these procedures are used. In the manufacturer’s SOPs are not provided then procedures developed in-house are used. For example, base metal catalysts are activated after the reactor has been prepared and mounted in the reactor furnace. This is usually conducted at 450°C with hydrogen flowing at an equivalent gas hourly space velocity (GHSV) of 13.7/hr at 100 psi for 4 hours. Catalysts that require sulfiding are treated after they are loaded into the reactor tube, but before the tube is installed in the reactor furnace. Briefly, the reactor is mounted in a horizontal furnace, supplied with a flowing mixture of hydrogen and hydrogen sulfide at atmospheric pressure, and heated gradually over 2 hours to 350°C, held for 6 hours, cooled to room temperature, and then capped for transfer to the reactor furnace.
For each test run the reactors are typically operated with a LHSV of 1/hr based on the dry mass of the catalyst. Hydrogen is supplied at a molar ratio of 10:1 based on the known or estimated mean molecular weight of the liquid feed. To measure H2 consumption N2 (%volume ~2%) can used as internal reference gas. The sulfided reactors and the base metal reactors are typically operated with 800 psi of hydrogen. The sulfided reactors are conditioned at the initial temperature for the test sequence with flowing liquid feed for 1.5 hours. The product collected during this period is then drained from the product tank, and a test volume collected for 1.5 hours. This sample is then drained from the tank and stored for analysis. The temperature is then set for the next test point and the reactor allowed to run until the reactor temperature is steady, typically about 1.5 hours. The product tank is then drained, and the process repeated until the test sequence is completed. The reactor temperature is normally increased from the starting temperature to the final temperature in predetermined steps. However, in some cases the temperature may be decreased from some upper starting temperature to a lower ending temperature. To measure the hydrogen consumption the reactor effluent rates are recorded and averaged, then the effluent gas is analyzed using the online 3000 Micro GC (Inficon).
Reviewing the available reactors, Reactors #2 and #6 were selected for the heteroatom removal since these systems have been used previously with sulfided catalysts. Reactor #1 was selected for the upgrading of the product from the heteroatom removal as this system has been previously used as a hydrotreating reactor and has not been operated with sulfided catalysts.
To quantify the performance of the Stage-1 reactor, the conversion of selected heteroatom components was measured using gas chromatography–mass spectrometry (GC-MS). Furthermore, selected samples were submitted for quantitative elemental analysis. Briefly, the analysis was used to measure the relative concentration of carbazole (N), dibenzothiophene (O), and dibenzofuran (S) between the product and feed. This analysis was relatively quick and could be performed in-house and served as a general-purpose screen tool. The quantitative analysis of selected elements (C, H, N, O, and S) was conducted by an analytical services laboratory (Galbraith Laboratories, Inc., Knoxville, TN) which gave more comprehensive results for the heteroatom conversion, but was more costly in terms of time and resources.
The performance of the Stage-2 reactor was quantified using GCxGC to conduct a hydrocarbon type analysis of the reactor feed and product. This analysis quantifies the major hydrocarbon classes in the sample by type such as the number of aromatic rings, cyclic or linear and branched alkyl aromatics, and cyclic, linear, or branched paraffins. Of particular interest for this program is the conversion of polynuclear aromatic hydrocarbons to single-ring aromatic hydrocarbons and paraffins (linear, branched, and cyclic). From this analysis, the degree of HDA could be determined as well as measuring the concentration of various aromatic components in the product. The latter is of interest as jet fuel specifications allow up to 25% single-ringed aromatics and 3% di-aromatics (alkyl naphthalenes).
7.5 Materials.
The feedstocks used in this program consisted of model fluids composed in-house at UDRI and actual syncrudes provided by Battelle. The model fluids consisted of Aromatic-200 (Exxon Mobile), a complex petroleum distillate of C12–C15 aromatics in the form of alkylnaphthalenes. This was used either as-received or blended with selected heteroatom components. Based on a review of the literature on petroleum refining and the production of ultra-low sulfur diesel fuel, dibenzothiophene (DBT) and carbazole (see Figure 26) were selected as model sulfur and nitrogen containing species, respectively. These were blended with the Aromatic-200 to give a total sulfur content of 1% mole/mole and a nitrogen content of 0.01% mole/mole. After blending, the Aromatic-200 was designated Aromatic-200SN. Note that the molecular structures of these model compounds are very similar to dibenzofuran, which was later used as a marker species for the presence of oxygenates in the actual syncrudes, but not used in the model feedstocks.
Figure 26. Molecular structures of dibenzothiophene (left) and carbazole (right).
The actual syncrudes were initially made available in relatively small volumes and were the products of an evolving process. In total, 6 syncrudes were delivered in two broad versions. The first two syncrudes (SC-1 and 2) were from an early development production batch (Quantex Run #2). These were very viscous and had a significant solid or gel phase that required sonication and/or mild heating (80-90°C) to homogenize. The syncrudes WFE SC-1 through 3 are from the same syncrude batch but filtered on different days as needed; these were from a more refined production process (Quantex Run #3; feed to the wiped film evaporator – WFE). WFE SC-4 is distillation fraction of WFE (Quantex Run #3) below 450°C. These were less viscous than the first set, though they also included a second phase that required sonication and/or mild heating (50-80°C) to homogenize.
GC-MS analysis of the syncrudes showed that their overall composition was very similar. All of the syncrudes were very complex mixtures of polynuclear aromatic hydrocarbons (PAHs) spanning from about naphthalene (2 rings) to benzopyrene (5 rings) with an average density of 1.102 g/mL. Figure 27 includes the GC-MS analysis of the Aromatic-200SN showing that this model feed consists primarily of alkyl naphthalenes plus the added dibenzothiophene and carbazole (average density 0.982 g/mL). The elemental analysis and density of the syncrudes are summarized in Table 24. It shows that there was some variability in the heteroatom content of the various syncrudes, though the density was fairly consistent. Taking 10ppm as a nominal target for the final concentration of N, O, and S, suggested that an average heteroatom removal efficiency on the order of 99.9% was required.
Figure 27. GC-MS analysis of the various syncrudes and the model feed.
Table 24. Syncrudes
The catalysts used in this subtask were commercial and developmental catalysts. We are required by non-disclosure agreement (NDA) to not disclose the identity of these catalysts. A list of the catalysts tested is provided in Table 25.
Table 25. Catalysts Tested for Upgrading of CTL Syncrudes.
7.6 Results and Discussion.
The Stage-1 catalyst evaluation was conducted in four broad phases largely driven by the availability of syncrude and catalysts. The first phase was conducted using the model Aromatic-200SN feed with catalysts available in-house. This verified the operation of the reactors, calibration of the system and established the initial operating conditions for subsequent analyses. The second phase used the first samples of syncrude. These were processed through Stage-1 and provided information on the operating conditions and challenges associated with these feeds and analysis. The third phase focused on screening the remaining Stage-1 catalysts using a standardized set of conditions based on the initial work with the first syncrude material. In the fourth phase the most promising candidate was selected for more detailed evaluation and to produce a finished Stage-1 product for upgrading in Stage-2.
7.6.1 Stage-1, Series 1 - Aromatic-200SN Model Feed.
For this initial test sequence the reactor was prepared and catalyst sulfided as described above using the as-received extrudates that were crushed, sieved 32/60 mesh, weighed (approximately 2.6 g, 3.5 mL), diluted 1:1 with 54 mesh SiC, loaded into the reactor tube, sulfided, and then transferred to the reactor assembly. The catalyst was initially conditioned using the neat Aromatic-200, and then with the Aromatic 200SN. Tests were conducted from 300°C to 400°C at 5°C intervals using a LHSV of 1/hr (nominally 5.2-5.5 L/min) and 800 psi hydrogen flowing at a molar feed ratio of 10:1 (nominally 46-49 mL/min at STP) assuming a mean molecular weight for the feed of 252.54 g/mol. At each temperature, a sample of the product liquid was analyzed by GC-MS. The conversion of DBT and carbazole was measured using the peak areas from the extracted ion chromatograms taking the 184 m/e and 167 m/e mass fragments as being characteristic of DBT and carbazole, respectively. The sulfur and nitrogen concentration was estimated from the measured conversion and the known starting concentrations of DBT and carbazole and assuming that no other organic sulfur or nitrogen compounds were produced as products.
The initial tests were performed using catalyst A (CoMo) and the Aromatic-200SN feed. The overall conversion and heteroatom results are summarized in Figure 28. These results show that the concentration of DBT steadily declined as the temperature increased from 300°C to 400°C, while the carbazole passed through a minimum at approximately 350°C. Conversion of the Aromatic-200 to naphthalenes and tetrahydronaphthalenes was also observed. Small amounts of alkyl benzenes and decalin are also present in the product.
As the tests with Catalyst-1 were concluding the commercial catalysts became available for testing. Catalyst “C” was selected for the first series of tests using the Aromatic-200SN feed as there was interest in comparing this NiMo catalyst with the CoMo catalyst used previously. Catalyst “D” was also selected for evaluation at the manufacturer’s literature indicated that this catalyst may show an activity lower than Catalyst “C”.
The results for the catalysts ‘C” and “D” are summarized in Figures 29 and 30, respectively. These show that the performance of these two catalysts is indeed very similar to each other. Both showed excellent activity towards the conversion of carbazole, reducing it to below the detection limit of the GC-MS by approximately 320°C. They also showed good activity towards the conversion of DBT up to about 350°C, after which the activity slightly declined. Both of these catalysts showed a much higher activity towards the reduction of carbazole as compared to catalyst “A”, while they showed a similar level of activity towards the conversion of DBT throughout most of the temperature range used here.
Figure 28. Summary of the results for Aromatic-200SN using catalyst “A” (CoMo) from 300-400°C with 800 psi hydrogen and a LHSV of 1/hr.
Figure 29. Summary of the results for Aromatic-200SN using catalyst “C” (NiMo) from 300-400°C with 800 psi hydrogen and a LHSV of 1/hr.
Figure 30. Summary of the results for Aromatic-200SN-2 using catalyst “D” (NiMo) from 300-400°C with 800 psi hydrogen and a LHSV of 1/hr.
Figure 31. Comparison of the overall results for S (left) and N (right) from the CoMo (Catalyst A) and NiMo (C and D) catalysts.
Series-2 included the initial tests using the early syncrudes SC-1 and -2 and WFE SC-1. These syncrudes were produced using the baseline solvent, i.e., soybean oil. The SC-1 and -2 were small volumes of a very similar syncrude that had a coarse solid phase at room temperature that required sonication and/or mild heating (80-90°C) to homogenize. In the initial Phase the catalysts were crushed, sieved, and diluted 1:1 with SiC before being sulfided. The goal of dilution of catalyst with SiC is to better control exothermic reaction and prevent hotspots. The runs included an initial conditioning period of 150 hours under a nominal startup condition; typically 360°C with 700 psi of hydrogen and a LHSV of 0.3/hr. After this initial conditioning period the reactor performance was evaluated at various temperatures and pressures at a fixed LHSV. The primary measure of reactor performance was the analysis of the N, O, S marker components (carbazole, dibenzofuran, and dibenzothiophene) as described above as well as the product density and appearance. Selected samples were submitted for elemental analysis (N, O, S). Series-2 ended with the migration to a simpler standardized test sequence using the as-received catalysts without dilution, a LHSV of 0.15/hr, a fixed temperature (380°C) and pressure (950psi), and a fixed run time of 100 hours.
The product from the Series-2, Stage-1 reactors was typically very dark and often contained what appeared to be very fine particulate that slowly settled. The product could be clear when the catalyst was relatively new or when the temperature and pressure were increased. However, the quality of the product as indicated by the GC-MS or elemental analysis did not seem to correlate with the visual appearance of the product.
The GC-MS analysis of the Phase-2, Stage-1 reactor products for WFE SC-1 feed shown in Figure 32 illustrate that there were essentially no differences in the principal organic constituents in the reactor products, though the relative concentrations of the individual components did show some variability. These results also show that the reactor feed was almost universally partially hydrogenated. Specifically, while the feed was composed almost entirely of PAHs, the reactor products show varying degrees of ring saturation. However, there is little evidence for ring opening. It is also interesting to note the presence of dibenzofuran in all of the reactor products, illustrating the limited conversion of this specific heteroatom component. Similarly, pyrene was noted as one of the few PAHs present in both the feed and product, suggesting that this component is exceptionally resistant to hydrogenation.
Figure 32. GC-MS analysis of selected products and the WFE SC-1 feed.
The overall results summarized in Table 26 suggest that the most promising candidates from the Phase-2 evaluation were the “I” and “F” catalysts based largely on the GC-MS analysis of the marker components for nitrogen (carbazole) and sulfur (dibenzothiophene) as well as the elemental analysis for sulfur. Only modest reduction in the marker component for oxygen (dibenzofuran) was noted from any of the catalysts. This is consistent with literature that suggests that dibenzofuran is exceptionally stable. The decrease in the density of the product relative to the feed indicates a modest degree of hydrogenation consistent with the GC-MS analysis described above. Note that in the absence of significant ring-opening the lightest product that can be produced from these syncrudes is decalin with a density of 0.895 g/mL.
Table 26. Summary of Stage-1, Series-2 Results
7.6.3 Stage-1, Series-3 - Syncrudes WFE SC-2, 3, and 4
Series 2 showed that relatively aggressive conditions of temperature and pressure would be needed to successfully remove the heteroatom components from the coal-derived syncrudes. It was also desirable to standardize the exposure conditions to make the results more comparable. For these reasons, the exposure conditions were fixed at 380°C and 1250 psi. Furthermore, the Series 3 catalysts were loaded as-received (extrudates) and without dilution. This doubled the amount of catalyst in the reactor, and by leaving the flow rates of feed and hydrogen unchanged the LHSV dropped from 0.3/hr to 0.15/hr without decreasing the production rate of Stage-1 product. It was observed that the hydrotreatment of syncrude is not excessively exothermic and therefore the catalyst can be used catalyst without SiC dilution
Photographs of selected Series-3 products are shown in Figures 33 and 34. The GC-MS total ion chromatograms of selected products are shown in Figure 35. The overall results from the Phase-3 evaluation are summarized in Table 27.
The photographs shown in Figures 33 and 34 illustrate that the product from the Series-3, Stage-1 reactors were relatively clear, though some contained what appeared to be very fine particulate that slowly settled. The fine material could also agglomerate with the water phase, making it somewhat difficult to separate.
Figure 33. Selected products from the processing of the WFE SC-3 syncrude over the “E” NiMo catalyst at 380°C and 1250 psi with a LHSV of 0.15/hr.
Figure 34. Selected products from the processing of the WFE SC-3 syncrude over the “H” NiMo catalyst at 380°C and 1250 psi with a LHSV of 0.15/hr.
The GC-MS analysis of the Series-3, Stage-1 reactor products shown in Figure 38 illustrate that there were essentially no differences in the principal organic constituents in the reactor products after 100 hours of operation with the syncrude, though the relative concentrations of the individual components did show some variability. These results also show that the reactor feed was almost universally partially hydrogenated. Specifically, while the feed was composed almost entirely of PAHs, the reactor products show varying degrees of ring saturation. However, there is little evidence for ring opening after the first few hours of operation with the syncrudes. It is also interesting to note the presence of dibenzofuran in all of the reactor products, illustrating the limited conversion of this specific heteroatom component. Similarly, pyrene was noted as one of the few PAHs present in both the feed and product, suggesting that this component is exceptionally resistant to hydrogenation.
Figure 35. GC-MS analysis of selected products and the WFE SC-3 feed.
The overall results summarized in Table 27 suggest that the most promising candidates from the Series-3 evaluation were the “E” and “H” catalysts. These showed the best reduction in heteroatom content based on the elemental analysis. The GC-MS analysis of the marker components also showed excellent reduction in nitrogen (carbazole) and sulfur (dibenzothiophene). Only modest reduction in the marker component for oxygen (dibenzofuran) was noted from any of the catalysts after 100 hours of operation with the syncrudes. This is consistent with literature that suggests that dibenzofuran is exceptionally stable. The decrease in the density of the product relative to the feed indicates a modest degree of hydrogenation consistent with the GC-MS analysis described above. Note that the greatest reduction in the product density was also found with the “E” and “H” catalysts. Note that in the absence of significant ring-opening the lightest product that can be produced from these syncrudes in decalin with a density of 0.895 g/mL.
Table 27. Summary of Stage-1, Series-3 Results
Note that the products from Series-3 were much lighter in color and clean compared to those from Series-2. The key reason for this is that Series-2 syncrude was made using soybean oil, which did not turn out to be a hydrogen donor, so the resulting syncrude was very heavy. On the other hand, Series-3 syncrude was made using a bio-solvent from the Battelle. CTL process.
7.6.4 Stage-1, Series-4 - Production of Stage-1 Product.
Based on the evaluations described above, the “H” sulfided NiMo catalyst was selected to produce Stage1 product for upgrading in Stage-2. To maximize the heteroatom removal from the Stage-1 process the approach was modified by conducting it in two steps referred to as Stage-1a and Stage-1b. The Stage-1a reactor was operated as described above. However, the elemental analysis of the reactor product as shown in Table 27 suggests that significant amounts of N remain in the product and the level of S is not low enough to permit the use of a noble metal catalyst in Stage-2. Therefore, the product from the Stage-1a reactor was sparged with nitrogen and washed twice with water to remove residual NH3 and H2S. (Sparging was conducted with nitrogen for 1 hour. Washing was with 1:1 HPLS water with gentle swirling and decanting in a separatory funnel.) This washed product was then processed through a Stage-1b reactor to complete the heteroatom removal. The catalysts and processing conditions for both reactors were the same; “H” NiMo catalyst at 380°C, 1250 psi, and with a LHSV of 0.15/hr. The syncrude used as the feed to the Stage-1a reactor was WFE SC-4.
Photographs of selected Stage-1a products are shown in Figure 36. This illustrates that the Stage-1a product is quire dark. These contrast with the product from the earlier evaluation of this catalyst using the WFE SC-3 syncrude (see Figure 34) which were relatively clear. As was previously observed, apparently small changes in the feed can make significant changes in the appearance of the product. Photographs of the Stage-1a and Stage-1b products given in Figure 37 shows that the Stage-1b process gives a relatively clear product.
Figure 36. Selected products from the processing of the WFE SC-4 syncrude over the”H” NiMo catalyst at 380°C and 1250 psi with a LHSV of 0.15/hr (Stage-1a).
Figure 37. The original WFE SC-4 feed and the final Stage-1a and Stage-1b products after they have been sparged and washed.
Table 28. pH of the Stage-1a and 1b Reactor Products
Analysis of the products by GC-MS (see Figure 38) shows that the process of sparging and washing does not have a significant effect on the principal organic components of the product, so the fuel value of the product stream is preserved. This analysis also shows that the Stage-1b process reduces the high molecular weight fraction to a small degree and increases the hydrogenation of the process stream. The extent of hydrogenation was quantified by conducting a hydrocarbon type analysis using comprehensive two-dimensional gas chromatography (GCxGC) as summarized in Table 29. This shows that the overall aromatic content after Stage-1a was approximately 60%, and after Stage-1b 42%. Recalling that the feed is essentially 100% aromatics in the form of PAHs, this data suggests that the that the HDA after Stage-1 was on the order of 40% and after Stage-1b it was on the order of 58%. Furthermore, the GC-MS analysis given in Figure 38 shows that the bulk of the product is heavier than decalin, indicating that the fuel value of the feed is preserved and that the hydrocarbon components of the process stream are well suited for upgrading in Stage-2. Finally, the density of the Stage-1a (0.97) and Stage-1b (0.93) products summarized in Table 30 shows that a reasonable degree of hydrogenation is occurring in the Stage-1 reactors and that the product stream is well suited for upgrading. The H/C ratio increases from 0.89 (feed) to 1.42 (Stage-1, Pass1) and 1.53 (Stage-1, Pass 2)
The choice of the catalyst to be used in Stage-2 will depend on the heteroatom content of the process stream leaving Stage-1. The preferred catalyst in Stage-2 would be a noble metal material such as Pt or Pt/Pd. However, if the heteroatom content of the process stream is not appropriate for a noble metal catalyst, then a hydrogenation catalyst such as a sulfided NiW may be necessary. The elemental analysis of the Stage-1 reactor products summarized in Table 29 and the GC-MS analysis of the heteroatom marker components suggest that the two-step Stage-1 process can indeed produce a product low enough in N and S to permit a noble metal catalyst to be considered for Stage-2.
Figure 38. GC-MS analysis of the WFE SC-4 feed and the Stage-1a and Stage-1b products.
Table 29. Stage-1a & 1b GCxGC Hydrocarbon Type Analysis
Table 30. Summary of Stage-1, Phase-3 Results
Characterization of Feed and Product with 1HNMR. The syncrude and products from Stage-1 were characterized with 1HNMR. Figure 39 is 1HNMR profile for SC-1 (blue) and product from Stage-1 (red) on “D” catalyst at 1250 psi, 380°C, hydrogen/syncrude weight ratio = 3000 and LHSV 0.3. There are two regions: aromatic and aliphatic. The hydrotreatment causes shift of aromatic to aliphatic. The aliphatic/aromatic ratio based on hydrogen is 0.41 in the feed (blue) and 2.5 in the product 1(red). Further treatment of product 1 with “J” catalyst under same conditions increases the ratio to 5.4 (green). Figure 40 is 1HNMR of WFE SC-2 and product over catalyst “J” catalyst under the same conditions. Same phenomena are observed: hydrogenation of aromatic and the aliphatic/aromatic ratio increases from 0.49 to 4.1. Figure 41 is 1HNMR from the hydrotreatment of WFE-4 over “H” catalyst at 1250 psi, 380°C, hydrogen/syncrude weight ratio = 3000 and LHSV 0.15 on two passes as described in Phase 4, Stage-1a. After Stage-1b the aliphatic/aromatic ratio increases considerably from 0.35 to 11.2 showing that changing the reaction condition and specifically LHSV improved considerably the hydrogenation of aromatics.
Figure 39. 1HNMR of SC-1 (blue) and product 1 over “D” catalyst at 1250 psi, LHSV: 0.3 and H/syncrude ratio 3000, and temperature of 380°C (red). Product 2 is processing of the product 1 under the same conditions with “J” catalyst (green).
Figure 40. 1HNMR of WFE-2 (blue) and product over “H” at 1250 psi, LHSV: 0.3 and H/syncrude ratio 3000, and temperature of 380°C (red).
Figure 41. 1HNMR of WFE4 (blue) and product over “H” at 1250 psi, LHSV: 0.1.5 and H/syncrude ratio 3000, and temperature of 380°C second pass (red).
7.6.5 Stage-2 Results
As described above, the function of the Stage-1 reactors is to remove the N, O, and S using a robust, sulfided catalyst to condition the feed for upgrading in Stage-2. This upgrading consists of converting the process stream into a mixture of hydrocarbons suitable for blending with conventional jet turbine fuels at levels up to 50% coal-derived fuel. The general requirements for this product includes that it be within the carbon number range of jet fuel (approximately C8 to C18) and be composed of at least 75% paraffins and no more than 25% aromatics, of which no more that 3% of the aromatics can be di-aromatics (naphthalenes). An overall analysis of the WFE SC-4 feed, the final Stage-1b product, and an example jet turbine fuel (JP-8) shown in Figures 42 and 43 along with the hydrocarbon type analysis summarized in Table 22 shows that the conversion of the coal-derived syncrude to a suitable blending stock has been partially completed during the heteroatom removal in Stage-1. Specifically, the hydrocarbon type analysis summarized in Table 22 shows that the aromatic content has been reduced from approximately 100% in the syncrude feed to 42% in the final Stage-1 product. Furthermore, the carbon distribution summarized in Figure 42 shows that the mass fraction of the syncrude heavier than the jet turbine range was reduced from 61% to 24%. This shows that a modest degree of hydrogenation and cracking is needed in Stage-2, therefore it is desirable to process the heavy fraction above the jet fuel range to bring the composition of the process stream to within the program goals.
Figure 42. GC-MS analysis of the WFE SC-4, Stage-1b product, and an example jet turbine fuel (JP-8).
Figure 43. Carbon distribution of the WFE SC-4 feed, Stage-1a and Stage-1b.
7.6.5.1 Stage-2 Testing with Model Compunds
For Stage-2 noble metal and sulfided catalysts can be used. The initial evaluation of the Nobel metal “K” catalyst was done with Aromatic-200 model as feed at 300-400°C, LHSV of 1/hr, 400 psi hydrogen, and a H:feed ratio of 10:1 mol/mol. Sulfided “J” catalyst was also evaluated with same model feed 300-380°C, 0.3 LHSV , 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v. The overall results of these analyses are summarized in the GC-MS data shown in Figures 44 and 45 for the ”K” and “J”, respectively, and the hydrocarbon type analysis summarized in Table 31. Briefly, it was found that the “K” catalyst was very efficient at hydrogenating the model feed, but showed little cracking. In contrast, the “J” showed a very high degree of cracking to the point that at temperature above 320°C only light, largely incondensable products were produced. As shown in Figure 44, even at 320°C the product showed a very high degree of cracking with a modest degree of hydrogenation. Therefore, 300°C was chosen as the basis for comparing these two catalysts. As shown in Table 31, the “K” catalyst gave a very high degree of hydrogenation with an HDA of 96% as compared with 40% for “J”. This suggests that “K” would be preferred as a hydrogenation catalyst whereas “J” may serve as a cracking catalyst. Therefore, a combination of these two may be effective in processing the Stage-1 product as a combination of hydrogenation and cracking is indicated.
Figure 44. GC-MS analysis of the Aromatic-200 model feed and the Stage-2 products using the “K” catalyst.
Figure 45. GC-MS analysis of the Aromatic-200 model feed and the Stage-2 products using the sulfided “J” catalyst.
Table 31. Model Feed Stage-2 Hydrocarbon Type Analysis at 300°C
7.6.5.2 Initial Stage-2 Testing with Stage-1 Product
The first attempt to upgrade a Stage-1 product was made using the WFE SC-2 syncrude processed through a Stage-1 reactor and the “I” NiMo catalyst. Briefly, all of the product from this reactor was combined to form a composite Stage-1 product. This product was used without any additional cleaning steps described above. This gave a Stage-2 feed with 7400 ppm N and 359 ppm S. The Stage-2 reactor configured with the sulfided “J” catalyst and operated the at 380°C with a LHSV of 0.3/hr, 1250 psi hydrogen, and a H feed ratio of 3000:1 v/v. Figure 46 shows a photograph of the Stage-2 feed and product and the overall results are summarized in the GC-MS data shown in Figure 47 and the hydrocarbon type analysis summarized in Table 32. Briefly, it was found that the “J” catalyst gave a product that was clear and lightly colored. The GC-MS analysis shows that the product was a very complex, partially hydrogenated mixture. The hydrocarbon type analysis also shows a modest degree of hydrogenation with an HDA (based on the Stage-2 reactor feed and product) of 21%.
Figure 46. Photograph of the Stage-1 product (Stage-2 feed) and the Stage-2 product using the sulfided “J” catalyst at 380°C with a LHSV of 0.3/hr, 1250 psi hydrogen, and a H:feed ratio of 3000:1 v/v.
Figure 47. The composite product from the SC-2/catalyst “I” (bottom) that was used as the feed to the Stage-2 reactor using the sulfided “J” catalyst at 380°C with a LHSV of 0.3/hr, 1250 psi hydrogen, and a H:feed ratio of 3000:1 v/v and the product from the Stage-2 reactor (top) at 380°C, 1250 psi hydrogen and 0.3/hr LHSV.
Table 32. Stage-2 Hydrocarbon Type Analysis at 300°C, 1250 psi and 0.3/hr with a “J” Catalyst
The second attempt to upgrade a Stage-1 product was made using the WFE SC-1 syncrude processed through a Stage-1 reactor and the “I” NiMo catalyst. Briefly, all of the products from this reactor was combined to form a composite Stage-1 product. This product was then sparged with nitrogen for 1 hour, but it was not washed with water as was later adopted for the Stage-1 product. This gave a Stage-2 feed with <5000 ppm N and 167 ppm S. The Stage-2 reactor was configured with the “K” catalyst and was operated the at 300-360°C with a LHSV of 0.3/hr, 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v. Figure 48 shows a photograph of the Zone-2 feed and product and the overall results are summarized in the GC-MS data shown in Figure 49 and the hydrocarbon type analysis summarized in Table 33. Briefly, it was found that the “K” catalyst gave a product that was surprisingly dark. However, as noted previously, the visual appearance often does not correlate with the product quality. The GC-MS analysis shows that the product was a very complex, partially hydrogenated mixture. The hydrocarbon type analysis also shows a modest degree of hydrogenation with an HDA (based on the Stage-2 reactor feed and product) of 22% was observed at 300°C, though this increased to 45% at 360°C. Recall that the refined Stage-1 product described above had an aromatic content of 42%. Reducing this to 25% to meet the goal for a blending stock would require an HDA on the order of 40%, which compares well with the HDA measured for the “K” catalyst. This suggests that a noble metal catalyst should be capable of upgrading the refined Stage-1 product to a suitable blending stock, though some cracking may be needed to reduce the carbon distribution to within the jet-fuel range.
Figure 48. Photograph of the Stage-1 product (Stage-2 feed) and the Stage-2 product using the “K” catalyst at 300-360°C with a LHSV of 0.3/hr, 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v.
Figure 49. The composite product from the WFE SC-1/ “I” catalyst (bottom) that was used as the feed to the Stage-2 reactor using “K” catalyst at 300-360°C with a LHSV of 0.3/hr, 950 psi hydrogen, and a H:feed ratio of 3000:1 v/v and the product from the Stage-2 reactor at 300°C (middle) and 360°C (top).
Table 33. Stage-2 Hydrocarbon Type Analysis at 950psi and 0.3/hr with “K” Catalyst
7.6.5.3 Hydrocracking of Stage-1 Product from Intertek Run #1 Over Proprietary Bi-functional Catalyst
One L of Stage-1 product from Intertek Run #1 (hydrotreated syncrude) was sent to UDRI for further treatment at 1250 psi, LHSV = 0.3 hr-1, Hydrogen to liquid volume ratio = 1000. The test was performed at two temperatures (320ºC and 340ºC) over a proprietary, NiW, bi-functional catalyst, obtained by Battelle. The temperature was increased from 320°C to 340°C at 48.4 hours TOS. The sample taken at 9.4 hrs was clear and lightly colored with some sediment. The sample taken at 19.0 hrs was clear and somewhat darker, with significantly more sediment. The samples taken from 27.0 to 65.6 hrs were clear, without any significant sediment. The samples taken at 70.9 and 77.2 hrs appeared dark and possibly cloudy. Some light, non-condensable product was noted in all of samples.
The volumetric product yield was between 90 wt% and 97 wt%, the balance being light hydrocarbons. Effectively, the online GC analysis (not described in this report) showed formation of C1 through C6 products. The density dropped significantly from 0.95 g/cm3 to a value below 0.86 g/cm3, but it increased with TOS. Increasing the temperature from 320ºC to 340°C dropped the density from 0.85 g/cm3 to 0.82 g/cm3, but again started increasing with TOS. Two explanations are possible:
- Since the run was performed for less than 80 hrs (TOS), the catalyst surface structure may still have been adjusting to the reaction conditions and it did not reach steady state.
- There was progressive carbon deposition on the catalyst and it continued to deactivate.
The GCxGC class fraction analysis (see Table 34) shows that hydrocracking of Stage-1 product significantly reduced the amount of total aromatic from 33 wt% to 15 wt%, which is in the range of commercial jet fuel. The tri- and multi-cycloparaffins were converted to mono- and di-cycloparaffins, which indicated that the catalyst was active in cleavage of the C-C bond.
Table 34. GCxGC Analysis Summary for Lab-scale Hydrocracking of Stage-1 Product from Intertek Run #1
Based on the results of laboratory-scale catalyst screening, the following catalysts are recommended for pre-pilot testing:
- Stage 1. Catalyst “I” (NiMo) is recommended in a two-pass configuration. After Pass 1, the NH3 and H2S are separated from the liquid and the liquid is then re-treated in Pass 2.
- Stage 2. It is recommended that both a noble-metal catalyst similar to Catalyst “K’ and a bi-functional catalyst similar to Catalyst “J” be tested. The first one is a better hydrogeneration catalyst, so a very low total aromatic content can be achieved, while the second one is a better cracking catalyst, which leads to a higher aromatic content.
8.0 PRE-PILOT SCALE: SYNCRUDE HYDROTREATMENT AND HYDROGENATION TO JET FUEL/DIESEL
8.1 Objectives
The objectives of this effort were as follows:
- Demonstrate that syncrude from the Battelle CTL process can be upgraded to jet fuel and diesel by hydrotreatment/hydrogenation and distillation, at pre-pilot scale
- Characterize the upgraded product before and after distilling into various fuel cuts
- Prepare a jet fuel fraction and a diesel fraction for detailed analysis by UDRI and others.
Battelle selected two syncrudes that where produced from Quantex Run #5 and #6B for conducting two tests at Intertek for upgrading these to distillate products. The details of the liquefaction tests were described in Section 6.0. The ultimate analyses of these two feedstocks are provided in Table 35. The key difference between the two is that the bio-solvent for Syncrude 6B had a higher H/C ratio and its density was lower.
Table 35. Ultimate Analysis of Syncrudes Used for Intertek Run #1 (Syncrude 5)
8.3 Standard Operating Procedures for Handling and Hydrotreatment of Syncrude
Before shipment of syncrude to Intertek for hydrotreatment, a Standard Operating Procedures (SOPs) critical to the safe hydrogenation operations was discussed with Intertek personnel. Some of the SOPs discussed include Safety, Chemical Hygiene, Spill Response, Fire Safety, First Aid, Lockout/Tagout, PPE, Work Place Air Monitoring, and Emergency Shutdown Procedures. In addition, controls to mitigate potential syncrude hydrogenation safety hazards were identified and added to the Test Plan and Environmental Health and Safety Plan.
8.4 Description of Pre-Pilot-Scale Reactors
Hydrotreatment operations were conducted at Intertek in two pilot plant units: P63 and P67. Each unit is composed of two trickle-bed reactors assembled in series. Figure 50 is photo of the P63 which is similar to P67. Each trickle reactor can hold up to 2.5 L of catalyst and it is composed of 6 heating and quenching zones. The temperature of the reactors is monitored at different zones with several thermocouples (TC). Figure 51 shows the catalyst loading and TC locations for Reactor #1 and Reactor #2.
Figure 50. Intertek’s continuous syncrude hydrotreatment system P63 unit.
Figure 51. Thermocouples (TCs) location and reactor loading. In Reactor #1 there are 4 catalyst beds each with 3 TCs. Reactor#2 has one catalyst bed with 5 TC.
8.5 Pretreatment of Syncrude 5
Battelle shipped 162 kg of syncrude 5 to Intertek. The syncrude was filtered via 1- micron filter at 65°C to remove any residual material left from its preparation at Quantex. The filtered syncrude was loaded into a Fractioneer distillation unit, but the unit plugged and was not able to achieve the desired 500°C cut point. The bottoms material from the Fractioneer unit was therefore transferred to a smaller, single-stage Sarnia unit to achieve the desired 500°C cut point. Distillate from the Sarnia and the Fractioneer were combined and filtered at 65°C and 0.5 micron. About 136 kg of distillate was recovered after filtration and was then hydrotreated.
8.6 Hydrotreatment of Syncrude 5 (Intertek Run #1)
As discussed in Section 7.0, the upgrading was performed in two stages: hydrotreatment, followed by hydrogenation. The details for each stage are provided below.
8.6.1 Stage-1 Reaction (Run #1)
During the first stage, heteroatoms, specifically sulfur, nitrogen and oxygen, are removed by hydrodesulfurization (HDS), hydrodeoxygenation (HDO) and hydrodenitrogenation (HDN). These reactions take place over a sulfided catalyst at a temperature of 380ºC, Liquid Hour Space Velocity (LHSV) of 0.15 hr-1, a pressure of 1300psi and a hydrogen/syncrude volume ratio of 1000. Table 36 provides the reaction conditions used. Figure 52 displays the temperature profile of the reactors during four TOS (time on stream) periods during hydrotreatment. It can be divided into four sections with respect to time on stream.
- 0< TOS<100hrs: Initiation step where the catalyst was not at steady state and the temperature profile was not uniform due to the high activity of fresh catalyst and the exothermic hydrogenation reaction.
- 100<TOS<130hrs: The catalyst was at steady state. The temperatures were uniform throughout the reactors.
- 130<TOS<160hrs: An external problem due to a malfunction in the hydrogen compressor forced a shut down. The operator decreased the temperature to protect the catalyst and resumed the reaction after fixing the compressor.
- 160 <TOS<230hrs: The system was restarted and the remaining feed was processed.
- 230<TOS<300 Select material was reprocessed to ensure adequate sulfur and nitrogen removal.
Table 36. Reaction Conditions for Stage-1 (Same Catalyst in Reactors #1 & #2)
The analysis of Stage-1 product showed a substantial (>99.9%) removal of S and N, with the product values being 2 ppm and 7 ppm, respectively.
Figure 52. Reactor #1 (R1) and Reactor #2 (R2) temperature profile.
8.6.2. Stage 2 Reaction (Run #1)
The goals of this Stage-2 (hydrogenation) reaction was to reduce the concentration of aromatics present in syncrude from approximately 100 wt% to less than 20 wt% and to reduce the density from ~0.98 g/cm3 to less than 0.90 g/cm3. This reaction was carried out in unit P63, using a proprietary noble-metal catalyst. This catalyst was tested earlier this year at UDRI with syncrude and with model compounds, as discussed in Section 7.6.5. It has a good hydrogenation activity at temperatures between 200ºC and 360°C. Therefore, we conducted the reaction at low temperatures (240°C) and at a relatively high Liquid Hourly Space Velocity (LHSV) of 0.6 hr-1, as compared to hydrotreatment reaction LHSV of 0.15 hr-1at a pressure of 1300 psi and a hydrogen/syncrude volume ratio of 1,000. Around 32 gallons of syncrude was processed over 2.2 L catalyst volume, filled in Reactor #1 of the P63 unit while bypassing Reactor #2. Table 37 summarizes of reaction conditions. The temperature profile of Reactor #1 is displayed in Figure 53. The target temperature was 240°C, but given the exothermicity of the reaction and short time on stream (TOS) of 90 hrs., the temperature varied between 220°C and 260°C.
Table 37. Reaction Conditions for Stage 2 (Hydrogenation of Hydrotreated Syncrude)
Figure 53. Temperature profile of Reactor #1 (P63 unit) during hydrogenation of hydrotreated syncrude produced in syncrude 5.
8.7. Analysis of Feed, Stage-1 and Stage-2 Products from Intertek Run #1
8.7.1 Feed and Product
Figure 54 shows photos of feed and products from the two stages of upgrading. The hydrotreatment converted a heavy crude to refined liquids. The final product was a yellow, transparent liquid, while the feed was composed of mostly heavy, aromatic compounds.
Figure 54. Photos of feed and products from Stage-1 and Stage-2 for Run #1.
8.7.2 Ultimate Analysis of Stage-1 and Stage-2 Products
The elemental analyses of the feed and products from Stage-1 (hydrotreatment) and Stage-2 (hydrogenation) are shown in Table 38.
Table 38. Elemental Characteristic of Feed, Stage-1 and Stage-2 Products
The analyses show the following:
- The H/C atomic ratio increased from 0.83 (feed) to 1.5 (Stage-1) and 1.7 (Stage-2). Both catalysts thus have good hydrogenation activity.
- The ash content of the feed to the reactor was <0.08 wt, compared to 0.14 wt% in Syncrude 5. This indicates that our pretreatment of feed by distillation and filtration removed most of the ash that can plug and/or poison the catalyst.
- The hydrotreatment catalyst (Stage 1) dropped the S and N from 5,800 ppm and 4,760 ppm, respectively, to less than 2 ppm and 7 ppm, respectively. The oxygen concentration dropped below detection limit for the instrument (Thermo Finnigan Flash EA).
- The density dropped from 1.13 g/cm3 to .97 g/cm3 in Stage-1 and to 0.93 g/cm3 in Stage-2, which indicates that both catalysts have moderate hydrocracking activity.
The level of hydrogenation of aromatics in Stage-2 was determined by HPLC. As shown in Figure 55, the aromatics content dropped by 96%. The active-metal catalyst used in Stage-2 is known to be superactive; it completely hydrogenated the double bounds in the aromatics present after Stage-1 reaction.
Figure 55. Concentration of mono, di, and tri aromatics in Stage-1 and Stage-2 products as determined by HPLC.
8.7.4 Simulated Distillation of Feed, Stage-1 and-2 Products (Run #1)
Figure 56 shows the simulated distillation of feed as well as Stage-1 and Stage-2 products. It shows that the hydrotreatment in Stage-1 dropped the boiling point significantly. This result is explained by the cleavage of C- heteroatom bonds. In Stage-2, there is only the hydrogenation of double bonds which explains the only slight change in the simulated distillation curve.
Figure 56. Simulated distillation of feed (syncrude), Stage-1 (hydrotreated syncrude) and Stage-2 (hydrogenated Stage-1 product) for Run #1.
8.7.5 Comparison of Batch and Pre-Pilot Results
The results of laboratory testing were compared with the results from pre-pilot testing, with respect to sulfur and nitrogen removal. As shown in Figure 57, for any given LHSV value, the sulfur and nitrogen removal was better in pre-pilot testing. This is likely because the longer catalyst-bed height in the pre-pilot test improved the gas-solid contact.
Figure 57. Residual “S” and “N” at different LHSV in lab & pilot scale reactors.
8.8 Fractionation of the Stage-2 Product (Run #1)
Around 32 gallons of distillate was produced by hydrogenation of hydrotreated syncrude. It was then divided into three portions:
- Portion #1: 8 gallons was distillated at Intertek to produce two distillate fractions: a fraction boiling below 337°C (called diesel fraction) and a fraction boiling above 337°C.
- Portion #2: 8 gallons was distillated at Intertek to produce two fractions: a fraction boiling below 295ºC (called jet-fuel fraction) and a fraction boiling above 295ºC.
- Portion #3: 16 gallons was left without distillation for further study, if necessary.
Table 39. Distillation of Stage-2-Product (Intertek Run #1)
8.9 Hydrotreatment of Syncrude 6B (Intertek Run #2)
A total of 92.3 kg of Syncrude 6B was hydrotreated in the P67 pre-pilot reactor, for Intertek Run #2. The unit details and operating procedures were reported in section 8.3. The first reactor was loaded with a sulfided catalyst and second reactor was loaded with a proprietary, bi-functional hydrocracking catalyst. During hydrotreatment, NH3 is produced, which may negatively impact the performance of hydrocracking catalyst. Therefore, the upgrading was performed in two stages with intermediate N2 purge at atmospheric pressure to minimize the amount of NH3 and H2S, as described below. In Stage-1, heteroatom removal was achieved using Reactor #1 while Reactor #2 was in bypass mode (under hydrogen atmosphere at room temperature), as indicated in Figure 58A. The reaction conditions are reported in Table 40. The Stage-2 operation was performed using two reactors: Reactor #1 to further reduce the heteroatoms, followed by hydrocracking in Reactor #2 (Figure 58B). The general reaction conditions are reported in Table 41.
Figure 58. Run #2 feed path for Stage-1 (hydrotreatment) and Stage-2 (hydrocracking).
Table 40. Stage-1 General Reaction Conditions (Reactor #2 in Bypass Mode) for Run #2
Table 41. Stage-2 General Reaction Conditions for Run #2
The Stage-1 reaction was carried out in Reactor #1 during the first 112 hrs. The temperature profile can be divided in two sections: first 30 hrs. heat up period and the period between 30 and 180 hrs where the reactor reached the target temperature (380ºC). The syncrude was injected at temperature around 150ºC. At a total on-stream time (TOS) between 40 and 48 hrs, a hydrogen leak occurred in the lab, which required shut down of the system and resumption of the reaction after controlling the leak. Therefore, the initial, off-spec material was reprocessed. The temperature profile is shown in Figure 59.
Figure 59. Temperature profile for Stage-1 and Stage-2 of Run #2.
The liquid from Stage-1 was processed in Reactor #1 (380°C) and Reactor #2 (320°C) at TOS between 180 hrs. and 300 hrs. (blue zone in Figure 11). The target temperature for Stage-2 was 320°C. However, given the limited amount of feed and the high space velocity (0.4 hr-1) we were not able to adjust the temperature, so we decided to run it at a slightly higher temperature (360°C).
8.10 Analysis of Syncrude and Distillate Products for Run #2
Table 42 summarizes the ultimate analysis of the feed and products from Stages-1 and -2. As shown, the H/C mole ratio increased by 50% in Stage-1 and by an additional 20% in Stage-2. Both sulfur and nitrogen dropped by more than 98% in Stage-1 and by more than 90% of the remaining portion in Stage-2. The density decreased by 8% in Stage-1 and by an additional 8% in Stage-2. The liquid wt% in Table 42 represents the mass recovery from the process.
Table 42. Ultimate Analysis of Feed (Syncrude), Stage-1 Product, and Stage-2 Product
Table 43 shows the aromatic contents of the syncrude, Stage-1 product, and Stage-2 product. The di-, tri-, and total aromatic concentration decreased with upgrading in Stage-1 and Stage-2. The tri-aromatics and di-aromatic were converted to mono-aromatics. A large portion of the total aromatics was converted to cycloparaffins.
Table 44 summarizes the GCxGC-MS of Stage-1 and Stage-2 products. The Stage-2 catalyst dropped aromatics concentration by at least 60%, increased the cycloparaffins by 100%, and reduced the paraffins by 14%. It looks like the hydrocracking catalyst has good activity in reducing double bonds but limited activity for the cleavage of C-C bonds.
Figure 60 provides the Simulated-Distillation curves of feed, Stage-1 product, and Stage-2 product. The amount of jet fuel (fraction below 300oC) in the feed is 20%, but increases to 50% and to 70% after Stage-1 and Stage-2, respectively.
Table 44. GCxGC-MS for Stage-1 Product and Stage-2 Product (wt%) for Run #1
Figure 60. Simulated distillation curves for feed, Stage-1, and Stage-2 for Run #2.
8.11 Fractionation of the Distillate from Run #2
A total of 27.3 kg of Stage-2 product was batch distilled to obtain cuts representative of jet fuel products. The material was first cut at 205°C at atmospheric pressure to prevent loss of the light fraction. Afterwards we dropped the temperature to 100ºC and then the unit was brought down to 100 mm Hg absolute pressure. Then, 1-liter volume cuts were taken until the overhead vapor temperature reached 217°C at 100 mm Hg, which is equivalent to 295ºC at atmospheric pressure. A total of 14 (1-liter) cuts were taken in this way. Table 45 summarizes the performed cuts. Figure 61 is a photo of jet fuel fraction and the still pot liquid fraction (residue) with boiling point above 295ºC.
To prepare a jet-fuel fraction that meets flash point specifications, we removed by distillation the light fraction that has boiling point below 160°C. This light fraction corresponds to 4.5 percent volume of the liquid product. A synthetic jet fuel blend was thus prepared using 65% volume from the upper boiling range from cut #1 along with 100% each from cut #2 through cut #7. A detailed characterization of this “neat” jet-fuel product along with a blend with a commercially-used jet fuel is provided in Section 9.1.
The synthetic diesel fraction was prepared by removing the 35% volume from the lower-boiling range from cut #1 along with 100% each of all other cuts. A detailed characterization of this “neat” diesel product is provided in Sec 9.2.
Table 45. Distillation of Stage-2 Product (Jet Fuel Fraction Cut below 295°C)
Figure 61. Photo of jet fuel fraction and still pot (residue) fraction for Run #2.
8.12 Comparison of Distillates from Runs #1 and #2
Due to differences in the syncrude properties as well as due to differences in the Stage-2 catalysts, the two upgrading runs produced distillates that had significantly different properties. The results for jet fuel fractions for Run #1 and Run #2 are shown in Table 46. As shown, the jet fuel produced in Run #2 has better properties than jet fuel produced in Run #1: lower freezing point; lower viscosity; lower pressure drop cross the filler; and slightly lower density. It has significantly higher aromatic and n-paraffin content with a concurrent reduction in tri-cycloparaffinic composition. Based on the GCxGC analysis (Table 47), it appears that there is less severe hydrogenation of the Stage-2 product. Although the total aromatic content of Run #2 jet fuel (~17%) is similar to conventional jet fuels, there is an appreciable quantity of 2- and 3-ring aromatic compounds, which is higher than in typical aviation fuels. The distillation step for the Run #2 jet fuel resulted in an improved distillation profile and carbon number distribution as compared to the Run #1 jet fuel sample (Figure 62 and Table 48). The higher concentration of lower molecular weight compounds (e.g., < C9) in Run #2 helped reduce the flash point to 38.9°C from 61.1°C in Run #1.
Table 46. Selected Jet Fuel Specification Tests Performed for Jet Fuel Fractions from Run #1 and Run #2
Table 47. GCxGC Run #1 vs. Run #2 Product (Stage-2 Product)
Figure 62. Simulated distillation of jet fuel fraction (≤295°C) for Run #1 and Run #2.
Table 48. Jet Fuel Carbon Distribution for Run #1 and Run #2 (wt%)
A comparison of diesel fractions from Runs #1 and #2 is provided in Table 49. A more detail analysis of the diesel product from Run #2, which was significantly better than from Run #1, is provided in Section 9.2.
Table 49. Comparison of Diesel Cuts from Run #1 and Run #2 Compared to ASTM 975 Specifications
9.0 CHARACTERIZATION OF DISTILLATE – PRODUCT FRACTIONS AS JET FUEL OR DIESEL
The distillate product from Intertek Run #2 was fractionated into a jet-fuel cut as well as to a diesel cut. The jet-fuel cut was analyzed extensively because jet fuel was the primary focus of this project. However, standard specification testing also was done to analyze the diesel fraction.
9.1 Evaluation of Battelle-CTL-Derived Neat Synthetic Fuel and Synthetic Fuel Blend for Use in Aviation Applications
9.1.1 Introduction
In recent years, significant efforts have been made to develop, evaluate and certify synthetic (e.g., non-petroleum derived) jet fuels for use in commercial and military aircraft. Initial focus was related to the approval of synthetic formulations which could be blended with conventional fuels for use. These efforts resulted in the certification and approval of several types of synthetic fuels as blending feedstocks in commercial and military aviation fuels. Commercial Jet A (ASTM D7566) and military JP-8 (MIL-DTL-83133J) fuel specifications detail requirements for both the currently approved synthetic blending feedstocks and the resulting fuel blends [13-14]. The properties of the synthetic fuel blends must conform to those required for typical petroleum-derived fuels. In addition, each synthetic blending feedstock has specific property requirements included in appendices of the fuel specifications. These latter requirements provide confidence that the fuel blend will conform to all operational and safety needs of current aircraft fuel systems and engines and insure process quality control during production.
Guidance on recommended protocols and methodologies for the evaluation and certification of synthetic aviation fuels and fuel additives was formalized and documented in ASTM D4054-09 and MIL-HDBK-510A. These protocols were developed to facilitate the approval process in a time- and cost-effective manner. The overall process is divided into ‘Tiers’; the proposed test program from ASTM D4054 is shown in Figure 63 [15]. Initial tiers focus on evaluation of chemical and physical properties of the fuel candidate to determine potential suitability prior to larger-scale turbine hot section and component/system-level testing. Fuel specification testing (referred to as Tier 1) is initially performed to determine conformity of the synthetic fuel candidate to physical property specification requirements. Upon determination of acceptable property conformance, more detailed testing (referred to as Tier 2) is performed to evaluate select “Fit-for-Purpose” (FFP) properties. FFP refers to a property required for safe operation which is not directly controlled by the respective fuel specification; a petroleum-derived fuel which meets all fuel specification requirements will inherently satisfy all FFP property needs. The FFP evaluations include several temperature-dependent properties (e.g., density, specific heat, thermal conductivity). Evaluation of all fuel specification and FFP properties outlined in ASTM D4054 and MIL-HDBK-510 requires several gallons of the neat synthetic fuel and blending feedstock [16].
Battelle produced a synthetic aviation fuel candidate from its CTL process based on the use of biomass-derived solvents. The ‘final’ synthetic jet fuel produced by Battelle was provided to the UDRI for preliminary evaluation of the suitability for use as either a neat ‘drop-in’ jet fuel or synthetic bending component. The evaluation approach was based on guidance in the aforementioned certification protocols; however, it was not possible to complete all required fuel specification and FFP property evaluations for the neat synthetic fuel and blend due to insufficient total available volume. Therefore, fuel specification and select FFP properties were evaluated to provide a quantitative basis for preliminary evaluation of the potential suitability of the submitted synthetic fuel for use in aircraft systems. The following sections discuss the analyses and results performed on the neat synthetic fuel and fuel blend.
* Testing must be performed at P&W, GE, Rolls Royce, Snecma, Honeywell or in other locations per OEM agreement due to proprietary concerns and test methods. Note 1-Additive testing to be performed at 4x the concentration being requested for approval except for filtration.
Figure 63. Proposed test program for qualification and approval of new aviation turbine fuels and fuel additives (Ref: ASTM D4054-16).
9.1.2 Evaluation of Neat Synthetic Fuel Formulation
The ‘final’ neat synthetic fuel in the DOE program was produced at Intertek Laboratories in a pilot scale unit using Battelle catalysts and processing conditions and was given the internal designation “54486-38-22”. It was the composite of distillation fractions of cut #1 (65% heavy fraction) trhough cut #7 (see Table 45). The synthetic fuel was analyzed for selected specification (and additional elemental analysis) properties using ASTM test methods by Intertek and analyzed for Hydrocarbon-type (HC-type) composition and carbon number distribution by UDRI. HC-type and carbon number analyses were performed using Two-Dimensional Gas Chromatography (GCxGC) with simultaneous Mass Spectrometry (MS) for species identification and Flame Ionization Detection (FID) for quantitation.
The specific ASTM tests performed were selected to provide guidance regarding suitability of the neat synthetic fuel as a direct ‘drop-in’ aviation fuel (i.e., blending with petroleum-derived fuel not necessary). The results from these ASTM tests are shown in Table 50. The HC-type composition is shown in Table 51 and the carbon number distribution is shown in Table 52. The analyses of the neat synthetic fuel indicated that it was not feasible to use this formulation directly as a ‘drop-in’ synthetic fuel; several properties did not conform to the required Jet A/A-1 and JP-8 property requirements. Specifically, the density (0.885 g/mL) was greater than the maximum specification requirement for aviation fuels (0.840 g/mL), the smoke point (17 mm) did not satisfy the sooting requirement (25 mm), and the hydrogen content (12.48% by mass) was below the military fuel requirement (13.4%).
With respect to other properties, the low temperature viscosity (7.3 cSt) was near the specification requirement (8.0 cSt), but the fuel had an acceptable measured freeze point (< -60 C). Nitrogen and trace elemental analyses showed the fuel had minimal residual inorganics from the production and upgrading processes employed; additional sulfur and nitrogen analyses were performed using non-ASTM test methods (reported in Section 4.2.2/Table 9) and indicated very low heteroatomic content in the synthetic fuel. These results are notable based on the elemental composition of the feedstocks used to produce this synthetic fuel. The distillation profile was similar to conventional aviation fuels; with a T90-T10 of 69.2 C. This range is significantly higher than the current synthetic fuel blending feedstock minimum limit of 22 C, which was implemented to insure a blending feedstock would not result in a significant discontinuity in the fuel distillation profile. The flash point was higher than typical Jet A/A-1 fuels due to the lower concentration of low molecular weight species. However, as discussed in Section 8.11, this can be brought down closer to typical Jet A/A-1 fuel values by not removing some of the <160°C fraction.
The thermal-oxidative stability of the neat synthetic fuel was evaluated via testing with the Jet Fuel Thermal Oxidation Tester (JFTOT) with a test temperature of 325 C. This test condition is very aggressive and has been used for evaluation of previously approved synthetic aviation fuel blending feedstocks; a satisfactory result is an indication that the synthetic fuel should not detrimentally affect the thermal stability characteristics of the fuel blend (per ASTM D7566). The coal-derived synthetic fuel did not pass the JFTOT requirements under this test condition (details regarding potential causes for this behavior will be discussed in Section 9.1.3.1.5). However, this result does not preclude the possible use of this synthetic fuel as a blending feedstock as the specification test temperature for a synthetic fuel blend is 260 C.
Table 50. Select Jet Fuel Specification and Elemental Analyses for Neat Synthetic Fuel
Table 51. Hydrocarbon Type Analysis of Neat Synthetic Fuel by GCxGC (Intertek Run #2)
Table 52. Carbon Number Distribution of Neat Synthetic Fuel (Intertek Run #2)
The HC-type analysis shown in Table 51 indicted the fuel had a total concentration of aromatics (~18.1% by mass) similar to typical aviation fuels; however, there was a higher quantity of 2- and 3-ring cycloaromatics (i.e., partially hydrogenated) than in typical aviation fuels. Petroleum-derived aviation fuels typical contain appreciable concentrations of iso- (~30-40%) and n-paraffin (~20-25%); however, the neat synthetic fuel had very low concentrations of branched and linear paraffins (approximately 0.7 and 2.2%, respectively). The synthetic fuel was primarily comprised of cycloparaffins (79% by mass), with a very high concentration of di- and tri-cycloparaffins (51.1% and 15.2%, respectively). Petroleum-derived aviation fuels typically contain ~30% cycloparaffins, which are primarily monocycloparaffins. The larger molecular weight cycloparaffins in the synthetic fuel were most likely produced directly from the coal feedstock via fragmentation and hydrogenation of the high molecular weight coal moieties. Likewise, the cycloaromatics in the neat synthetic fuel were produced via incomplete hydrogenation of the upgraded feedstocks. The high concentrations of the di-/tri-cycloparaffins and cycloaromatics are the primary cause for the high fuel density and most likely the reduced smoke point value (e.g., increased sooting tendency).
9.1.2.1 Estimation of Blending Ratio for Synthetic Fuel
Based on properties and compositional analysis of the neat synthetic fuel, it was determined that the viable approach for use in aviation applications was as a blending feedstock with petroleum-derived aviation fuel. Therefore, determination of optimal and maximum blending ratios is necessary to allow detailed evaluation of the resulting specification and selected FFP properties of the synthetic fuel blend. The target blend ratio must be specified such that the synthetic fuel blend conforms to all specification requirements identified for commercial (ASTM D7566/D1655) and military (MIL-DTL-83133J) aviation fuels. The maximum allowable blend concentration is dependent on the specific chemistry and properties of the synthetic fuel. For example, Synthetic Paraffinic Kerosenes (SPKs) derived from either Fischer-Tropsch (FT) Synthesis or Hydoprocessed Esters and Fatty Acids (HEFAs), which are primarily comprised of iso- and n-paraffins, can be blended at concentrations up to 50% by volume provided all specification requirements are satisfied. Depending on the properties of the fuel to which these are blended, the density or aromatic content (or both) of the refined fuel may limit the amount of SPK that can be added to the final blend to less than 50% (per D7566). On the contrary, iso-paraffinic compounds produced via oligomerization of iso-butanol (termer Alcohol-to-Jet Synthetic Paraffinic Kerosene [ATJ-SPK]) can only be blended to a maximum concentration of 30% by volume for use.
Based on the specification and compositional properties shown in Tables 1 and 2, the high density of the neat synthetic fuel (0.885 g/mL) is the primary property limiting the maximum blending ratio for application. Specifically, the synthetic fuel blend must have a final density ≤ 0.840 g/mL for use. Improved thermal stability and smoke point are also expected upon blending (due to dilution). The density (@ +15 C) of typical aviation fuels range from approximately 0.780 to 0.830 g/mL, with an average of approximately 0.803 g/mL. The resulting density of a synthetic and petroleum-derived aviation fuel blend is expected to be linear with the volumetric blending ratio as the fluids are expected to behave as ideal fluids; this has previously been shown during blending of FT-derived SPK with aviation fuels [17]. Calculations were performed to investigate the effect of blending ratio and petroleum-derived fuel density on the resulting synthetic fuel blend density. Figure 64 shows the estimated density values of the fuel blend as a function of blending ratio with aviation fuels with densities of 0.780, 0.803 and 0.830 g/mL, respectively. As shown, the primary limiting factor occurs when the petroleum-derived fuel has a high density (e.g., maximum blend percentage of ~18% by volume when petroleum fuel has a density of 0.830 g/mL). However, fuels with lower density values (which comprise the majority of aviation fuels) allow higher blend percentages with the neat synthetic fuel. Based on the density trends shown in Figure 2, the recommendation was made to evaluate the corresponding specification and FFP properties for a 20% blend by volume of the synthetic fuel with an ‘average’ (or ‘nominal’) jet fuel. This provides guidance on the expected performance when using the synthetic fuel as blending feedstock and will most likely mitigate potential issues related to the concentration of certain compound classes which are higher than typically present in aviation fuels (e.g., di-/tri-cycloparaffins).
Figure 64. Effect of volumetric blending ratio and corresponding density of petroleum-derived fuel on calculated density for synthetic fuel blend.
Evaluation of the specification and selected Fit-For-Purpose properties of a 20% blend by volume of the synthetic fuel with a petroleum-derived aviation fuel was performed to provide preliminary guidance regarding the potential suitability for use in aviation applications. Subsequent testing and evaluation as outlined in ASTM D4054 and MIL-HDBK-510 is necessary to provide sufficient data to determine suitability for pursuing certification and approval of the synthetic fuel as a blending feedstock. Blending was performed using a Jet A fuel provided by the Fuels & Energy Branch of the Air Force Research Laboratory. This specific Jet A (with internal identification POSF 10325) has been termed an ‘average’ (or ‘nominal’) jet fuel as the specification properties are very close to historical averages for aviation fuels. This specific Jet A has been used as the ‘average’ fuel in the Federal Aviation Administration (FAA) National Jet Fuel Combustion Program (NJFCP). Results obtained for the synthetic fuel blend will be compared to those for the neat Jet A and are discussed in the following sections.
Due to the limited quantity of batch 54486-38-22, Battelle prepared a “second batch” of neat synthetic fuel from the same process streams; this batch of neat synthetic fuel was given the internal designation “54486-39-16”. The composition of 54486-39-16 was compared to the initial batch (54486-38-22) to determine suitability for testing. HC-type composition and carbon number distribution analyses (shown in Section 4.1) indicated the compositions of the two batches were sufficiently similar to proceed with blend preparation and preliminary evaluation for this program.
Battelle had a limited overall volume of the neat synthetic fuel 54486-39-16 available for blend preparation and testing, which precluded completion of all recommended Tier 1 and 2 testing and evaluations in this effort. Therefore, the decision was made to evaluate all fuel property specification requirements; specific FFP evaluations were selected which would provide detailed insight regarding the suitability of the blend for use. The 20% blend by volume of the neat synthetic fuel (54486-39-16) with the ‘nominal’ jet fuel (POSF 10325) was given the internal designation “54486-39-26”. Results from these tests and pertinent discussion are included in the following sections.
9.1.3.1 Chemical Composition and Specification Properties
The chemical composition and aviation fuel ASTM specification fuel properties were evaluated for the 20% by volume synthetic fuel blend. HC-type analysis obtained using GCxGC analysis and ASTM D2425/6379 are shown in Tables 53 and 54 (a more detailed summary of HC-type analysis is included in appendix of report), carbon number distributions are shown in Table 55, and ASTM specification properties are presented in Table 56. Results for the Jet A used for blending and the neat synthetic fuel(s) are included for comparison.
Table 53. Comparison of Hydrocarbon Type Analyses of Fuels by GCxGC
Table 54. Comparison of Hydrocarbon Type Analyses of Fuels via ASTM D2425/D6379
Table 55. Carbon Number Distributions of Fuels
Table 56. Jet Fuel Specification Test Results of Fuels
9.1.3.1.1 Composition
Blending of the neat synthetic fuel at 20% by volume with petroleum-derived Jet A resulted in an overall chemical composition similar to typical aviation fuels, with the exception of the cycloaromatic and di-/tri-cycloparaffins content. The concentration of these latter compound classes is higher than typically observed in aviation fuels. This result was expected based on the coal feedstock used for production of the synthetic blend feedstock. Differences were observed between hydrocarbon type analyses performed on the neat synthetic fuel and blend using GCxGC and ASTM D2425/D6379. However, there was good agreement between the techniques for the ‘nominal’ Jet A fuel. The differences most likely occur since the ASTM techniques were developed for petroleum-derived aviation fuel; therefore, the GCxGC results provide improved accuracy of the overall fuel composition. A comparison of the carbon number distribution data from Table 55 is shown in Figure 65. The synthetic fuel had a carbon number distribution slightly narrower than typical aviation fuels, but with an acceptable range of molecular weights. The higher concentration of C10-C14 compounds in the synthetic fuel and blend is due to high concentrations of multi-ring cycloparaffins and cycloaromatics (e.g., decalin, tetralin, dodecahydro-acenaphthylene, perhydrophenalene). The impact of increased concentrations of these higher molecular weight cycloaromatics and cycloparaffins will be more apparent when comparing combustion and FFP characteristics of the synthetic fuel blend. If necessary, subsequent processing could be performed to reduce the concentration of these compound classes in the neat synthetic fuel via either further distillation or hydrocracking.
The compositional characteristics determined by ASTM specification testing shown in Table 56 indicated the synthetic fuel blend conformed to the primary compositional properties required for aviation fuels containing synthesized hydrocarbons. More specifically, the acid number, total and mercaptan sulfur and hydrogen content all met current specification requirements for synthetic fuel blends. The total aromatic content for the synthetic fuel blend (15.6%) was lower than expected based on the composition of the neat synthetic fuel and Jet A; a repeat analysis via ASTM D1319 resulted in a total of 18% aromatics by volume. These quantified aromatic values are within the measurement uncertainty of the ASTM technique employed.
Figure 65. Comparison of carbon number distribution for neat synthetic fuel, Jet A, and 20% by volume synthetic fuel blend.
The volatility characteristics of the synthetic fuel blend satisfied all current fuel specifications and were similar to a ‘nominal’ aviation fuel. Although the flash point of the neat synthetic fuel (62.5 C) was near the high end of typical Jet A fuels, the synthetic blend had a flash point similar to the Jet A used for blending. This behavior is a result of the flash point being primarily dependent on the most volatile components in the fuel. The distillation profile of the blend was very similar to the nominal Jet A, which is a favorable characteristic during application. The distillation profiles for the synthetic blend and ‘nominal’ Jet A are shown in Figure 66. Included are distillation profiles for two additional jet fuels, a JP-8 (POSF 10264) and JP-5 (POSF 10289), which have also been used in the FAA NJFCP. The JP-8 has been termed a ‘best case’ jet fuel (e.g., low viscosity, aromatics and flash point with high H-content) while the JP-5 is a ‘worst case’ jet fuel (e.g., high viscosity and flash point with low H-content). As shown in Figure 4, the volatility profile of the synthetic fuel blend is within the bounds of the two reference fuels. The blend density (0.820 g/mL) was within the specification limits and is identical to the value calculated based on the blending ratio of the two fuels. A favorable characteristic of the synthetic aviation fuel produced in this program is that blending will increase the density relative to the neat petroleum-derived fuel.
Figure 66. Distillation (D86) profiles for 20% by volume synthetic fuel blend and ‘Nominal’, ‘Worst’, and ‘Best-Case’ jet fuels from FAA NJFCP program. ‘Nominal’ jet fuel used for blend preparation; value in parentheses corresponds to AFRL/RQTF identification number.
The fluidity characteristics of the synthetic fuel blend satisfied the aviation fuel specification requirements for fuel freeze point and low temperature viscosity (at -20 and -40 C). The freeze point behavior was expected due to the low value for the neat synthetic fuel. Blending of the synthetic fuel with the Jet A only slightly increased the viscosity relative to the neat petroleum-derived fuel (from 4.4 to 4.9 cSt at -20 C and 9.1 to 10.3 cSt at -40 C). This is a favorable result as the resulting viscosity was within specification requirement. The low temperature viscosity characteristics of the synthetic fuel blend will be further discussed below (See Section 9.1.3.2.1).
9.1.3.1.4 Combustion
The primary combustion characteristics in the aviation fuel specification requirement pertain to the energy content (i.e., Heat of Combustion) and sooting propensity. The synthetic fuel blend heat of combustion (43.1 MJ/kg) satisfied the specification requirement, as expected based on the high energy content for the neat synthetic fuel (45.1 MJ/kg). As discussed in Section 3.0, the high concentrations of high molecular weight cycloparaffins and cycloaromatics were most likely the cause for the low smoke point of the neat synthetic fuel (17 mm). However, blending with the petroleum-derived fuel resulted in an improvement to the smoke point (20.9 mm). This increased value with the corresponding concentration of naphthalenes (1.2% by volume) satisfy the fuel specification requirement. Further evaluation of the impact of the higher molecular weight cyclic compounds on combustion performance/stability and sooting propensity, as recommended in the ASTM D4054 and MIL-HDBK-510A processes, may be necessary to insure suitability of the fuel for use.
9.1.3.1.5 Thermal Stability
The thermal-oxidative stability of aviation fuels is evaluated using the JFTOT; the specification requirement for a synthetic fuel blend is a passing rating at 260 C. The synthetic fuel blend passed the JFTOT specification at this test temperature with a Visual Tube Rating (VTR) of < 1 and Filter pressure drop of 14 mm Hg. The VTR result demonstrated negligible tube deposition during stressing. However, the filter pressure drop of the blend was higher than observed for the neat Jet A. As discussed in Section 9.1.2, the neat synthetic fuel did not obtain a passing rating at 325 C, which is the test condition previously used for qualifying synthetic fuel blending feedstocks. However, passing at 325 C may not be necessary for eventual certification and use as the specific requirements for a synthetic fuel blending feedstock are determined based on the feedstock and production processes employed. Further evaluation, including determination of the JFTOT breakpoint (highest test temperature at which there is a passing result), is necessary to provide improved insight into the overall thermal stability characteristics of this fuel. Potential causes for the observed behavior for the neat synthetic fuel and the blend may be related to the presence of high molecular weight polar compounds (e.g., oxygenated multi-ring cycloparaffins and cycloaromatics). Further evaluation of fuel blend thermal-oxidative stability characteristics was performed using a Quartz Crystal Microbalance, and will be discussed in Section 9.1.3.2.3.
9.1.3.1.6 Contaminants/Corrosion/Wear
Several aviation fuel specification tests are performed to verify the fuel has acceptable compatibility and performance characteristics with minimal contamination. The synthetic fuel blend passed the existent gum, microseparator (MSEP; measure of impact on water coalescing/removal performance) and copper corrosion requirements. As expected, the blend had zero conductivity due to the low heteroatomic composition. If the conductivity of the blend would need to be increased, this can be addressed via use of static dissipater additive. The lubricity of the fuel blend (wear scar of 0.64 mm) is below the maximum allowable value of 0.85 mm.
9.1.3.2 Fit-For-Purpose Properties
The synthetic fuel blend met all current specification requirements for an aviation fuel which contains synthesized hydrocarbons. However, further evaluation of the performance and compatibility of the synthetic fuel blend beyond those characteristics evaluated by the specification properties is necessary to determine suitability for implementation and use on aviation platforms. These are referred to as “Fit-for-Purpose” (FFP) properties. FFP properties refer to a property required for safe operation which are not directly controlled by the respective fuel specification; a petroleum-derived fuel which meets the fuel specification will inherently satisfy all FFP property requirements. ASTM D4054 and MIL-HDBK-510 provide guidance on pertinent FFP tests to be performed when evaluating synthetic fuels and additive for use in aviation applications. Specific areas of recommended tests include those shown in Figure 1. Additional recommended testing may arise depending on results from the specification and compositional property results. FFP properties do not have well defined limits; rather the effect of the proposed fuel or additive on the corresponding FFP property must fall within the scope of experience of the engine and airframe manufacturers. The results from the FFP testing provide the basis for the FAA, DOD and aircraft engine/airframe OEMs to evaluate the potential suitability of the proposed fuel/additive for use and determine if further testing and evaluation is warranted. Completion of all recommended FFP tests requires several gallons the synthetic fuel blend for completion.
There was insufficient total volume of the synthetic fuel blend to complete all recommended FFP tests. Therefore, specific FFP tests were selected based on the available volume and composition/specification test results which would provide guidance regarding the potential suitability of the synthetic fuel blend for use in aviation applications. The selected FFP property tests performed in this effort are shown in Table 57. Subsequent evaluation of other FFP characteristics as outlined in the recommended approval guidelines must be completed to provide the basis for determination if the synthetic fuel blend is suitable for further evaluation and use. The following section discuss results from the specific FFP testing performed in this effort.
Table 57. Selected Fit-For-Purpose Property Testing Performed on Synthetic Fuel Blend
9.1.3.2.1 Low Temperature Viscosity
The synthetic fuel blend satisfied the specification requirements for viscosity at -20 and - 40 C; however, additional guidance regarding the corresponding fluid behavior within the low temperature regime is beneficial. Therefore, the low temperature dynamic viscosity characteristics of the neat Jet A and synthetic fuel blend were measured from - 20 to < -56 C using a Brookfield Scanning Viscometer with a cooling rate of -5 C/hr. The dynamic viscosity is converted to kinematic by normalizing to the corresponding temperature-dependent density. Results from these measurements are shown in Figure 67. Both the Jet A and synthetic fuel blend exhibit characteristics consistent with typical aviation fuels; the viscosity increases with decreasing temperature until there is a rapid increase near the phase transition (e.g., crystallization) temperature. For aviation fuels, this is typically due to crystallization of long chain n-paraffins in the fuel. The addition of the synthetic fuel suppressed this transition by approximately 2 C. The addition of the synthetic fuel slightly increased (~1-2 cSt) the kinematic viscosity relative to the baseline Jet A; however, this increase is minor and not expected to result in a significant performance impact.
Figure 67. Low temperature kinematic viscosity measurements of Jet A and synthetic fuel blend.
9.1.3.2.2 Analysis of Polars and S/N Content
Analysis of the heteroatomic and polar composition was performed to provide insight into the quantitative levels of these classes of species in the synthetic fuel blend. Heteroatomic and polar species are known to affect thermal-oxidative stability of aviation fuels and can impact material and storage compatibility. The sulfur and nitrogen content (total and polar) of the neat synthetic fuel, Jet A and neat synthetic blend were quantified using Gas Chromatography with Sulfur and Nitrogen Chemiluminescence Detection; results are shown in Table 58. The neat synthetic fuel had very low sulfur/nitrogen content, demonstrating that the processes and catalysts used in the fuel production were effective at removing the heteroatomic species.
The polar composition of the fuels was investigated via extraction using silica gel solid phase extraction. Retained polar compounds were eluted with methanol and analyzed using GCxGC-FID/MS to quantify the respective classes of polar compounds in the fuels; results are shown in Table 59. The Jet A fuel had ‘typical’ levels of polar compounds, primarily phenols and lower molecular weight compounds. Although the neat synthetic fuel had a similar total polar content, the respective species were of higher molecular weight (primarily 2-ring). The presence of these types of species is due to the composition of the neat synthetic fuel (e.g., tetralone related to high concentration of tetralin). The oxygenated species are either residual in the fuel following synthesis or formed via oxidation during storage. The presence of these types of species may impact the thermal-oxidative stability and seal swell propensity during application (discussed below). The polar content of the synthetic fuel blend was a result of blending of the neat synthetic and Jet A fuels.
Table 58. Total and Polar Sulfur/Nitrogen Content of Fuels using Gas Chromatography with Chemiluminescence Detection
Table 59. Polar Content of Fuels Determined using Solid Phase Extraction and Two-Dimensional Gas Chromatography
9.1.3.2.3 Thermal-Oxidative Stability via Quartz Crystal Microbalance
The thermal-oxidative characteristics of the neat synthetic fuel and blend were previously analyzed via JFTOT testing. The synthetic fuel blend passed the specification criteria at a test temperature of 260 C; however, the neat synthetic fuel failed at an elevated test temperature of 325 C. Improved insight into potential causes for these results would be useful regarding the suitability of the synthetic fuel for use in aviation applications. Therefore, the thermal-oxidative stability of the synthetic fuel blend was further evaluated using a Quartz Crystal Microbalance (QCM). QCM analysis has primarily been used to provide insight regarding oxidation and deposition tendencies of jet fuels due to differences in trace chemical composition (e.g., heteroatomic species and dissolved metals content).
The thermal-oxidative stability characteristics of the Jet A and synthetic fuel blend were assessed using the QCM operated under typical experimental conditions (i.e., 140°C, air saturated fuel, 15 hours). These experimental conditions are sufficient to identify inherent differences in oxidation rate and level of deposit accumulation of aviation fuels. Results from this testing are shown in Figure 68. The Jet A fuel is a slow oxidizer; deposit accumulation occurs as oxidized fuel species are produced and undergo further molecular growth reactions. The oxidation and deposition characteristics of the synthetic fuel blend were qualitatively similar to the Jet A fuel; however, the rate of oxidation and level of deposition were increased relative to the petroleum-derived fuel. Typical aviation turbine fuels produce about 1 to 6 g/cm2 of deposition under these test conditions; the synthetic fuel blend is near the high end of this range.
The impact of the addition of the synthetic fuel on the corresponding thermal-oxidative stability characteristics may be related to the polar composition and the primary species formed during oxidation of the synthetic fuel. As discussed in Sections 4.1 and 4.2.2, the synthetic fuel had a higher concentration of 2 and 3-ring cycloparaffins and cycloaromatics (and higher molecular weight polar compounds) than typically observed in aviation fuels. These species may be more prone to molecular growth via oligomerization reactions during oxidative stressing; similar behavior had been observed during a previous study investigating the impact of aromatic addition to SPK fuels [18]. Pretreatment of the synthetic fuel prior to blending (e.g., extraction of polars via silica or clay filtration) and addition of an antioxidant could possibly improve the corresponding thermal stability behavior. Although the neat synthetic fuel may be inherently more prone to deposit formation than a typical aviation fuel, the corresponding behavior is within ranges of typical aviation fuels when blended at reasonable concentrations (i.e., ≤ 20% by volume). Further evaluation of the thermal stability characteristics, such as JFTOT breakpoint or flowing system testing, would provide improved insight regarding the impact of the synthetic fuel during use. This could assist in determining if processing modifications or post-production treatments, such as filtration or addition of antioxidant, could further improve the thermal stability characteristics of the synthetic fuel blend.
Figure 68. Deposition (solid lines) and oxidation (dashed lines) profiles for thermal stressing of fuels using quartz crystal microbalance at 140 C.
The Derived Cetane Number (DCN) of a fuel is primarily relevant to use in diesel engines. DCN is a measure of the combustion speed during compression needed for ignition. This is an important property for use of military aviation fuels in ground support systems and vehicles. The DCN can be measured using an Ignition Quality Tester (IQT) per ASTM D6890. An acceptable DCN range of 40-65 has been recommended during qualification of synthetic aviation fuels (per MIL-HDBK-510A). The Jet A (POSF 10325) used in the blend formulation had a measured DCN of 48.3, while the synthetic fuel blend had a measured value of 45.5. The synthetic fuel blend value is within the recommended range for DCN and would be expected to perform satisfactorily during use.
9.1.3.2.5 Initial Material Compatibility (Seal Swell)
Compatibility of candidate synthetic fuels and additives with fuel system materials is critical to insure suitability for use without adverse effects. Materials of interest include metallic/non-metallic components and elastomers. The commercial and military fuel certification processes describe recommended approaches for evaluating the physical properties of materials following extended exposure to the test fuel. This includes aging the materials in the test fluids for extended durations and comparing the impacts to those observed with ‘typical’ petroleum-derived fuels. Compatibility testing for all fuel system materials requires very large volumes of fuel for completion. However, initial material compatibility evaluations can be performed via investigation of seal swell characteristics of elastomers. Elastomer seal swell has been shown to be critical during the certification of synthetic fuel blends to insure there is sufficient swell to prevent leakage in aircraft fuel systems. Therefore, the volume swell characteristics of selected O-rings exposed to the synthetic fuel blend were evaluated and results were compared to a historical reference data sets for normal volume swell behavior in aviation turbine fuels. This testing has previously been used to provide detailed information regarding material compatibility of potential synthetic fuel formulations while requiring very small fuel volumes for testing.
The volume swell of the synthetic fuel blend was evaluated using three different types of elastomers which are the most prevalent fuel wetted o-ring materials in conventional aircraft. Specifically, nitrile rubber (Identification Number N0602), fluorosilicone (L1120), and fluorocarbon (V0747) O-rings manufactured by Parker Hannifin were used in this evaluation. The plasticizer was removed from samples of the N0602 nitrile rubber using acetone solvent extraction and designated N0602e. For each analysis, two size -001 O-rings were cut in half with one section from each O-ring being used for this analysis. Volume swell characteristics were measured by performing optical dilatometry at room temperature. UDRI has used optical dilatometry for evaluating the impact of synthetic fuels on the seal swell of various elastomeric materials, including the initial evaluation of synthetic SPK fuel blends during certification for use in B-52 and C-17 aircraft. This technique requires a very small volume of fuel for evaluation and significantly shorter test durations than conventional ASTM soak tests. Briefly, two samples were placed in a reservoir with 5 mL of the test fuel. Starting at 1 minute after being immersed in the fuel, the samples were digitally photographed every 20 seconds for the next 3 minutes. At 5 minutes total elapsed time, the samples were photographed every 30 minutes for the next 40 hours. After the aging period was completed, the cross-sectional area was extracted from the digital images and taken as a characteristic dimension proportional to the volume. The results reported below are the average values obtained from the two samples. At the completion of the aging period, the absorbed fuel was extracted and the composition was analyzed by GC-MS. By comparing the GC-MS analysis of the fuel absorbed by the O-ring with that of the neat fuel, the relative solubility of the major classes of fuel components were summarized in terms of their respective polymer-fuel partition coefficients (Kpf). The major class fractions were taken as the normal-, and iso-alkanes, normal- and iso-alkyl benzenes, naphthalene, and alkyl naphthalenes. These were isolated from the GC-MS total ion chromatograms using ions 57 (n,i-alkanes), 105 (principally the C3 substituted alkyl benzenes), 128 (naphthalene), and 141 (C1 and C2 substituted naphthalenes). Furthermore, cycloalkanes and tetrahydronaphthalene (tetralin) were analyzed using ions 83 and 104, respectively.
The basis for comparison was the average values and 90% prediction intervals for the volume swell and partition coefficients obtained from 12 reference JP-8s with aromatic contents from 10.9% to 23.6% by volume (v/v) (prior studies have shown that the volume swell behavior of JP-8 is similar to Jet-A). It should be noted that the prediction interval is a statistical estimate for the range of values that would be exhibited by 90% of all individual JP-8s, and therefore reflects the estimated range of behavior for ‘normal’ JP-8s with 10-25% aromatics. The results summarized in Table 60 and Figure 69 show that the volume swell behavior of the synthetic fuel blend and the Jet A used for blend preparation were well within the normal range typically observed for JP-8 for the nitrile rubber and fluorosilicone materials. The volume swell of the fluorocarbon was somewhat lower than average, likely due to the lack of Fuel System Icing Inhibitor additive (Di-Ethylene Glycol Monomethyl Ether) which is required in military aviation fuels. However, the absolute difference is very small and does not indicate a potential issue with either the Jet A or synthetic fuel blend.
The overall absorption of fuel by the sample O-rings is summarized in Table 61 and Figure 70. These results show that the average solubility of all major class fractions examined are within the range typically observed for JP-8s. Note that the solubility of the cycloalkanes in the test fuels with nitrile rubber is about 10% higher than exhibited by linear and branched alkanes, which is consistent with prior work. The solubility of tetralin, a significant component of the synthetic fuel, was modestly higher than the C3 alkyl benzenes, but lower than the alkyl naphthalenes. This suggests that the presence of the synthetic fuel components slightly elevates the overall solubility in the elastomer which increased the volume swell character of this fuel blend as compared to the neat Jet A.
Overall, the volume swell character of the synthetic fuel blend is within the range typically observed for aviation turbine fuel and should therefore be compatible with the O-ring materials studied. Further evaluation of the compatibility characteristics of the synthetic fuel with other fuel system and engine materials may be warranted.
Table 60. Summary of Volume Swell of Selected O-rings Aged in the Test Fuels and the Average Values for JP-8
Figure 69. Volume swell and 90% confidence intervals of selected O-ring materials in the synthetic fuel blend and neat Jet A compared with the 90% prediction regions for JP-8.
Table 61. Average Polymer-Fuel Partition Coefficients (Kpf) of Major Fuel Class Fractions in Selected O-Rings Aged in the Synthetic Fuel Blend and Neat Jet A Compared to Average Values for JP-8
of synthetic fuel blend and fuel components absorbed by elastomers (see Table 11
for Parker Identification Number). Primary compounds are identified.
chromatograms scaled for comparison. Note: Peak assignments tentative based on
NIST mass spectral library.
Advertisement
The 10 largest coal producers and exporters in Indonesia:
- Bumi Resouces (BUMI)
- Adaro Energy (ADRO)
- Indo Tambangraya Megah (ITMG)
- Bukit Asam (PTBA)
- Baramulti Sukses Sarana (BSSR)
- Harum Energy (HRUM)
- Mitrabara Adiperdana (MBAP)
- Samindo Resources (MYOH)
- United Tractors (UNTR)
- Berau Coal
