Technical Memorandum

To:		EPA

Title:	Techno-economic analysis of current technology for Fischer-Tropsch fuels production

Author:		Ryan Davis

Date:		August 14, 2009


Abstract
The purpose of this study is to evaluate the economics of a biomass-based Fischer Tropsch (FT) plant, based on publicly available literature and engineering data.  The focus for the plant is on maximum production of liquid fuels (specifically diesel) consistent with recent developments in FT synthesis.  The basis for the process was the 2007 NREL thermochemical ethanol design report [1], which utilized low pressure indirect steam gasification in a 2,000 dry tonne/day plant processing hybrid poplar wood chips, and serves as the same starting point for the FT model.  The design report established specific design targets thought to be achievable by the year 2015, while the FT modifications contain somewhat optimistic assumptions that would be reflective of a "best-case" type of scenario; thus, it was decided that this study would be appropriate for operation commencing in the year 2022.  Upon completing the base case scenario, the minimum required diesel selling price to achieve a target 10% IRR was found to be $2.06/gallon.  Of the various sensitivity parameters evaluated, the areas with the largest impact on diesel price are synthesis conversion extent, feed price, and coproduct values.

Keywords: 	Fischer Tropsch, state of technology, techno-economic, diesel, fuel

Introduction
As noted above, the basis for this study was the 2007 NREL thermochemical ethanol design report.  Based on previous NREL studies comparing the various gasification options, it was decided that the same gasification scheme would be used here, thus the design report model served as a starting point and was modified as appropriate for Fischer Tropsch fuels.  Modifications were made to the tar reformer, gas cleanup, and steam sections, while most of the synthesis and separation sections were replaced altogether as described below.  Very few changes were required in the front-end sections of the plant and the same base feedstock type and cost ($50.70/dry ton) was assumed.  Additionally the same plant capacity factor (96% = 8,406 operating hours/year) was assumed, and the original assumption of 100% equity financing was not changed. 

The basic flow scheme for this process is as follows: the feed is delivered to the plant, dried and gasified in the same manner as described in the design report.  The dirty syngas is fed to the tar reformer which operates under the same conditions as the base model, with the exception that more steam is fed to the reformer to adjust the H2/CO ratio to that required for FT.  The syngas then passes through the quench section and compressors (which were left unchanged), before feeding to the amine scrubbing unit which removes H2S to a lower level (20 ppmv vs 50 ppmv previously).  H2S is then further removed to sub-ppm levels using a zinc oxide guard bed, and a small slip-stream is sent through a PSA to provide hydrogen for downstream hydroprocessing.  The majority of the material is fed to a slurry column synthesis reactor, which converts the syngas to a range of fuel-range hydrocarbons and wax.  The wax is sent to a hydrocracker to upgrade it to additional fuel-range material, while the lighter hydrocarbons are cooled and knocked out from the unconverted syngas, the latter of which is sent to an expander and recycled to the tar reformer area, as was done in the ethanol model.  The liquid hydrocarbons and hydrocracker products are sent to a fractionator to separate the products into light gases, naphtha, diesel and residual wax.  Each of these steps, as well as the underlying process and costing assumptions, are described in more detail in the following sections. 

Process Descriptions and Major Assumptions for Model/Cost
As described previously, the front-end sections (feed preparation, gasification and char combustion) were left essentially unchanged, and their process descriptions can be found in the NREL thermochemical design report.  The feed is assumed to enter the plant at a cost of $50.70/dry ton, consistent with the updated price for the base model.  The following sections describe changes made to the downstream units of the model and their underlying assumptions.

Area 300: Conditioning
The assumptions made in A300 are summarized as follows and described in more detail below:
   * Tar reformer: Larger steam feed than necessary for reforming is not a problem for catalyst
   * Still meeting all steam/power requirements internally, even though steam demand is higher here and requires combustion of more syngas at the tar reformer regenerator
   * MEA is still preferred amine type; MEA CO2/H2S selectivity = 1
   * Increased amine unit cost associated with more acid gas removal was incorporated by scaling the original cost by the ratio of new/original total inlet flow (syngas + amine)
   * Steam methane reforming unit is not warranted (some studies include it to increase syngas yield, but CH4 is not present in substantial amounts here)
   * NH3 removal is sufficient in existing quench scrubber
   * No COS hydrolisation unit required
   * No dedicated CO2 removal step required 

The "Conditioning" section includes the tar reformer, quench, acid gas removal, and sulfur removal.  The quench and sulfur removal sections were left unchanged, and any differences in those areas relative to the original model were a result of new inputs to either section.  The tar reformer operating conditions and specifications were also left unchanged, however the steam feed was increased substantially.  While the syngas into the tar reformer has a H - 2 - /CO ratio close to 0.6, cobalt-based FT synthesis requires a ratio of 2.1 [2].  This can be met by utilizing a water-gas shift (WGS) reactor to favor H2 production at the expense of CO via the reaction: CO + H2O <--> CO2 + H2 .  However, the tar reformer catalyst also maintains WGS activity, so it was decided it would be beneficial to save on the capital cost for a dedicated WGS reactor by utilizing the reformer catalyst's WGS capability to meet the synthesis composition requirements.  Thus a design specification was created to increase steam to a rate that shifts the equilibrium for this reaction to achieve the proper syngas ratio.  An implicit assumption here is that excess steam does not hinder the reactor's capacity to crack heavy components (tars), which should be valid as steam is also required to reform the hydrocarbon components. 

The tar reformer catalyst regenerator was also left unchanged, but operates at a higher temperature to satisfy increased steam demands and close the heat balance; the assumption here is that the plant remains electricity-neutral and meets all internal power requirements by burning process material.  Thus since steam demands are higher in this model, more of the raw syngas as well as the recycled unconverted gas from the synthesis reactor must be diverted to the regenerator.  This ultimately results in lower product yields and decreased overall plant efficiency relative to the ethanol model.

The acid gas cleanup unit was also left mostly unchanged, but has higher internal traffic.  While the ethanol synthesis reactor had a sulfur limit of 50 ppmv H2S, the FT synthesis catalyst has a higher sulfur sensitivity and requires cleanup to ppb levels [3].  There are several options to achieve this, the most common being amine scrubbing followed by zinc oxide polishing, or physical absorption processes such as Selexol or Rectisol.  The amine/ZnO process was chosen here as Selexol has not been found to be favorable for H2S removal and Rectisol is too costly at the smaller scale of a biomass plant, compared to a larger coal-based FT plant [2].  A ZnO guard bed can readily remove H2S to the required level, but has an inlet sulfur limit of about 20 ppmv; thus the amine scrubbers must remove sulfur to this level first.  

For consistency with previous NREL models, MEA was still assumed to be the chosen amine here.  MEA also removes the most CO - 2 of the conventional amines, which is useful here to minimize inerts (CO2) in the synthesis reactor, and thus reactor size.  Although MEA is slightly more selective for CO2 than H2S [4], selectivity was assumed to be 1 here as complete CO2 removal wouldn't be realistic.  As a result of a significantly larger H2S (and associated CO2) removal, the amine scrubber traffic and thus amine reboiler heating steam demand increased substantially over the base ethanol model.  Since the amine unit was originally costed as a total package, the associated increase in cost was calculated by scaling the original cost in the design report by new/original total traffic (inlet syngas + amine flow).

Additional cleanup considerations for selected components are shown in Table 1.  Although some studies include separate provisions for removing some of these components, it is assumed here that the existing units remove them sufficiently.



            Table 1: Considerations for additional FT contaminants
                                  Contaminant
                                FT requirement
                                  Discussion
                                  N (NH3/HCN)
                            <20 ppb total N [5]
Most other studies assume nitrogen compounds can be sufficiently removed in the existing quench scrubber; assuming the same here.
                                      COS
                            <10 ppb total S [5]
The NREL studies do not include a provision for COS formation, and it is assumed COS is not produced in significant amounts.  If trace amounts do form, COS can be removed to an extent in the ZnO bed [6].  If there were significant COS formation, the alternative would be to add a hydrolisation unit; it is assumed here that this is not necessary.
                                      CO2
                                      NA
CO2 is not a contaminant, but as an inert can increase reactor costs if it is present in large amounts.  Thus some studies have included provisions such as a Sulfinol unit to remove CO2.  However, CO2 removal may only be justified if over ~10% [5].  In this study, the concentration is only 1% due to co-removal with H2S in the MEA-based amine scrubber.


A400: Fuel Synthesis
The assumptions made in A400 are summarized as follows and described in more detail below:
   * No secondary compression required; compressors before amine section pressurize the syngas to an adequate pressure for proper FT performance
   * Zinc oxide beds: remove H2S to adequate levels, also remove any COS that may be present; equipment cost scaled from NREL H2 report (assuming two beds required); catalyst cost and amount also based on NREL H2 report
   * PSA unit: feed is set by H2 required for hydroprocessing, assuming 85% H2 recovery and >99.9% purity.  Cost is for total PSA "package" (assuming this accounts for total number of beds required as well as sorbent cost), and can be scaled from the NREL H2 study which was based on a much higher flowrate.  Since cost is on a flowrate basis, it is assumed that any periodic sorbent changeout (if required) is already included in the cost and is not costed separately.  
   * FT synthesis reactor: 446F and ~27.5 bar operating conditions; conversion = 80% and selectivity α = 0.9, both set independent of reaction conditions but  based on "typical" performance that can be achieved in modern FT synthesis.  Reactor is cobalt-based slurry column reactor, and capital cost scaled from parameters given in other studies.  Catalyst cost was calculated using GHSV = 5800 std L/kg cat/hr, 3 year lifetime, $25/lb based on various sources.
   * Wax leaves synthesis reactor as a separate stream from the rest of the product and feeds to wax hydrocracker; hydrocracker cost was scaled from other studies and is assumed to include all equipment "inside battery limits" for the entire plant as well as associated catalyst costs.  Hydrocracker wax conversion = 90% with 70/30 wt% diesel/naphtha yield and 4% light ends.

The first A400 unit in the base model was a 2[nd] compression section to pressurize the clean syngas to 1,000 psia as required for alcohol synthesis.  However, modern slurry reactors for FT synthesis can operate as low as 20-25 bar (~300-350 psia) while still maintaining high selectivity and conversion [7].  Given that the first compression section in A300 pressurizes the syngas to 425 psia, the feed can already be delivered to the synthesis reactor at 400 psia after including pressure drop through the intermediate equipment; thus the second compressor section was eliminated.  

The first new unit in A400 is the zinc oxide guard bed, which treats the residual H - 2S down to ppb levels according to the reaction  ZnO + H2S <--> ZnS + H2O .  The setup for this unit was based on the 2005 NREL H2 study [6], which set the inlet temperature to the reactor at 707 °F so that the reaction closely approaches equilibrium.  Thus the original reactor feed heat exchangers were used to increase the temperature to 707 °F before entering the guard bed.  Although there would still be traces of sulfur leaving the ZnO unit, complete removal was assumed for the purposes of the model.  Additionally, although COS is not expected to be present in significant amounts, the ZnO bed would remove small amounts of this species as well.  Cost for the ZnO reactor itself was scaled from the H2 study according to inlet rate, and it was assumed two beds would be required to allow for constant operation if one bed needs to be changed out.  The ZnO catalyst assumptions were also the same as in the H2 study, namely that the catalyst has a 5-year lifetime and costs $4.67/lb, and the required catalyst volume was calculated using GHSV=4000 hr[-1] based on standard flow conditions.

Following the ZnO beds, a small slipstream of clean syngas is diverted to a PSA unit to purify hydrogen for downstream hydroprocessing uses.  The PSA was also set up similar to the NREL H2 study, where the feed is first cooled to 110 °F and any water is knocked out, and H2 is produced at >99.9% purity (assumed 100% here) and 85% recovery.  Some of the purified hydrogen is recycled back to the PSA inlet given that typical PSA operation requires >70% hydrogen purity [6].  The PSA tail gas combines with the unconverted syngas from synthesis after the expander, and is recycled to the tar reformer section (this also sets the tail gas pressure).  The PSA cost was also determined by scaling the cost in the H2 study, which uses a $/scfd basis; it is assumed here that since the cost is already on a flow rate basis, any periodic sorbent changeout (if required) is already included in the overall cost, and there is not a separate provision for ongoing sorbent material expense.  Although the PSA section would contain several beds, this cost basis is for the total PSA "package."

The majority of the cleaned syngas leaving the ZnO unit feeds to the FT synthesis reactor.  The operating temperature of the synthesis reactor is set at 446 F based on various sources [5], and the operating pressure is 404 psia as discussed above.  It was decided to use a slurry reactor instead of a fixed-bed type due to its many advantages, such as much higher single-pass conversions (80% or more) in a relatively small reactor volume and superior control of the heat released from the exothermic reaction  CO + 2.1H2 --> -(CH2)- + H2O  [8].  Additionally, a cobalt-based catalyst was chosen over the iron-based alternative due to many advantages such as higher activity, higher diesel yields, and much longer catalyst life, which justifies its initial higher price [9].  As is done in most other FT studies, it is assumed that hydrocarbon component selectivity follows the Anderson-Schulz-Flory chain growth equation, which states that the molar yield of a given component with carbon number n is a function of α in the equation:
Cn=αn-11-α

The parameter α is dependent on the reactor conditions and type, and in modern cobalt-based slurry systems is around 0.9 or higher [7], where a higher value implies greater selectivity towards high carbon numbers (diesel and wax); for this study, a value for α = 0.9 was assumed.  The yields of C1-C4 were re-distributed to favor more methane and less of the others as is closer to reality [5].  The conversion was set to 80% which is also achievable in modern slurry reactor systems [8].  The operation of a slurry column reactor entails suspending fine catalyst particles in the product wax and bubbling syngas through the mixture, while removing heat through boiler feed water tubes running through the reactor to produce steam.  While the mechanism to separate product wax from the catalyst is not immediately clear, it is assumed that this operation is included in the "inside battery limits" cost for the FT reactor.  The cost was taken by scaling the size and cost based on parameters found in other studies, most of which reference a Bechtel analysis done for slurry-based FT systems [8].  The basis for cost scaling is described in the ethanol design report, and was also done here according to:

New Cost= Base CostNew SizeBase Sizef

As noted above, a cobalt-based catalyst was chosen for this study.  The required catalyst amount was calculated based on the report by Larson et al, which assumed a GHSV = 5800 std L/kg cat/hr for the FT reactor [8].  The cost for cobalt-based FT catalyst varied over a number of studies from $10-30/lb, with an industry source quoting $25-30/lb [10]; given this info, the cost assumed here was $25/lb.  Finally, the catalyst lifetime varied by source from 2.5-5 years; the lifetime assumed here was three years as this was the most common figure. 

Following the fuel synthesis reactor, the wax material is sent to a wax hydrocracker while the rest of the product is cooled in a series of heat exchangers and knocked out to split the unconverted and light gases from the liquid product.  This is a similar scheme as the original base model, with the exception that the knockout drum is now three-phase as a result of large amounts of water produced from the FT reaction.  As this is a comparatively small cost, the new cost for a three-phase separator was estimated by adding 20% to the cost of a two-phase drum.  The water product is recycled back to the steam cycle, while the gas product is heated and expanded to recover electricity similar to the base model (the expander was left unchanged, but produces less electricity since the inlet pressure is lower than the ethanol case).  The PSA tail gas then joins this stream and the mixture is recycled back to the tar reformer and regenerator.  The liquid product is sent to a low-pressure flash drum prior to distillation, as was done before.

A400: Hydroprocessing
The wax produced from the synthesis reactor is sent to a wax hydrocracker, where it is upgraded to additional fuel-range material.  The hydrogen rate from the PSA was set such that the H2/liquid feed ratio was 1200:1 nm[3]/m[3], and the hydrocracker operates at relatively mild conditions, ~710F and 500 psia with a 90% wax conversion [11].  The product slate was set at 70/30 wt% diesel/naphtha with approximately 4% light ends [2, 11].  Although the hydrocracking plant was modeled in detail, there is insufficient literature data on an accurate cost for the individual hydrocracking reactor and catalyst; thus the unit as a whole was costed based on literature data for a hydrocracking plant as is done in other studies [8].  It is assumed that the hydrocracking catalyst expense is included in this "inside battery limits" cost, given that the cost is on an inlet flow rate basis.  The outlet streams from the hydrocracker are a heavy hydrocarbon liquid which feeds directly to distillation, a lighter liquid which combines with the synthesis liquid stream at the low pressure flash drum, a small flue gas stream produced as a result of the hydrocracker furnace (combines with the other flue gas streams), and a small purge offgas stream which is burned in the tar reformer regenerator.

A500: Product Separation
The assumptions made in A500 are summarized as follows and described in more detail below:
   * A single steam-stripped column will adequately separate the products into offgas, naphtha, diesel, and residual wax
   * Column specified to produce 95% purity diesel; naphtha purity not as important as it requires additional refining to be considered "gasoline" and wax is considered "low-grade" paraffinic wax
   * 28 actual stages required assuming 50% tray efficiency; column was costed in Aspen IPE
   * Total column feed is heated to 600F and heat is supplied by FT reactor feed cooler; this coupled heat exchanger was costed in Aspen IPE
   * Remaining heat exchangers scaled by duty based on another exchanger in similar service.  The pinch analysis was not re-evaluated as the time required is not justified based on the cost impact; even doubling the cost of every heat exchanger changes the diesel price by 3 cents.
   * For the diesel product to be considered "finished" (ready for use as transportation fuel), a diesel hydrotreater is required.  This was not modeled in detail, but was costed based on an "inside battery limits" estimate for a hydrotreating plant.

 The separation section consists mainly of a single distillation column to separate the various products, thus is simpler than the base model as the molecular sieve unit and 2[nd] distillation column are not necessary.  As mentioned previously, the liquid synthesis products as well as the light liquid from wax hydrocracking are flashed in a low-pressure drum and the offgas is recycled to the tar reformer.  The liquid product is preheated to 250F before mixing with the heavy liquid stream from wax hydrocracking, and the combined stream is heated to 600F and fed to the distillation column.  This column operates at higher temperature than the base model because the products have significantly higher boiling ranges; as such, a reboiled column would result in bottom temperatures in excess of what could be met by heat exchange with steam or most other process streams.  Thus the column was converted to a steam-stripped fractionator as is common in the petroleum industry; this resulted in a more reasonable temperature profile and good internal column traffic (the alternative would be to draw diesel+ material from a reboiled column and split diesel/wax in a 2[nd] vacuum column).  Since this is a more ideal mixture, the column decreased in diameter to 5 ft and number of trays required was 28 actual trays (assuming 50% tray efficiency).  This new column was costed in Aspen IPE based on the resulting dimensions.  

The fractionator was specified to produce diesel at 95% purity, while the naphtha and wax material was closer to 90%.  It was assumed that the purity of these byproduct streams is not as important since the naphtha requires additional upgrading before it can be sold as "gasoline" and the wax does not have stringent component requirements, but would be sold as a "low-grade" paraffinic wax.  The offgas from the column (consisting mostly of C1-C4 components) was recycled to the tar reformer.  The product run-down lines were left unchanged, and the wax coolers and product tank were costed in the same manner as the existing equipment.  

According to various studies, the diesel cannot be considered "finished" (ie, ready for use as transportation fuel) without final upgrading to improve various engine performance properties and to crack any residual wax [8].  Thus a diesel hydrotreater was included in the capital cost calculations, although it wasn't modeled in detail as the base cost for this unit was another "inside battery limits" package cost based on diesel flow rate.  It was assumed that the final product flowrate would be the same as that leaving the distillation column, which should be reasonable given that this would be fairly mild hydrotreating and the goal is not to break the material down into lighter fractions. 

It should be noted that the pinch analysis was not re-evaluated from the base model as this would take an unreasonable amount of time to adjust the heat exchanger network, relative to the impact on the economics.  Thus, all heat exchangers were costed based on the same pinch network with updated duties, and were checked to ensure there would not be temperature crosses at the beginning or end of a heat exchanger series.  The new FT feed cooler/fractionator pre-heater was coupled and costed as a single unit in Aspen IPE, as well as new utility coolers.  Any other new exchangers were costed based on a similar exchanger that would have somewhat similar U and LMTD values.  This approximation to heat exchanger costing is reasonable in the scope of this study, given the fact that even doubling the cost of every heat exchanger would only change the diesel selling price by ~3 cents/gallon.

A600: Steam System
The assumptions made in A600 are summarized as follows and described in more detail below:
   * High-purity water produced as a result of FT synthesis is recovered and recycled to the steam cycle
   * A new medium-pressure steam loop was created to accommodate a lower FT synthesis operating temperature and thus steam quality (relative to the original ethanol synthesis reactor).  This included a new medium-pressure boiler feed water pump, preheater, and steam superheater, all of which were costed by scaling to the same piece of equipment in the high-pressure loop.
   * To accommodate the lower steam pressure generated at the FT synthesis reactor, the steam turbine was modified to accept the medium-pressure steam at the 2[nd] stage while still extracting low pressure steam at the 3[rd] stage for the various process and heat demands in the plant.  The anticipated higher cost for this turbine was estimated by assuming the total (HP+MP) steam rate is theoretically introduced at the 1[st] stage.

The steam system was slightly modified to include recycle streams from new water knockout sources added as a result of FT water production, as well as medium pressure steam produced from the FT synthesis reactor.  As noted previously, the pinch analysis was not re-evaluated in this study.  Since the pinch point was originally set by the alcohol synthesis reactor temperature, an adjustment to the steam heat balance was necessary to accommodate the lower operating temperature of the FT synthesis reactor.  Originally, steam was generated at 527F based on a 43F temperature approach to the 570F operating temperature of the ethanol synthesis reactor.  Assuming the same approach temperature with the 446F operating temperature in the FT model, steam is generated at 403F which results in a lower saturation pressure, and thus lower quality steam to drive the turbine generator.  As this would result in an unacceptably low pressure differential and thus power generation at the turbine, the turbine was modified to accept the same high-pressure steam generated elsewhere in the plant at the first stage, and the new medium-pressure steam generated from the synthesis reactor at the second stage.  Low-pressure steam is still extracted from the third stage to meet the various process and heat demands throughout the plant as described in the original design report.  Since steam is both injected and extracted at intermediate stages in the turbine, this modified configuration is called an "induction/extraction turbine."  

The new medium-pressure steam cycle includes a separate boiler feed water pump, boiler feed water preheater, and medium pressure steam superheater to bring the steam to the proper 2[nd]-stage turbine injection temperature.  All of these units were costed by scaling the power (for a pump) or duty (for an exchanger) based on the same piece of equipment in the high-pressure steam loop.  The medium pressure steam is generated at saturation conditions via tubes immersed in the synthesis reactor shell, and this cost is assumed to be included in the overall reactor cost described above.  The cost for an "extraction/induction" turbine was not immediately available in literature, so the new cost was estimated by assuming the full flowrate into the first stage of the turbine and scaling based on the original estimate.  Although the first stage does not actually see this full flow, this is a reasonable way to reflect a higher turbine cost due to a more complex design, when the cost for such a design was not available.

Operating Costs
The assumptions made for operating costs are summarized here and described in more detail below:
   * ZnO and FT catalysts require a "disposal cost" at the point when they are replaced; cost was assumed to be the same as the tar reformer catalyst disposal cost
   * Naphtha coproduct credit = $3.27/gallon, estimated by discounting projected gasoline price in 2022 by $0.40/gal
   * FT wax coproduct credit = $0.49/lb, estimated by scaling the 2006 paraffin wax spot price by 2022 projected oil price and discounting by 50%.
   * All other operating cost estimates are assumed to be on the same basis as in the ethanol design report

For this study, zinc oxide and FT synthesis catalyst were both assumed to require a "disposal cost" at the point when they are replaced, similar to the tar reformer catalyst.  The cost was assumed to be the same as that for tar reformer catalyst disposal, and is based on a basic "solid waste disposal" rate; this does not account for any environmental or metal recovery premiums that could be associated with either of these catalysts.  However, this assumption has a negligible impact on overall economics, as it was determined that even at 100x this base cost, the minimum diesel selling price was the same rounded to the nearest cent.

The diesel price is much more dependent on assumed co-product credit for naphtha and residual wax.  Although some studies assume the naphtha price can be approximated by "gasoline", this is not valid unless the plant includes numerous naphtha upgrading units as the naphtha produced by FT synthesis is low-octane and generally low-quality [12].  Thus the coproduct credit for naphtha was based on an analysis by Reed et al which discounted the price of gasoline by $0.40/gallon, which in turn was based on an average estimated cost required to upgrade FT naphtha for blending in the gasoline pool [12].  The base gasoline selling price was taken from EIA projections for gasoline in year 2022, $3.67/gal in 2007 dollars [13].  Thus the net assumed naphtha coproduct credit was $3.27/gallon.  It should be noted that as expected, the gasoline price is projected to continue increasing into the future, thus if a shorter timeframe was chosen earlier than 2022, the required diesel selling price would be higher and vice-versa, but this is an artificial influence on economics as it only reflects fluctuating energy prices and not physical advances in the research and technology.

There is a small amount of residual wax leaving the plant as a result of <100% conversion at the wax hydrocracker.  The wax selling price was estimated by assuming a refinery "paraffin wax" price in 2022 and discounting by 50% [14].  The projected price for paraffin wax was estimated by first taking the paraffin wax spot price in 2006 [15] and dividing by the US imported oil average price in the same year [13].  This ratio was then applied to the EIA projected oil price for 2022 [13] to arrive at the projected paraffin wax value, given that this price is largely a function of base oil prices.  The resulting estimate is $0.49/lb of FT wax in 2007 dollars.

Sensitivity Analysis
As noted previously there were numerous assumptions made to reasonably define the system, both from a process and costing standpoint.  Given that each assumption has a different degree of impact on the bottom line economics, an analysis was run to determine the sensitivity of the diesel selling price on a particular variable, while all other factors were held constant.  The results are shown in Figure 1.

                               Financial/ Market
Process

                   Figure 1: Results of sensitivity analysis
Figure 1 is divided into two categories, process assumptions and financial/market assumptions, and each factor was varied over a reasonable range (either according to values used in similar studies or elsewhere in literature).  The synthesis conversion extent was the largest process-related assumption evaluated, from the base value of 80% (achievable in modern slurry reactors) down to 40% which was chosen as this is typical for fixed-bed FT systems [8].  Various alternatives for the gas cleanup section were evaluated, as this section varies widely between studies.  A rough estimate was made to determine the cost of removing H2S and CO2 using a Rectisol unit based on parameters from [8], and was compared to the more standard schematic of amine/ZnO cleanup.  Additionally, the use of DEA instead of MEA in the amine unit was evaluated, but both of these changes had a relatively small effect on diesel price.  A somewhat larger impact was realized by supplementing the increased steam heat demands with natural gas (up to 5 MMSCFD) instead of relying purely on burning process streams and thereby sacrificing product yields; this improves economics but hurts the overall plant sustainability and self-sufficiency.  

The largest financial/market assumptions to have an impact on diesel price were feed cost and assumed coproduct credit.  Feed cost has been well-documented to have the single largest impact on product selling price in many studies, and the price range evaluated here covers the range typically assumed to be realistic.  The coproduct credit is also an important variable on bottom-line economics as discussed previously, and the range evaluated here was set to +25% to account for uncertainties in future oil prices.  Additionally, the stream factor (online production time) was varied over the range typically assumed in other studies to demonstrate the effect this factor has in artificially influencing diesel price.  Finally, FT catalyst cost was varied over the range cited by industry [16, 10], but diesel price was not affected over this range.

Concluding Remarks
This study essentially served as a "best-case" scenario for FT diesel production; underlying process assumptions for achievable conversion and yields were higher than some studies but equivalent to others, and are based on literature sources citing recent developments on both the research and industry-level scale.  Additionally, a 96% capacity factor (operating time of 8406 hours/year, chosen here for consistency with the ethanol model) is somewhat optimistic relative to other studies, and artificially improves economics relative to other studies with lower capacity factors.  It should be noted that this model was developed from NREL's thermochemical ethanol design report model, and all assumptions in the front-end section of the plant such as feedstock cost and gasifier performance are based on the same "future" design targets.  Additionally, the FT model was developed to maintain similarity with the ethanol model (same recycle scheme, cleanup using amine scrubbing with MEA, no net electricity production etc); this is by no means the only feasible approach, and some of the various alternatives are presented in the "Sensitivity Analysis."  Still other options exist as well, such as eliminating recycle and focusing on net power generation as some other studies have investigated.  Given this information, a comparison against similar biomass-based FT studies is shown below for reference [16], as well as a brief explanation of differences. 

     Table 2: Comparison against other FT studies (other values from [16])
                                   Parameter
                               Larson et al [8]
                              Tijmensen et al [2]
                              Swanson et al [16]
                                 Present study
Feed rate (dry tons/day)
                                     5000
                                     1920
                                     2205
                                     2205
Operating time (hours/year)
                                     7000
                                     8000
                                     7446
                                     8406
Feed cost ($2007/dry ton)
                                     51.80
                                     39.54
                                     75.00
                                     50.70
Diesel yield (gal/dry ton)
                                      25
                                  45  [5][2]
                                    32-41 
                            (LT-HT gasification)[3]
                                     43.3
Total fuel yield (GDE/dry ton[1])
                                     38.9
                                     55.7
                                     42.3
                                     62.7
Total Efficiency (%, LHV)
                                     57.3
                                     50.1
                                     42.7
                                     46.7
Product Value ($/GDE[1])
                                     2.32
                                     2.56
                                     5.40
                                     2.51
   1. GDE = Gallons Diesel Equivalent (all products normalized to diesel according to heating value)
   2. Specific diesel yield value from Hamelinck et al; this report expanded on the Tijmensen study and used similar parameters, but broke out fuel yield into individual products while Tijmensen did not (all other values in the column are from the Tijmensen study)
   3. This study looked at both low-temperature (LT) and HT gasification schemes, and the specific diesel yield is reported for both cases.  All other values in the column are for the LT case.

As evident in Table 2, the results of this study are reasonable compared to the others, keeping in mind that the results vary according to different production scenarios.  For example, the Larson et al study did not employ a recycle loop (thus lower normalized yields) but burned un-converted syngas in a turbine to generate extra electricity to sell (thus higher total efficiency values, which include electricity).  The economics may be slightly better in the Larson study due to a larger scale as well as inclusion of all refining units to upgrade the naphtha value.  The Tijmensen et al study is relatively similar to this case in that it focuses on maximum fuel production, and shows similar yield results after normalizing operating time.  The Swanson et al study is not quite the same, in that it employs a fixed-bed synthesis reactor and thus much lower per-pass conversion, which results in lower yields and less attractive economics.  Note that the $/GDE figure in the present study is provided to establish a common basis for comparison; it was calculated by normalizing the coproducts to diesel according to heating value and thus eliminates uncertainty associated with projected coproduct credit.  The key economic metric in this study is the minimum diesel selling price, which is $2.06/gallon of diesel fuel for the base case.

The intent of this report was to provide a snapshot analysis of FT technology and economic viability by applying the most recent developments in the FT industry (typically based on coal or gas feeds) to a biomass feedstock.  As such, the results have a higher degree of uncertainty given that a commercial-scale biomass-based FT facility does not yet exist.  There is room for further development of this study in the future, through means such as a more thorough reactor model that calculates specific selectivities and conversions, evaluation of alternative process layouts and gas cleanup options, validation of the larger capital equipment costs, and more individual costing for areas such as the hydrocracker plant.  However, the assumptions employed here regarding such parameters are standard across most studies, thus the reliability of these results are comparable.
References
[1]  Phillips, S. et al., "Thermochemical Ethanol via Indirect Gasification and Mixed Alcohol Synthesis of Lignocellulosic Biomass" (2007), http://www.nrel.gov/docs/fy07osti/41168.pdf
[2]  Tijmensen, M. et al., "The Production of Fischer Tropsch Liquids and Power Through Biomass Gasification" (2000), Utrecht University, NWS-E-2000-29, ISBN 90-73958-62-8
[3]  Spath, P. et al., "Fischer Tropsch Synthesis Technology Brief" (2002), http://devafdc.nrel.gov/bcfcdoc/6985.pdf
[4]  Amine website, http://www.amines.com/mdea_comp.htm
[5]  Hamelinck, C. et al., "Production of FT Transportation Fuels from Biomass; Technical Options, Process Analysis and Optimization, and Development Potential (2003), Utrecht University, NWS-E-2003-08, ISBN 90-393-3342-4
[6]  Spath, P. et al., "Biomass to Hydrogen Production Detailed Design and Economics Utilizing the Batelle Columbus  Laboratory Indirectly-Heated Gasifier" (2005), http://www.nrel.gov/docs/fy05osti/37408.pdf
[7]  Van Vliet, O. et al., "Fischer Tropsch Diesel Production in a Well-to-Wheel Perspective: A Carbon, Energy Flow and Cost Analysis."  Energy Conversion and Management 50 (2009) 855-876.
[8]  Larson, E. et al., "Large-Scale Gasification-Based Coproduction of Fuels and Electricity from Switchgrass."  Biofuels, Bioproducts and Biorefining 3 (2009) 174-194 (electronic version; contains report and supporting info).
[9]  "Fischer Tropsch Catalyst Test on Coal-Derived Synthesis Gas."  Syntroleum public document.
[10]  "Syntroleum Analyst Day" powerpoint (2006), http://www.syntroleum.com/Presentations/Analyst-Day-June-2006-Final.pdf
[11]  Leckel, Dieter.  "Low-Pressure Hydrocracking of Coal-Derived Fischer-Tropsch Waxes to Diesel."  Energy & Fuels 21 (2007) 1425-1431.
[12]  Reed, M. et al., "Baseline Technical and Economic Assessment of a Commercial Scale Fischer-Tropsch Liquids Facility" (2007).  DOE/NETL document 2007/1260.
[13]  Energy Information Administration price forecasts, http://www.eia.doe.gov/oiaf/forecasting.html
[14]  Personal communication with Rich Bain (NREL), 7/24/09.
[15]  From SRI report on US Paraffinic wax prices, http://www.sriconsulting.com/CEH/Public/Reports/595.5000/
[16]  Swanson, R. et al., "Techno-Economic Analysis of Biomass Gasification Scenarios" (2009), ConocoPhillips/Iowa State FT study.























FT model inlet/outlet heat and material balance summary
 
                            Energy Flow [MMBTU/hr]
                           Material Flowrate [lb/hr]
                              Major Energy Inlets
Biomass Feedstock (WET)
                                    -1699.4
                                    367437
Natural Gas
                                      0.0
                                       0
Air
                                     -78.3
                                    703501
Olivine
                                      0.0
                                      538
MgO
                                      0.0
                                       7
Water
                                    -1375.2
                                    199430
Tar Reforming Catalyst
                                      0.0
                                       1
Purchased Electricity
                                     -0.1
                                      NA
Other
                                      0.0
                                       3
Total
                                     -3153
                                    1270917
                             Major Energy Outlets
Diesel
                                     -21.9
                                     24550
Naphtha
                                     -9.6
                                     9885
Wax
                                     -1.3
                                     1620
Cooling Tower Evap.
                                    -591.6
                                    107139
Flue Gas
                                    -1938.3
                                    948007
Sulfur
                                      0.0
                                      133
CO2 Vent
                                    -313.0
                                     81664
Byproduct Electricity
                                      0.0
                                      NA
Air Coolers
                                     244.9
                                      NA
Compressor Heat
                                     135.1
                                      NA
Vent to Atmosphere
                                     -0.5
                                      331
Sand and Ash
                                     -2.2
                                     2428
Catalyst Purge
                                      0.0
                                       1
Wastewater
                                    -378.1
                                     55376
Ambient Heat and Work Losses
                                      3.1
                                      NA
Other
                                    -273.2
                                     39765
Total
                                     -3147
                                    1270900
% difference (Note 1)
                                     0.2%
                                     0.0%
   1.  Slight discrepancy due to thermodynamic property changes within model and overall convergence tolerance








FT model emission sources and flowrates [lb/h]

