Improving the Interface between Fischer Tropsch Synthesis and Refining

Transcription

1 This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes. pubs.acs.org/ef Improving the Interface between Fischer Tropsch Synthesis and Refining Daniel F. Rodríguez Vallejo, and Arno de Klerk*, Instituto de Energía Materiales y Medio Ambiente, Universidad Pontificia Bolivariana, Medellín, Circular , Colombia Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB T6G 2 V4, Canada *S Supporting Information Downloaded via on August 13, 2018 at 15:55:17 (UTC). See for options on how to legitimately share published articles. ABSTRACT: In a typical industrial Fischer Tropsch process the hot reaction product from Fischer Tropsch synthesis is stepwise cooled, condensed, and recovered, before being separated by distillation into different cuts. A different design was proposed whereby the hot reaction products are directly introduced into a pressure distillation unit, which combines syncrude recovery and distillation. Process simulation was employed to evaluate the proposal. It was found that integration of syncrude recovery and distillation was technically viable and that it had a number of benefits compared to stepwise cooling and recovery of syncrude prior to distillation. These positive outcomes were independent of the type of Fischer Tropsch technology used. Some notable benefits included a decrease in heating/cooling duty, improved liquid recovery, and reduced loading of tail gas separation. The proposed design also enabled other improvements, such as a strategy to improve catalyst-wax separation from slurry bubble column reactors and a strategy to reactively improve distillation performance and increase liquid yield. 1. INTRODUCTION Industrial gas-to-liquids (GTL) and coal-to-liquids (CTL) facilities are presently based on mainly indirect liquefaction technology. A generic indirect liquefaction process consists of four important elements: (1) generating syngas, either by natural gas reforming or by coal gasification, (2) cleaning and conditioning of the syngas, (3) converting the syngas into products, and (4) refining. It is a complex process. On its own, each of the four technology areas is well developed and efficient, but the same is not true of the interfaces. The inefficiency at the interfaces is best illustrated by the temperature differentials when material passes from one technology area to another (Figure 1). In this work we are focusing specifically on the interface between syngas conversion by Fischer Tropsch (FT) synthesis and FT product refining. The syncrude from FT synthesis is a complex product, with three to four different phases at ambient conditions: gas, organic liquid, aqueous liquid, and organic solid. The composition and temperature of the syncrude product that leaves the FT reactor depend on the FT technology. 1 2 The next step is to recover the syncrude from the unconverted syngas. Current industrial practice relies on a stepwise cooling and phase separation strategy. 3 In some designs part of the energy is recovered by feed-product heat exchange, but for the most part the energy is removed by water or air cooling. Once the syncrude is separated from the unconverted syngas, the syncrude is refined, employing refining techniques analogous to that found in crude oil refineries. The first step is fractionation of the syncrude in an atmospheric distillation unit (ADU), with the different syncrude-cuts sent to appropriate refining units, such as oligomerization, hydrotreating, and hydrocracking. The interface between syngas conversion and syncrude refining is therefore the syncrude recovery section, which is between FT synthesis and the ADU (Figure 1). In a crude oil refinery the interface between crude oil production and crude oil refining is physically separated, often by thousands of kilometers. The ADU is seen as the entry point into the refinery, even though in practice the ADU in a crude oil refinery is usually preceded by a desalter unit. The operation of an ADU is pertinent to this work and will be briefly described. The ADU in a crude oil refinery does not employ a standard reboiler, as is found in standard distillation columns. 4 An ADU employs a fired feed preheater. The hot partially vaporized crude oil leaving the feed preheater is then allowed to dissipate the added heat to drive the distillation process in a conventional way but without a separate reboiler. In a typical crude oil refinery, the ADU accounts for about 40% of the total energy consumption during refining. 5 The large amount of energy consumed by the ADU fired preheater is understandable: it is the entry point of the refinery, which represents the highest flow rate of material, and the liquid feed must be heated up and partially vaporized from near ambient conditions. In a FT facility, the division between synthetic crude oil production and refining is arbitrary, and the interface does not have to be segregated. Current design practice to perform stepwise cooling and syncrude recovery, only to be followed by reheating the feed before the ADU. This seems wasteful. Furthermore, the purpose of syncrude recovery and that of the ADU is very similar, namely, to separate syncrude from syngas and to separate syncrude into useful fractions for further refining. By introducing the hot product from FT synthesis directly into a distillation unit, it may be possible to reduce the overall energy need for syncrude separation. Considering the high consumption of energy by a refinery ADU, reducing the Received: March 29, 2013 Revised: May 10, 2013 Published: June 7, American Chemical Society 3137

2 Figure 1. Temperature differentials between technology areas illustrated by a coal-to-liquids design employing high temperature Fischer Tropsch synthesis. The downstream refinery after the ADU is not shown. size, or even eliminating the preheater before the distillation unit, will have a meaningful impact on the overall energyrelated environmental footprint of synthetic fuel refining. It is the purpose of this work to investigate the feasibility of improving the interface between FT synthesis and refining. It is suggested that the stepwise cooling, syncrude recovery, and atmospheric distillation steps can be combined into a single unit. By directly feeding the syncrude into a distillation unit, the energy can be dissipated through product fractionation, thereby avoiding successive cooling and heating by utility streams. 2. DESIGN CONCEPTS FT technologies can be broadly speaking classified into two categories (Table 1), which are different in important ways for the proposed integration of syncrude recovery and product fractionation. 3 Table 1. Characteristics Typical of High Temperature Fischer Tropsch (HTFT) and Low Temperature Fischer Tropsch (LTFT) Processes description HTFT LTFT operating temperature ( C) a operating pressure (MPa) reaction phase gas gas+liquid Fischer Tropsch catalyst Fe Fe or Co organic product distribution (wt%) b light gas (C 1 C 2 ) liquid petroleum gas (C 3 C 4 ) naphtha (C C) distillate ( C) residue/wax (>360 C) < water-soluble oxygenates 5 10 <5 a The high temperature slurry Fischer Tropsch process (HTSFTP) that operates at 270 C. Water is coproduced with the organic product during Fischer Tropsch synthesis. High temperature Fischer Tropsch (HTFT) synthesis employs a fused Fe-based catalyst and synthesis takes place in the gas phase at around 340 C and 2.5 MPa. At reaction conditions no liquid phase is present in the reactor. The HTFT syncrude is rich in low boiling material, with <5 wt % of the organic material boiling above 360 C. The heavy material (atmospheric residue) is rich in olefinic and aromatic compounds. Low temperature Fischer Tropsch (LTFT) synthesis employs either Fe- or Co-based catalysts, and the reaction medium contains both liquid and vapor phases. The operating temperature for Fe-LTFT typically varies from 200 to 250 C, but higher temperatures are possible, such as in the high temperature slurry Fischer Tropsch process (HTSFTP) developed by the Chinese Academy of Sciences and Synfuels China. 2 The operating temperature for Co-LTFT is usually in the range C. Irrespective of LTFT technology and operating conditions, LTFT syncrude is rich in high boiling material, with >40 wt % of the organic product boiling above 360 C. The heavy material is a paraffinic wax High Temperature Fischer Tropsch (HTFT) Design Concept. The stepwise cooling of the reactor product from HTFT synthesis and syncrude recovery that is typical of industrial operation is shown in Figure 2. Although this is a generic flow diagram, the main design elements are the same, irrespective of whether HTFT synthesis was performed in a fixed fluidized bed reactor or in a circulating fluidized bed reactor. 3 The first step is to cool the hot gaseous product, typically by feed-product heat exchange, to around 150 C. This is sufficient to condense the heaviest oil fraction. The heavy oil is used as a washing medium to remove the small amount of catalyst that was not removed by the cyclones in the HTFT reactor from the stream. The solids contaminated heavy oil is allowed to settle, producing clear oil, called decanted oil, and a small waste stream containing heavy oil and solids, called gunk. The decanted oil has a broad carbon number distribution, typically C 11 C Figure 2. Configuration for stepwise cooling and syncrude recovery from a typical industrial HTFT facility. 3138

3 The second step is to cool down the gaseous product further to condense the lighter oil fraction and the aqueous product. On a molar basis about half of the product from Fischer Tropsch synthesis is water and on a mass basis it is even more, so that this step produces most of the liquid product. The temperature to which the stream is cooled down has an impact on downstream refining, because it determines the composition of the light oil and the oxygenate partitioning between the light oil and aqueous product phases. 3 This is equivalent to a single equilibrium stage, and the light oil has a broad carbon number distribution, typically C 4 C The aqueous product, which is also called reaction water, contains most of the polar oxygenates in the C 1 C 4 range. The efficiency of oxygenate partitioning is better at lower temperatures, with industrial separation taking place in the range C. The tail gas is still laden with organic products in the C 1 C 7 range. 7 Further cooling and separation is performed on the tail gas to recover the light hydrocarbons. In the oil refinery the light oil and decanted oil is separated using a standard ADU. In its simplest form it is suggested to replace the design shown in Figure 2 with feeding the hot vapor phase reactor product directly into a distillation unit (Figure 3). The design industrial designs for primary separation of the Fischer Tropsch aqueous product. 3 (b) The vapor traffic at the feed position in the bottom of the column is the total volumetric product from HTFT synthesis. Operating under pressure is important to keep the column diameter as small as practicable, as well as to avoid tail gas recompression. (c) The column design is complex, because there are two liquid phases present on most trays. (d) Tray design must make provision for a small amount of solids in the feed as well as the surface-active nature of the product, which may result in foaming. (e) Even though the bottom product contains solids and will therefore require further separation, the distillate/gas oil is a small fraction of the overall HTFT syncrude product (Table 1). (f) The overhead condenser temperature is likely to be determined by the most cost-effective utility for cooling due to the high condenser duty. In Figure 3 the overhead temperature was chosen based on cooling water as utility Low Temperature Fischer Tropsch (LTFT) Design Concept. Unlike HTFT synthesis, the LTFT reaction product contains a liquid phase product at reaction conditions. The product is phase separated in the reactor, and the solids content of the liquid product depends on the Fischer Tropsch technology employed. The liquid product (wax) from synthesis in a slurry bubble column reactor contains some solids, whereas the product from fixed bed reactor synthesis is almost free of solids. A typical industrial LTFT stepwise cooling and syncrude recovery process is shown in Figure 4. Figure 3. Simplified process flow diagram of the direct integration of a distillation unit for cooling and recovery of syncrude from the reactor product of HTFT synthesis. shown is a simplified configuration and in practice the pressure distillation unit will be designed with side-draws to obtain different naphtha and potentially kerosene cuts. The distillation unit (Figure 3) has neither a reboiler nor a feed preheater. It is not an ADU but a pressure distillation, so that the bottom product is not an atmospheric residue but a distillate that includes the solids carryover from HTFT synthesis. All of the feed is already in the vapor phase and has the highest latent energy content possible. Not only does this avoid any reboiler temperature constraints typical of syncrude distillation 3 but also has other potential benefits, such as sharper cuts, better light oil recovery, more equilibrium stages for organic-aqueous phase product partitioning, and potentially a better removal of carboxylic acids, metal carboxylates, and solids materials. There are some salient issues to point out for the implementation of such a concept: (a) The feed is corrosive, and the column metallurgy must be appropriately selected. An appropriate base may be considered as cofeed to neutralize the carboxylic acids (and absorb some CO 2 ), as was employed to reduce corrosion in some of the Figure 4. Generic industrial fixed bed LTFT cooling and syncrude recovery process. It is possible to condense the oil product from the vapor phase reactor product in two steps, but only a single cooling and condensation step is shown in Figure 4. The light oil, also called condensate, is phase separated from the aqueous product and uncondensed vapor product. The cooling and syncrude recovery is analogous to the second cooling and recovery step after HTFT synthesis, and the discussion will not be repeated. The condensation temperature affects the product partitioning between the organic and aqueous phases, but it is less critical because LTFT syncrude usually contains less light oxygenates. Replacing the LTFT cooling and syncrude recovery design in Figure 4 with cooling and recovery in a distillation unit (Figure 5) is at the detail level different from that for HTFT. Foremost are the lower feed temperature and the presence of part of the syncrude as a liquid product. The liquid reflux from the top of the column is supplemented by the liquid wax feed. The wax is fed at a position above the gaseous reactor product feed, so that some stripping of lighter compounds in the wax may take place to produce a sharper cut. The temperature of the feed restricts the distillation cut point and also affects the energy content available for distillation. 3139

4 Figure 5. Simplified process flow diagram of the direct integration of a distillation unit for cooling and recovery of syncrude from the reactor product of slurry phase LTFT synthesis. The potential benefit from applying direct distillation instead of cooling and recovery to LTFT syncrude has less benefit than for HTFT on account of the lower operating temperature. In the case of LTFT synthesis in slurry bubble column reactors, the distillation may be modified to include a catalyst separation step (Figure 6). Separation of catalyst from wax is challenging, 3 Figure 6. Details of pressure distillation column tray configuration to facilitate primary recovery of catalyst in the product from a slurry bubble column LTFT reactor for return to the reactor. and adding an additional separation step may improve the overall performance of solids removal from the wax. The separation occurs by settling in the dead-area of each sieve tray below the wax feed tray. Since the catalyst particles are quite fine (<100 μm), settling is not very efficient. Catalyst recovery from wax in the column may be better applied as primary separation step to reduce the solids loading of subsequent catalyst removal technologies, which can be applied for solids separation from the column bottoms (Figure 5). 3. PROCESS MODELING The design concepts were evaluated using the commercially available process simulation software HYSYS. The process flow diagrams constructed in the process simulator to represent the designs in Figures 2 5 are shown in the Supporting Information (Figures S1 S4). In principle any of the reputable process simulators can be employed as long as an appropriate thermodynamic model is selected. The reactor product from Fischer Tropsch synthesis contains mainly water gas shift products (H 2, CO, CO 2, and H 2 O), oxygenates and hydrocarbons (Table 1). In order to model Fischer Tropsch syncrude condensation and phase separation, the thermodynamic model must be capable of accurately predicting the vapor liquid liquid equilibrium (VLLE) data of a system that contains water, oxygenates, and hydrocarbons Selection of the Thermodynamic Model. The Peng Robinson Stryjek Vera (PRSV) cubic equation of state (EOS) is the thermodynamic model selected to represent the multicomponent equilibrium of the system studied. This equation is a 2-fold modification of the original EOS (eqs 1 4) proposed by Peng and Robinson 8 P = where RT a V b VV ( + b) + bv ( b) (1) N b = xb i i i= 1 (2) RTci bi = Pci (3) and a N N = xxa i j ij 1 1 (4) i= j= The first modification consists of a different set of mixing rules for mixtures. The mixing rule for the cross term, a ij, adopts a composition dependence form (eqs 5 7), which in HYSYS is computed according to the Margules expression and ij i j 0.5 i ij j ji ij ji a = ( aa) (1 xk xk ) k k where a = a α a (5) i ci i (6) ci RT = ( ci ) P ci 2 The second modification involves the variation of the term α i with the acentric factor, ω i, and an empirical factor, k 1i, which is used to fit pure component vapor pressures (eqs 8 10). 9 α i = [1 + k i(1 T ri 0.5 )] 2 (8) k = k + k (1 + T )(0.7 T ) i 0i 1i ri 0.5 ri 0.5 (9) k0i = ωi ωi ω i 2 (7) (10) 3140

5 Table 2. Bubble Point Temperatures from the Literature (T B ) and Calculated by the PRSV Model (T B(PRSV) ) for the VLE of the Water (1), n-butanol (2), and n-hexane (3) Ternary System a liquid phase mole frac vapor phase mole frac temp (K) x 1 x 2 x 3 y 1 y 2 y 3 T B T B(PRSV) % error a The liquid phase mole fractions, x i, and vapor phase mole fractions, y i, at pressure kpa are shown. These modifications extend the application of the Peng Robinson EOS to model the thermodynamic equilibrium of moderately nonideal systems. It has been successfully employed in nonideal systems, giving results that are as good as methods based on excess Gibbs energy functions like Wilson, NRTL, and UNIQUAC. The advantages of the PRSV equation include the following: (a) It predicts the phase behavior of hydrocarbon systems more accurately, particularly those that involve dissimilar components. (b) Its application can be extended to model the phase behavior of nonideal systems, and it performs at least as good as traditional activity coefficient models. The model contained in the commercial software HYSYS performs rigorous three-phase flash calculations that can handle aqueous systems containing water, methanol, or glycols as well as systems containing hydrocarbons and non-hydrocarbons in the second liquid phase Evaluation of Model Performance. The experimental ternary VLE data of the system water/n-butanol/nhexane reported in the study performed by Gomis et al. 10 was employed to verify the performance of the PRSV thermodynamic model with a system containing water, oxygenates, and hydrocarbons. Model performance was evaluated in the following way. For a given phase composition the temperature prediction of the PRSV thermodynamic model, T B(PRSV), was compared to 3141

6 the experimentally reported bubble point temperatures, T B,of Gomis et al. 10 (Table 2). The PRSV calculations had a positive temperature bias, and for all compositions T B(PRSV) > T B. The average relative error in temperature prediction was 1.3%. This indicated that the PRSV thermodynamic model was likely to provide an adequate description of the behavior of mixtures containing water, oxygenates, and hydrocarbons Modeling of Fischer Tropsch Synthesis. The HTFT and LTFT processes were modeled with a syngas feed having a H 2 :CO ratio of 1:1. This is typical of gasification processes employing feed materials with a low effective H:C ratio, such as coal, and that are operated with a high syngas outlet temperature. 11 Consequently Fe-based catalysts were selected for Fischer Tropsch synthesis, because such catalysts are active for catalyzing the water gas shift (WGS) reaction. Instead of using published selectivity data on Fischer Tropsch syncrude, it was decided to make use of the built-in HYSYS model for Fischer Tropsch synthesis. There were three reasons for doing so: (a) The specific kinetics of Fischer Tropsch synthesis was not relevant. (b) It provided an independent description of Fischer Tropsch syncrude selectivity that is not associated with any specific Fischer Tropsch catalyst or technology. (c) The model description kept the complexity manageable for process simulation purposes. The model described the reaction network during Fischer Tropsch synthesis in terms of a limited number of reactions. It considers the production of hydrocarbons: α-olefins (eq 11) and paraffins (eq 12) with carbon numbers ranging from C 1 to C 30. It also considers the WGS reaction (eq 13). nco+ 2nH2 CnH2n + nh2 O (11) nco + (2n + 1)H C H + nh O 2 n 2n+ 2 2 (12) CO + H2O H2 + CO2 (13) The reaction kinetics are computed according to the model proposed by Lox and Froment, 12 with the rate expressions repeated here for ease of reference (eqs 14 16): R R R CH n 2n = CH n 2n+ 2 = k HC6 1 + = khc1pco khc1pco + khc5ph 2 α 1 n 1 n = 2,..., 30 khc1pco 1 khc1pco + khc5ph 2 α (14) k HC5PH PCOPH k 2 O v 0.5 PH Kv khc1pco khc1pco + khc5ph 2 α 1 1 n 1 n = 1,..., 30 khc1pco khc1pco + khc5ph 2 α (15) 0.5 PCO 2 PH 2 KWGS CO2 2 PH 2 O 0.5 PH 2 (16) This model incorporates an expression to calculate the probability chain growth, α, which does take into account the apparent change in α-value of LTFT syncrude with carbon number. 13 In this study the carbon number distribution was modified to incorporate two α-values. The operating conditions employed in the simulation of Fischer Tropsch synthesis are summarized in Table 3. Table 3. Operating Conditions and Reaction Descriptors Employed in the Simulation of Fe-Based Fischer Tropsch Synthesis description HTFT LTFT operating temperature ( C) operating pressure (MPa) syngas H 2 :CO (mol/mol) 1:1 1:1 syngas conversion (mol %) carbon number distribution α 1 a 0.5 α a HTFT has only a single-α-value distribution, with higher C 1 and lower C 2. The model description had some limitations, which should be borne in mind when interpreting the results from the process simulations: (a) The carbon number distribution was limited to C 30. This is an acceptable carbon number range for HTFT but not for LTFT where carbon numbers higher than C 100 can be produced. 2,3 (b) The carbon number distribution during Fischer Tropsch synthesis is affected by temperature. However, this temperature dependence was not reflected in the model, and the α-values for LTFT synthesis were kept constant. This was deliberate so that the effect of process temperature could be studied independently from differences in LTFT syncrude composition. (c) The reaction network considered only hydrocarbon synthesis. Oxygenates play a significant role in nonideal VLLE of real Fischer Tropsch syncrude. Oxygenate recovery and partitioning affects refining efficiency 3 and could not be evaluated with this model. (d) Other syncrude characteristics that were not captured in this work are aliphatic branching, olefinic isomers, and aromatics. This simplification was not likely to affect the outcome of the comparison. (e) The constant syngas conversion of 85% is realistic for single-reactor HTFT synthesis, but it is usually not a realistic assumption for single-reactor LTFT synthesis. 2,3 However, since a low H 2 :CO feed was employed, the water partial pressure at high conversion is moderated by the WGS equilibrium (eq 13) making high conversion feasible. (f) Only a comparative evaluation of the proposed concept relative to the base case is possible, because of the assumptions made in the description of Fischer Tropsch synthesis Modeling of Stepwise Cooling and Syncrude Recovery. The base case stepwise cooling and syncrude recovery (Figures 2 and 4) are performed using two three phase separators, each one preceded by a cooler (Figures S1 and S2). The software performs rigorous three phase flash calculations, which generate three exit streams at thermodynamic equilibrium. In the first separator, the temperature is adjusted to obtain an 80% recovery of n-decane in the liquid organic phase as well as some aqueous product. The remaining gaseous product is further cooled to 35 C and sent to the second separator to generate a lighter organic phase, aqueous product, and tail gas streams. 3142

7 After the stepwise cooling is performed, the organic products are mixed and sent to an ADU (Figures S1 and S2). The ADU is modeled as a common distillation tower with reboiler, partial condenser, and reflux return to the column Modeling of Integrated Distillation and Syncrude Recovery. The proposed case that involved combined distillation and syncrude recovery (Figures 3 and 5) did not employ cooling or separation of the reactor product before it was introduced into the pressure distillation unit. The process flow diagram in the simulation represented this directly (Figures S3 and S4). The vapor phase product from the HTFT or LTFT reactor goes directly to the bottom of the distillation unit, replacing the vapor product conventionally produced by a reboiler, which is absent from this column. The liquid phase reactor product, which is only produced during LTFT synthesis, is fed higher up in the column. The performance of the distillation unit was not sensitive to the liquid feed position, although feeding the liquid product at the top of the column was marginally better than feeding at a lower position in the column, as discussed in the Supporting Information (Table S1). 4. RESULTS AND DISCUSSION 4.1. Comparison of Cases for HTFT Syncrude Recovery and Distillation. The benefit of the proposed case, which integrates HTFT syncrude recovery and fractionation (Figure 3) with respect to the base case, which employs stepwise cooling and recovery (Figure 2), was determined. In order to perform the comparison on the same basis, the base case calculations included the refinery ADU (Figure S1). For an industrial ADU the rule of thumb is that trays are required between each side-draw. 14 For the sake of simplicity, the number of trays were fixed at 20 trays, and no side-draws were included in the distillation columns of either case, even though side-draws would form part of a practical design. A single cut between naphtha and distillate at a normal boiling point of 175 C was considered. The outcome of the comparison of heating and cooling duties is provided in Table 4. The main benefit of the proposed case is in the heating requirements. The base case requires a heating duty of 56 kj kg 1, which needs to be provided by a fired heater (furnace) associated with the refinery ADU. Not only is the feed preheater to the ADU more costly than a simple heat exchanger or reboiler but also the energy input that is required is more than the duty. The cost ratio at equal duty Table 4. Heating and Cooling Duties Associated with the Base Case and Proposed Case for HTFT with a Reactor Exit Temperature of 340 C duty (kj/kg) a description of equipment base case proposed case heating duties reboiler (distillation) 56 - b cooling duties cooler b cooler b condenser (distillation) total cooling duty: a Energy flow per mass of total reactor product, that is hydrocarbons (24 wt %), water, CO 2, and unconverted syngas. b No duty, because the equipment is not needed. of a fired heater compared to a kettle reboiler (shell and tube exchanger) is around 5:1. 15 Furthermore, the thermal efficiency is typically in the range of 80 90%; 15 not all the energy released to produce the hot combustion gases can be recovered by heat exchange. In the proposed case, when syncrude recovery and fractionation are integrated, the cooling duty is 26 kj kg 1 higher than in the base case. This is an oversimplification, because part of the duty in the first cooler can be beneficially employed for heating. The value of the energy depends on the temperature at which heat exchange takes place, and this stream has an initial temperature of 340 C. In the alternative case it may not be necessary to use the stream at 340 C as feed for distillation, and some of that energy may be beneficially recovered, thereby decreasing the cooling duty of the alternative case. Realistically the difference in cooling duty between the two cases is more than is indicated, but energy optimization was not performed. A further difference between the two cases that affects the design and cost is in the nature of the cooling media that can be used. On account of the higher average temperature at which cooling takes place in the base case, air cooling can be used for part of the cooling duty. Overall the operating and capital cost associated with cooling is meaningfully lower in the base case than in the proposed case. The product separation and recovery efficiency was compared (Table 5). There are two aspects of particular importance. The first is the recovery of light organic products as liquids, and the second is the quality of the separation. Table 5. HTFT Syncrude Recovery and Separation for the Base Case and Proposed Case Expressed in Terms of Total Hydrocarbon Mass Flow Rate base case (kg/kg hydrocarbons) hydrocarbon products tail gas naphtha proposed case (kg/kg hydrocarbons) distil late tail gas naphtha distil late C 1 (methane) C 2 C 4 (light gases) C 5 C (naphtha) C 11 C 22 (distil late) C 23 C (residue) total: The recovery of liquid hydrocarbon products during the proposed integration of syncrude recovery and distillation is 46 wt %, which is better than the 40 wt % recovery in the base case (Table 5). This is not surprising, because in the base case the separation of the tail gas from the liquid products is effectively limited to a single equilibrium stage in the second phase separator (Figure 2). The recovered liquid includes some dissolved light gases, but the picture does not change when only the naphtha (C 5 C 10 ) is considered. In the proposed case the naphtha recovery is 79 wt %, whereas the base case naphtha recovery is 67 wt % (Table 5). Poor C 5 C 7 HTFT naphtha recovery is indeed industrially observed for a typical base case design. 3,7 A further consequence of the poorer naphtha recovery in the base case is that recovery of the lighter products from the tail gas becomes more onerous. This affects the overall downstream heating and cooling requirements, 3143

8 Table 6. Heating and Cooling Duties Associated with the Base Case and Proposed Case for LTFT with Different Reactor Exit Temperatures As Indicated duty (kj/kg) a LTFT at 200 C LTFT at 220 C LTFT at 240 C LTFT at 270 C description of equipment base case proposed base case proposed base case proposed base case proposed heating duties reboiler (distillation) b b b b cooling duties cooler b b b b cooler b b b b condenser (distillation) total cooling duty: a Energy flow per mass of total reactor product, that is hydrocarbons, water, CO 2, and unconverted syngas. b No duty, because the equipment is not needed. which is not taken into consideration in the comparison given in Table 4. The quality of the separation is not independent from the liquid recovery (Table 5). The sharpness of the naphthadistillate cutpoint is very similar between the two cases. The main difference is in the composition of the naphtha fraction, also referred to as unstabilized light oil (ULO). Due to the higher naphtha recovery in the proposed case, the naphtha contains more light straight run naphtha. It also contains more dissolved lighter (C 2 C 4 ) hydrocarbons, which is the consequence of pressure distillation at 2.5 MPa as opposed to atmospheric distillation. In neither of the cases was much methane retained in the naphtha. For HTFT synthesis the main advantages of the proposed integration of syncrude recovery and distillation over the base case that involves stepwise cooling, recovery, and separation by ADU in the refinery are as follows: (a) Less equipment and unit operations are required. (b) In particular the capital and operating cost associated with a fired heater for the ADU is avoided. (c) Overall liquid recovery improved by 6 wt %, and the naphtha recovery in particular was 12 wt % higher than in the base case. (d) The tail gas separation requirements become less onerous, which also translates into an operating cost saving. Since the tail gas flow is more than half the hydrocarbon flow, this is a significant potential saving The main disadvantage of the proposed case is the increased cooling duty Comparison of Cases for LTFT Syncrude Recovery and Distillation. The operating range for LTFT synthesis is wider than for HTFT synthesis. It was anticipated that the benefit of the proposed integration of syncrude recovery and distillation (Figure 5) may decrease as the temperature of the product from LTFT synthesis decreased. In this work we considered operating temperatures over the range C. When LTFT synthesis is used for the production of waxes, wax-products, or lubricant base oils, operation at lower temperature is preferred, because it enables higher wax yields. When LTFT synthesis is used in energy applications, there is benefit in operating at higher temperatures to improve the quality of the steam coproduced. 2 The effect of temperature on the carbon number distribution was not considered, and all of the LTFT cases were evaluated with the same carbon number distribution. However, the WGS equilibrium changed with temperature, and this caused a minor change in the product composition. The same number of trays in the distillation (20 trays) and same cutpoint between naphtha and distillate (175 C normal boiling point) were employed as for the HTFT cases. A comparison between the HTFT and LTFT cases is therefore in principle possible. The heating and cooling duties of the base case and proposed case are given (Table 6). The proposed integration of syncrude recovery and distillation had a benefit both in the heating and cooling duties over the base case involving stepwise cooling, recovery, and distillation. It was counterintuitive to find that as the operating temperature of LTFT synthesis increased, the heating duty of the base case ADU (Figure S2) also increased. This was a consequence of an increase in the required reflux ratio of the ADU to meet the same cut point specification. Even though the combined liquid product formed the feed in each instance, and the reactor product was forced to be the same (α-value was not changed with LTFT temperature), the efficiency of the stepwise cooling and syncrude recovery was affected. The dominant effect was that of the higher temperature of the initial vapor liquid equilibrium in the reactor. It did not meaningfully affect the overall syncrude recovery and sharpness of the cuts (Table 7), but it came at the cost of a more demanding separation in the ADU. With an increase in LTFT temperature, the cooling duty increased. This reflects the higher total amount of energy to be removed as well as the increased vapor fraction of the product. At the lowest LTFT synthesis temperature, the difference in cooling duties between the base case and proposed case is 28 kj kg 1. If the increased cooling duty of the base case due to the increased ADU reboiler duty is discounted, there is only a minor increase in the difference of the cooling duties between the two cases with an increase in LTFT temperature. Overall the operating and capital cost associated with cooling is lower in the proposed case than in the base case. This makes the proposed case more efficient for LTFT both in terms of heating and cooling duty. The same final product distribution was obtained for the base case LTFT separation, irrespective of the LTFT operating temperature (Table 7). In practice this will not be the case, because the α-value of Fischer Tropsch synthesis will decrease with increasing synthesis temperature. In the present work, by keeping the α-value constant, it was possible to show that the separation and syncrude recovery in the base case design was 3144

9 Table 7. LTFT Syncrude Recovery and Separation for the Base Case and Proposed Case Expressed in Terms of Total Hydrocarbon Mass Flow Rate for Different LTFT Operating Temperatures base case (kg/kg hydro carbons) hydrocarbon products tail gas naphtha proposed case (kg/kg hydrocarbons) distil late tail gas naphtha distil late LTFT at 200 C C C 2 C C 5 C C 11 C C 23 C total: LTFT at 220 C C C 2 C C 5 C C 11 C C 23 C total: LTFT at 240 C C C 2 C C 5 C C 11 C C 23 C total: LTFT at 270 C C C 2 C C 5 C C 11 C C 23 C total: not sensitive to the reactor temperature. For the proposed case, an increase in reactor temperature caused more products to be in the vapor phase and all of the products to be at a higher temperature. The energy available for integrated distillation and recovery of the syncrude consequently increased with increasing LTFT temperature, making it more efficient (Table 7). The naphtha recovery improved from 84 to 87 wt % as the LTFT temperature was increased from 200 to 270 C. The naphtha recovery in the proposed case was also much better than the naphtha recovery of the base case, which was 74 wt %. The improvement in overall liquid recovery was not as much, 83 versus 80 wt %, because the naphtha is a smaller fraction of the total LTFT syncrude product than in the case of HTFT synthesis. The sharpness of the cut in the proposed case was not as good as in the base case (Table 7), because there was less energy available for distillation. The smaller relative volatility difference during pressure distillation played a minor role, judging by the sharpness of cut that could be obtained with HTFT syncrude (Table 5). However, pressure distillation caused some of the light gases to be dissolved in the liquid products, which is undesirable. Even though it will add to the cost, it may be preferable to add a preheater before the pressure distillation unit to increase the inlet temperature of the feed and thereby obtain sharper cuts. This modification was not evaluated. Another approach that can be considered to increase the energy available for distillation is to increase the temperature at the bottom of the column through reaction. A catalytic section can be accommodated at the bottom of the pressure distillation unit (Figure 7). It is preferable to remove the liquid bottom Figure 7. Incorporation of exothermic catalytic distillation to increase the temperature of the vapor feed in the proposed process for the integration of syncrude recovery and distillation. product from the distillation above the catalyst bed, because the liquid may contain some materials that can over time cause catalyst fouling. By exposing the catalyst only to the vapor phase product from the Fischer Tropsch reactor, this is avoided. The nature and the extent of the exothermic reaction can be selected and moderated through an appropriate choice of catalyst. The main reactive species present in the vapor feed are H 2, CO, olefins, and oxygenates. Some of the candidate catalytic conversions that can be considered are hydrogenation, syngas conversion, and oligomerization. It is possible to perform a mild hydrogenation, but it is likely that methanation will be the main reaction. Methanation of CO is very exothermic (206 kj mol 1 ), and it would increase the bottom temperature. Usually methane is an undesirable product and methanation will not be a recommended reaction, because it undermines the Fischer Tropsch selectivity profile. However, there are instances where this may be considered, such as an open gas loop design or applications where the main use of the tail gas is as fuel gas. In addition to Fischer Tropsch synthesis, there are various other syngas conversions possible. 2 For example, methanol synthesis over a Cu-ZnO/Al 2 O 3 catalyst is possible if the inlet temperature is 240 C or higher, even though the pressure it too low to make this an efficient conversion. Methanol synthesis is exothermic (130 kj mol 1 ) and depending on the reaction extent, it may work well to increase the bottom temperature of the column. A water tolerant oligomerization catalyst can be employed to convert the olefins in the feed to heavier olefinic products. This reaction is exothermic ( kj mol 1 ), and oligomerization can improve the liquid product yield by converting normally gaseous olefins into liquid products. The increase of LTFT 3145

10 distillate yield by oligomerization is known and practiced, 3 although not in the way proposed here. The potential benefitof oligomerization is therefore 2-fold, namely, to increase the bottom temperature of the column and to increase the liquid yield. The additional benefit of employing a catalytic distillation section to increase the temperature at the bottom of the pressure distillation unit was not evaluated further. The main advantages of the proposed integration of syncrude recovery and distillation after LTFT synthesis compared to stepwise cooling, recovery, and atmospheric distillation are as follows: (a) Less equipment and unit operations are required. (b) There is no heating duty, and the cooling duty is less than in the base case. (c) Better naphtha recovery is obtained in the liquid product and the overall liquid recovery improved by 3 wt %. The main disadvantage of the proposed case is the poorer quality of the distillation cuts compared to that obtained by an equivalent ADU. 5. CONCLUSIONS The purpose of the investigation was to demonstrate the technical viability and advantages of integrating Fischer Tropsch syncrude recovery and pressure distillation unit (proposed case) compared to the current industrial practice of stepwise cooling, recovery, and atmospheric distillation unit (base case). The study was comparative in nature, and the results are best valued qualitatively. Although the advantages and disadvantages were quantified, these values are subject to the simplifying assumptions that were made in order to model the different processes. Care was taken to verify the validity of the thermodynamic model (Peng Robinson Stryjek Vera cubic equation of state) to describe the nonideal nature of the process streams, but the representation of Fischer Tropsch syncrude in the model only approximated the complexity of real syncrude. The most important outcomes of the investigation were as follows: (a) Process simulation indicated that the proposed case is technically viable and that it required less equipment and unit operations than the base case. These positive outcomes were independent of the type of Fischer Tropsch technology used. (b) Application of the proposed case to HTFT synthesis resulted in a lower heating duty (56 kj kg 1 ) by avoiding the use of a fired heater that is required by the atmospheric distillation unit in the base case. Additionally, the overall liquid recovery was improved by 6 wt %, mainly due to the 12 wt % higher naphtha recovery compared to the base case. The better naphtha recovery in turn reduced the load on tail gas separation. (c) The proposed case resulted in lower heating and cooling duties when applied with LTFT technology. As the operating temperature of the LTFT technology was increased from 200 to 270 C, the difference in heating duty increased from 150 to 250 kj kg 1 and the difference in cooling duty increased from 27 to 137 kj kg 1. The proposed case also improved the liquid recovery by 3 wt %. (d) A modified distillation tray design was suggested for the proposed case to improve catalyst-wax separation after LTFT synthesis and specifically for implementation in conjunction with slurry bubble column reactors (e) The main disadvantages of integrating syncrude recovery and distillation are higher cooling duty (26 kj kg 1 ) when applied to HTFT technology and less sharp distillation cuts when applied to LTFT technology. The latter was mainly due to the lower temperature of the feed going to the pressure distillation unit. (f) An exothermic catalytic distillation modification of the proposed case was suggested to increase the temperature of the vapor feed in the column and to improve liquid yield. Hydrogenation, syngas conversion, and oligomerization were the candidate catalytic conversions considered. ASSOCIATED CONTENT *S Supporting Information Details on the coding used in the process simulator are provided, including the process flow diagrams of the different case studies (Figures S1 S4). An evaluation of the feed stage for liquid LTFT product during integrated syncrude recovery and separation was presented (Table S1). This material is available free of charge via the Internet at AUTHOR INFORMATION Corresponding Author *Tel: , deklerk@ualberta.ca. Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS This work was made possible through a proof of concept grant of the Canada School of Energy and Environment (CSEE). The contribution of one of the authors was further supported through a scholarship received from ENLAZA-MUNDOS of Medellıń Mayor and Joven Investigador of COLCIENCIAS REFERENCES (1) Fischer Tropsch Technology (Stud. Surf. Sci. Catal. 152); Steynberg, A. P., Dry, M. E., Eds.; Elsevier: Amsterdam, (2) Greener Fischer Tropsch processes for fuels and feedstocks; Maitlis, P. M., De Klerk, A., Eds.; Wiley-VCH: Weinheim, (3) De Klerk, A. Fischer Tropsch Refining; Wiley-VCH: Weinheim, (4) Gary, J. H.; Handwerk, G. E.; Kaiser, M. J. Petroleum refining. Technology and economics, 5th ed; Taylor Francis: Boca Raton, FL, (5) Sittig, M. Petroleum refining industry. Energy saving and environmental control; Noyes: Park Ridge, NJ, (6) Leckel, D. O. Diesel production from Fischer Tropsch: the past, the present, and new concepts. Energy Fuels 2009, 23, (7) De Klerk, A.; Furimsky, E. Catalysis in the refining of Fischer Tropsch syncrude; Royal Society of Chemistry: Cambridge, UK, (8) Peng, D.-Y.; Robinson, D. B. A new two-constant equation of state. Ind. Eng. Chem. Fundam. 1976, 15, (9) Stryjek, R.; Vera, J. H. PRSV: An improved Peng-Robinson equation of state for pure components and mixtures. Can. J. Chem. Eng. 1986, 64, (10) Gomis, V.; Font, A.; Saquete, M. D.; García-Cano, J. LLE, VLE and VLLE data for the water-n-butanol-n-hexane system at atmospheric pressure. Fluid Phase Equilib. 2012, 316, (11) Higman, C.; Van der Burgt, M. Gasification, 2 nd ed.; Elsevier: Amsterdam, (12) Lox, E. S.; Froment, G. F. Kinetics of the Fischer Tropsch reaction on a precipitated promoted iron catalyst. 2. Reaction kinetics. Ind. Eng. Chem. Res. 1993, 32,