To evaluate biorefinery concepts, Excel based standalone process models that include mass and energy balances, technoeconomic analyses (TEA) and greenhouse gas (GHG) emission analyses are built by following the methodology described in previous work (Garcia-Nunez et al., 2016). Data needed to build standalone process models of a PR and six SAF technologies are described in detail in Petroleum Refineries and Sustainable Aviation Fuel Scenario. Integration concepts of co-location and repurposing were applied to generate alternative scenarios for the evaluations of costs and environmental impact reductions.

Unlike a corn ethanol mill or sugarcane mill, the existing PRs are not concentrated in a specific region in the United States (US Energy Information Administration, 2016). For this work, it was assumed that the existing PRs were located within the Midwest (PADD 2) (US Energy Information Administration, 2016), which allowed us to take advantage of corn stover-based SAF technologies that have been developed in previous work (Tanzil et al., 2021b). A PR with an atmospheric distillation capacity of 120,000 BPD (barrels per day) was used as the existing baseline capacity (Sun et al., 2018). This refinery accommodates an atmospheric distillation column that produces gas, light naphtha, heavy naphtha, gas oil, and heavy bottoms (Gary et al., 2010; Sun et al., 2018). Light naphtha, heavy naphtha, and gas oil are further processed (hydrotreatment, hydrocracking, isomerization, and catalytic reforming) in the refinery to produce jet/kerosene, diesel, and gasoline (Gary et al., 2010; Sun et al., 2018). The historic significance of heavy bottoms or residuals as direct fuel for other industries has dwindled over the past few decades in response to new environmental regulations (Gary et al., 2010). Therefore, further sequential processing of bottoms via vacuum distillation, fluid catalytic cracking (FCC), coking, and hydrotreatment must be completed (Sun et al., 2018). A schematic of this complicated process is included in the Supplemental Information. Table 1 shows the capacity of the major processing units for the PR scale used in this study, as well as utility consumption (Sun et al., 2018).

The PR facility was assumed to be located in the Midwest, so the most abundant lignocellulosic feedstock (corn stover) in that region (National Corn Growers Ass, 2016) was chosen as the feedstock for the SAF technologies. The standalone SAF technologies were termed VB_A, ATJ_A, DSHC_A, FP_A, and GFT_A, where A denotes the respective standalone technology. The conceptualized novel process, ICGFT, which has been described in detail in previous work (Tanzil et al., 2021a) was also modeled as a standalone scenario (ICGFT_A). This work investigated the integration opportunities offered for lignocellulosic processes. Although triglyceride-based HEFA (hydroprocessed esters and fatty acids) processes are readily available for integration as intermediates because of the low oxygen content (Starck et al., 2016), HEFA-based integration is only under construction by World Energy on a commercial scale (Lane, 2019) at a California site [City of Paramount (2020)., 2020]. This study focused on lignocellulosic SAF processes, which pose challenges as a result of their high oxygen content in both the feedstock and intermediates. These challenges are addressed on a case-by-case basis.

In two previous studies (Tanzil et al., 2021b; Tanzil et al., 2021c) integration scenarios were formulated based on the existing facilities capital structure. However, an existing PR facility has both a larger capacity and higher capital costs (Gary et al., 2010) than an existing corn ethanol mill (Wallace et al., 2005; Kwiatkowski et al., 2006) or sugarcane mill (Tanzil et al., 2021c). The capacity of standalone SAF scenarios was determined by the co-processing capacity of a PR, which typically ranges between 5 wt% to 15 wt% of the co-processing material (Gary et al., 2010; Pinho et al., 2015). In this work, this range was used to calculate the SAF capacity of each technology so that the corresponding feedstock capacity (corn stover) was maintained at under 2,000 metric tons per day (MTD). This can be regarded as a viable commercial-scale feedstock capacity (Swanson et al., 2010; Humbird et al., 2011; Jones et al., 2013; Quinn and Davis, 2015). Therefore, the co-processing ratio differed from 6 to 15%. For example, GFT_A-derived Fischer–Tropsch (FT) products were co-processed at a co-processing ratio of 7% to maintain the initial feedstock capacity under 2000 MTD. DSHC_A-derived farnesene was co-processed at a ratio of 6% in the hydrocracker. Farnesene is a C-15 unsaturated hydrocarbon molecule that is hydrogenated and cracked in the hydrocracker. Both VB_A and ATJ_A had a 10% co-processing ratio. For FP_A, a 15% co-processing ratio was used to limit the feedstock capacity to 1274 MTD. Because of the proposed high fuel yield (Tanzil et al., 2021a), ICGFT_A had a significantly lower feedstock capacity.

Table 2 shows the calculated SAF capacities [million liters per year (MLY)], corn stover capacity, and fixed capital investment (FCI). Capacities and FCIs of these standalone facilities were scaled from process models built in previous studies (Tanzil et al., 2021a; Tanzil et al., 2021b).

Depending on the SAF process pathway, the biomass-derived intermediates were co-processed with heavy gas oil (HGO) in the following upgrading units: hydrotreater, hydrocracker, and FCC.

Two types of integration strategies were included in the analysis (de Jong et al., 2015; Tanzil et al., 2019; Tanzil et al., 2021b), namely, co-location and repurposing. Co-location strategies explored the infrastructure of an existing PR without interruption of the production of petroleum products (de Jong et al., 2015). In this work, co-located scenarios were divided into two categories. In the first category, scenarios that utilized only OSBL infrastructures were defined (Table 3).

These were not directly involved in the conversion process of crude oil to various fuel products, e.g., buildings, yard improvements, and some of the service facilities. The capacities of the five components of service facilities—steam generation, power substation, power distribution, water distribution, and product storage capacity—were subject to co-located integration strategies. Due to the high capacity of PR infrastructure and well-established technological identities, a 20% cutoff margin was assumed for these service facilities to share with any of the SAF technologies. The core management group of plant managers and engineers was also considered to be shared.

In the second category of co-located scenarios, the co-processing capabilities of a PR were utilized, and these are detailed in Sustainable Aviation Fuel Scenario. Therefore, scenarios involving co-processing of compatible intermediates were generated (Figure 1). In addition to the OSBL component of the first category, these scenarios represent the conversion process equipment located ISBL.

Table 3 lists the integrated scenarios, which utilize existing infrastructure for either co-location or co-processing as well as defining each scenario as either purchasing or self-generating electricity.

Because of the feedstock limitations, the crude oil capacity of 120,000 BPD chosen for co-location was too large for the repurposing scenarios. Therefore, the SAF capacity for the repurposed scenarios (Table 4) remained the same as that in the co-located scenarios. Lignocellulosic SAF technology requires additional equipment not included in a PR. However, two large advantages for repurposing a PR are the avoidance of power generation module costs and not having to purchase hydrogen from an external source; note that these are required for co-located scenarios. However, a repurposed PR needs to be valued and added as an FCI component in the repurposed scenarios. In this work, this component was calculated to be $72 MM$ from the literature (Lane, 2019) assuming the six-tenth rule of scaling.

The technical data that were required to build the material and energy flows of the studied processes are given in the Supplemental Information (Supplementary Tables S1A–S1E; Supplementary Tables S2A–S2F); these data were used to build the material and energy flows of both the standalone and integrated scenarios.

TEA included capital and operational cost estimations, followed by a financial analysis to determine the minimum fuel selling price (MFSP) of each scenario. In this work, the MFSP was estimated for the SAF. Other fuel prices were determined based on the correlation between historic price data for the SAF and other fuels, which was carried out in previous work (Tanzil et al., 2021b). The methodology to conduct the TEA has been well documented in two previous studies (Tanzil et al., 2021a; Tanzil et al., 2021b). The set of assumptions for the financial analysis are given in Supplementary Tables S4 and S5. Reference equipment costs were taken from various sources (Davis et al., 2015; Jones et al., 2013; Swanson et al., 2010; Klein-Marcuschamer et al., 2013; Humbird et al., 2011) and were used to calculate fixed capital costs using ratio factors (Peters et al., 2004). The modified cost ratio factors for this work are given in Supplementary Table S3 in the Supplemental Information. A corn stover price of $70/dry metric ton (20% initial moisture) was taken from the literature (Edwards, 2014). Electricity sales price ($0.038/kWh) and purchase price ($0.069/kWh) were taken as 5 year averages (2013–2017) from the Energy Information Administration (EIA) (EIA. US, 2018; US EIA, 2020a). The 5 year average (2013–2017) of natural gas ($4.20/MMBtu) was also taken from the EIA (US EIA, 2018). Other raw material prices are given in the Supplemental Information (Supplementary Table S6). A levelized hydrogen price of $1.77/kg was taken from a United States Department of Energy (DOE) estimation that included capital and operational costs to produce hydrogen (Dillich et al., 2012). The reference salary structure was taken from the literature (Jones et al., 2013) (Supplementary Table S9). The methodology to determine the adjusted salary structure has been outlined in previous work (Tanzil et al., 2021b). All of the analyses were carried out for the cost year of 2017.

GHG emission profiles were developed for the integrated scenarios between the PR and SAF processes following an attributional life cycle assessment (ALCA) approach. A cradle-to-gate system boundary was established as in previous work (Tanzil et al., 2021b). The material and energy flow data are given in Supplementary Tables S7A–S7F. A list of emission factors is also given in the Supplemental Information (Supplementary Table S8). The functional unit selected was 1 MJ of the total fuel product.

For this study, a medium-sized PR with a heavy coking configuration that processes 120,000 BPD was analyzed (Sun et al., 2018). Figure 2 shows a flow diagram of the overall material and energy flow. The hydrogen flow represents a steam methane reforming facility inside the refinery that produces 8.5 MT H₂/hr as required. This configuration processes heavy fuel oil via further hydrotreatment and cracking (FCC) to produce gasoline and diesel.

Fig. shows the overall material and energy flow for the Midwest-based SAF standalone scenarios. The high co-processing capacity of the PR enabled the studied SAF technologies to increase the fuel capacity beyond that in previous work (Tanzil et al., 2021a; Tanzil et al., 2021b). The high H₂ consumption by ICGFT_A was caused by the steam methane reforming (SMR) facility that provided CO₂ for gasification to increase the fuel yield (Tanzil et al., 2021a). The material and energy flows in Figure 2 and Figure 3 were used to determine whether the OSBL-based co-located scenarios matched the 20% cutoff sharing infrastructures.

The large capacity of the PR (Figure 2) allowed the integrated SAF processes to utilize service facilities without surpassing the 20% cutoff requirement for the steam generation, power substation, distribution, and product storage. The water distribution facility was only utilized by two co-located scenarios, namely, FP_B1 and FP_B2, because of the lower water consumption in these two scenarios. Thus, the portion of the ratio factor that covered the service facilities was modified to be in the range of 27–29.5% for all co-located scenarios (Supplementary Table S3). The ratio factor was also reduced for buildings to 29% (Peters et al., 2004). However, for repurposed scenarios, this decreased to 7% for buildings (Peters et al., 2004). In addition, the yard improvement cost was assumed to be zero for all scenarios. More service facilities would be available for a repurposed scenario, and hence, a much lower ratio factor of 8.5% was needed. Details are given in Supplementary Table S3.

Unlike corn ethanol mills (Tanzil et al., 2021b) and sugarcane mill-based integration (Tanzil et al., 2021c), PR-based integration scenarios have higher capital and operational costs as a result of the higher production capacities (Table 2). Figure 4 demonstrates the FCI reduction opportunities for each integrated scenario from their respective standalone scenario. Sharing the cost of the OSBL infrastructure (VB_B1, ATJ_B1, DSHC_B1, FP_B1, GFT_B1, and ICGFT_B) reduced the costs by 6–10%. In addition to OSBL cost sharing, replacement of the power generation module with power purchases (VB_B2, ATJ_B2, DSHC_B2, and GFT_B2) reduced the costs by 26–33% compared with the standalone scenarios. The capital costs were reduced by 28–39% if co-processing was adopted (VB_B3, ATJ_B3, DSHC_B3, FP_B2, and GFT_B3). Repurposing strategies reduced the capital costs by 12–44% in comparison with the standalone scenarios (VB_C, ATJ_C, DSHC_C, FP_C, and GFT_C). In all repurposed scenarios, the cost of the PR (72 MM$) was added as the FCI component.

Although CAPEX reductions were realized, OPEX did not always decrease as shown in Figure 5. Four scenarios, namely, VB_B2, VB_B3, ATJ_B2, and DSHC_B2, had OPEXs that increased by 3–9% because of the purchase of electricity. Six scenarios that only utilized OSBL infrastructures from the PR (VB_B1, ATJ_B1, DSHC_B1, FP_B1, GFT_B1, and ICGFT_B) reduced OPEX slightly by 1–2% from the respective standalone scenarios due to salary reductions of 10%. For these scenarios, the maintenance cost (Supplementary Table S5) did not change from the corresponding standalone scenarios because these integrated scenarios did not have cost reductions from ISBL.

For the other scenarios, fixed OPEX was reduced by 3–41% from the corresponding standalone scenarios because of cost reductions from ISBL. Therefore, three co-processing scenarios (DSHC_B3, FP_B2, and GFT_B3) reduced the total OPEX by 10–13% from the respective standalone scenarios, and repurposed scenarios reduced OPEX by 8–32% from the respective standalone scenarios. It is noteworthy that the other OPEX (OPEX of raw materials and energy) of the two repurposed scenarios (VB_C and FP_C) were reduced by 47% because the levelized cost of H₂ was replaced by using already existing steam methane reforming inside the repurposed PR facility. In such a case, the purchase of natural gas (assuming a stoichiometric SMR reaction) nearly halved the other OPEXs.

The cost profiles of FCI and OPEX were reflected in their respective MFSP estimations (Table 5). Sharing only OSBL infrastructures (VB_B1, ATJ_B1, DSHC_B1, FP_B1, GFT_B1, and ICGFT_B) reduced MFSP by 3–6% in comparison with the respective standalone scenarios, while the non-power generating scenarios (VB_B2, ATJ_B2, DSHC_B2, and GFT_B2) reduced the costs by 2–14%. Co-processing scenarios reduced the MFSP by 10–19% compared with the respective standalone scenarios. Repurposed scenarios reduced the MFSP by 16–34%, following contributions from the reduced OPEX and FCI, as discussed above.

Single point sensitivity analyses of five parameters—co-processing ratio, feedstock cost, real discount rate, FCI, and equity (Davis et al., 2015)—were carried out in this work. The base values of the equity and real discount rate are taken as 30% of FCI and 10%, respectively (Davis et al., 2015). The rest of the base values are given in Sustainable Aviation Fuel Scenario. For favorable and unfavorable values of the equity and discount rate, ±50% of the base value was assigned, as taken from the literature (Humbird et al., 2011; Davis et al., 2015). A favorable value of $60/dry MT and an unfavorable value of $100/dry MT of corn stover was also taken from the literature (Thompson and Tyner, 2014; US Department of Energy, 2011). For the FCI sensitivity calculation, ±30% was used as the percent delivered method considering a ±30% estimation error in estimating the FCI (Peters et al., 2004). Sensitivity analyses for the new concepts ICGFT_B and ICGFT_C were not carried out because their MFSP values were close to conventional jet fuel prices (US EIA, 2020b).

Figure 6 shows the sensitivity plot of five integrated scenarios that utilized only the shared infrastructure from an existing PR. The favorable values of these parameters analyzed individually only reduced the MFSP values by 2–16%. In terms of the sensitivity ranking, the impact of the co-processing ratio was consistently among the two bottom parameters except for FP_B1.

The co-processing ratio can be directly correlated to both the fuel capacity and feedstock capacity. As described in Sensitivity Analyses, the choice of the base value of the co-processing ratio was dictated by the feedstock processing capacity, which was close to 2,000 MTD, except for FP_B1. Therefore, a ±50% variation indicated a feedstock range between 1,000 and 3,000 MTD (approximately). Previous work suggests (Tanzil et al., 2021a) that MFSP values do not change significantly after 1,000 MTD of feedstock capacity, and values tend to flatten after 2,000 MTD. Similar suggestions also have been made in the case of MFSP vs fuel capacity (Tanzil et al., 2021a). However, it also has been suggested that a high feedstock capacity or fuel capacity can increase the MFSP value (Tanzil et al., 2021a). Because in case of a low fuel yield scenario, high feedstock capacity can significantly increase the capital and operational cost. This explanation can be linked to the fact that DSHC_C (Figure 8) had a 30% increase in MFSP for a 50% increase in the co-processing ratio or fuel capacity.

The sensitivity of feedstock cost appears to be among the two top parameters for the majority of the repurposed scenarios (Figure 8).

Table 6 shows the GHG emission profiles of the integrated scenarios, which were categorized into the following three segments: feedstock usage, conversion site, and co-product credits. Greenhouse gas (GHG) estimations showed that 17 of 21 integrated scenarios resulted in GHG savings (7–92%).

Corn stover usage resulted in GHG emissions in the range of 4–15 g CO_2-eq/MJ of total fuel. The emissions from the conversion site were dominated by the energy consumption as well as hydrogen consumption. VB_B2, VB_B3, DSHC_B2, and DSHC_B3 showed higher emissions as a result of the high hydrogen consumption and fossil fuel-based electricity. The repurposed scenarios yielded lower emissions than co-located scenarios because the former took advantage of the onsite SMR plant to produce hydrogen, thus avoiding the high emission factor of purchased hydrogen. The co-product credit includes lignin sales, the electricity credit, and the displaced emission profile by hydrocarbon fuels other than SAF. Dry lignin fuel was assumed to replace the emissions caused by coal.

A displacement factor of 10 kg coal/kg lignin (Pourhashem et al., 2013) was used to calculate the emission credit by lignin fuel sales. Seventeen integrated scenarios resulted in GHG savings (Table 6) compared with the GHG emission value of 87.3 gCO_2-eq/MJ of conventional fossil fuel (GREET, 2018).

The estimated MFSPs (Table 5) and GHG emissions (Table 6) are two performance criteria used to evaluate the integrated scenarios. Economic performance largely relies on the cost structure (lower MFSP is desired), while environmental performance relies on process improvements in terms of less energy consumption and on the method of emission estimation (lower GHG emission is desired). Scores from 0% (highest MFSP/GHG emission) to 100% (lowest MFSP/GFG emission) were assigned to each of the estimated MFSP and GHG emission values.

The detailed methodology for depicting both types of performance (Figure 9) is adopted from elsewhere (Garcia-Nunez et al., 2016; Tanzil et al., 2021b) and outlined in the Supplemental Information (Supplementary Tables S10, S11). The spider plot in Figure 9 shows that the repurposed scenario (ICGFT_C) of the proposed novel technology ICGFT had the best performance in terms of both economic and environmental impacts. Although each of the integrated scenarios reduced the MFSP, only one scenario (ICGFT_C) from the proposed novel concept had an MFSP lower than that of conventional jet fuel ($0.54/L) (US EIA, 2020b).