Techno-economic analysis and life cycle assessment of a biorefinery utilizing reductive catalytic fractionation

Abstract

Reductive catalytic fractionation (RCF) is a promising approach to fractionate lignocellulose and convert lignin to a narrow product slate. To guide research towards commercialization, cost and sustainability must be considered. Here we report a techno-economic analysis (TEA), life cycle assessment (LCA), and air emission analysis of the RCF process, wherein biomass carbohydrates are converted to ethanol and the RCF oil is the lignin-derived product. The base-case process, using a feedstock supply of 2000 dry metric tons per day, methanol as a solvent, and H₂ gas as a hydrogen source, predicts a minimum selling price (MSP) of crude RCF oil of $1.13 per kg when ethanol is sold at $2.50 per gallon of gasoline-equivalent ($0.66 per liter of gasoline-equivalent). We estimate that the RCF process accounts for 57% of biorefinery installed capital costs, 77% of positive life cycle global warming potential (GWP) (excluding carbon uptake), and 43% of positive cumulative energy demand (CED). Of $563.7 MM total installed capital costs, the RCF area accounts for $323.5 MM, driven by high-pressure reactors. Solvent recycle and water removal via distillation incur a process heat demand equivalent to 73% of the biomass energy content, and accounts for 35% of total operating costs. In contrast, H₂ cost and catalyst recycle are relatively minor contributors to operating costs and environmental impacts. In the carbohydrate-rich pulps, polysaccharide retention is predicted not to substantially affect the RCF oil MSP. Analysis of cases using different solvents and hemicellulose as an in situ hydrogen donor reveals that reducing reactor pressure and the use of low vapor pressure solvents could reduce both capital costs and environmental impacts. Processes that reduce the energy demand for solvent separation also improve GWP, CED, and air emissions. Additionally, despite requiring natural gas imports, converting lignin as a biorefinery co-product could significantly reduce non-greenhouse gas air emissions compared to burning lignin. Overall, this study suggests that research should prioritize ways to lower RCF operating pressure to reduce capital expenses associated with high-pressure reactors, minimize solvent loading to reduce reactor size and energy required for solvent recovery, implement condensed-phase separations for solvent recovery, and utilize the entirety of RCF oil to maximize value-added product revenues.

Fig. 1. RCF biorefinery design. (A) Block flow diagram of an RCF-based cellulosic ethanol biorefinery. This process follows a configuration similar to that described by Humbird et al., with the dilute-acid pretreatment area replaced with RCF. The RCF area produces a lignin-rich RCF oil as a co-product and a carbohydrate-rich pulp. Carbohydrate pulp, the residual biomass solids after the lignin is removed during RCF, is isolated and saccharified to C₅ and C₆ sugars via enzymatic hydrolysis, fermented to ethanol, and recovered to produce fuel-grade ethanol. WWT is the wastewater treatment area. CHP is the combined heat and power area where a combination of natural gas, sludge from WWT, and residual solids from ethanol production are burned to generate steam for process heat and electricity. Excess electricity is sold to the grid as a co-product. (B) The process diagram highlighting major unit operations in the RCF area of the biorefinery for the methanol case. The solvent pump-around RCF reactor design is shown in Fig. S2 (ESI†). The gaseous products are sent to a PSA unit to recover H₂ to be recycled to the reactor, and the liquid and solid products are separated via centrifugation to isolate dilute RCF oil, which is concentrated in a crude distillation column, and water is removed from methanol in the water removal distillation column. After washing with methanol to recover additional lignin oil, the carbohydrate-rich pulp is isolated via a second centrifuge and the methanol rinse is recycled back to the RCF reactor. Major heating and cooling duties were heat integrated to reduce overall energy demand. A process flow diagram with labeled streams and a corresponding table of stream compositions is included in Fig. S1 and Table S2 (ESI†).

Fig. 2. Methanol case economics, GWP, and CED. (a) Contributions to the MSP of the lignin portion of RCF oil (MSP-lignin fraction). Fixed costs include labor and overhead. Grid electricity is the excess generated electricity sold to the grid, process electricity is the electricity requirement for that process area, and total plant electricity is the electricity generated to offset process electricity. Raw materials include biomass feedstock and process inputs such as makeup solvent, hydrogen, glucose for enzyme production, and natural gas. Capital recovery charge accounts for capital depreciation (capital costs), annual income tax, and return on investment. The cost categories on the y-axis are organized by process area, corresponding to the block flow diagram in Fig. 1, along with input and output costs such as feedstock, enzyme production, utilities, and ethanol sales that are not encompassed in a specific area. The sum of all individual contributions is the MSP-lignin fraction, or MSP of the lignin portion of the RCF oil product, at $1.74 per kg. *Cost of feedstock includes all upstream feedstock logistics, handling, and pre-processing steps up to the reactor. (b) Breakdown of contributions to GWP-lignin fraction. The results shown are normalized to the production of 1 kg lignin fraction and allocated according to mass. Economic allocation results are given in Table S5 (ESI†). Delivered feedstock includes CO₂ uptake by growing poplar. The boiler/turbogenerator (CHP) area is not shown in (b) or (c) because impacts for that area have been divided between process areas with steam demand according to the fraction of total steam used by each area: RCF area (89.6%), hydrolysis & fermentation areas (10.2%), and on-site enzyme production (0.2%). Electricity offset is the emissions credit assigned to the biorefinery from displacement of grid electricity. The total GWP, including the electricity offset and CO₂ uptake by growing poplar, is 0.079 kg CO₂-eq per kg under mass allocation and 0.131 kg CO₂-eq per kg under economic allocation. (c) Breakdown of contributions to CED-lignin fraction. Results shown are normalized to 1 kg lignin fraction and allocated according to mass. Economic allocation results are given in Table S6 (ESI†). As was done for GWP, CED for the CHP area was divided between steam-using process areas, and the CHP area is not shown separately in the figure. Electricity offset is the energy credit from displacement of grid electricity. The total CED is 74.03 MJ per kg under mass allocation and 122.04 MJ per kg under economic allocation. All data shown in Fig. 2 are included in numerical form in Table S6 (ESI†) for Fig. 2a and Table S7 (ESI†) for Fig. 2b and c.

Fig. 3. Sensitivity analysis of the methanol case. Results of single point sensitivity analyses for change to the (a) MSP-lignin fraction, or MSP of the lignin constituents in RCF oil (base normalized price = $1.74 per kg), (b) GWP-lignin fraction under mass allocation, and (c) CED-lignin fraction under mass allocation. For each sensitivity case, the key variable was modified to either its minima or maxima while holding all other variables constant to the methanol case. The minima and maxima values used in this analysis are shown in the vertical axis labels as low cost (light red): base: high cost (dark red). Low cost and high cost indicate variable modifications which led to respective net reductions or increases in the MSP-lignin oil. Reasonable minima and maxima were chosen to understand quantitative impacts based on expected uncertainty, prospective process modifications, or potential limits of each variable, and were then used to evaluate the change in MSP-lignin fraction, GWP-lignin fraction, and CED-lignin fraction. Full rationale for the selection of minima and maxima is given in Table S8 (ESI†). All data shown in Fig. 3 is included in numerical form in Table S9 (ESI†). Sensitivity results for change to MSP-crude RCF oil are given in Table S10 (ESI†). Sensitivity results for GWP and CED under economic allocation are given in Table S11 (ESI†).

Fig. 4. Sensitivity analysis on methanol solvent loading. Effects of solvent loading on (a) MSP-crude RCF oil and MSP-lignin fraction, (b) GWP-lignin fraction under mass allocation, and (c) CED-lignin fraction under mass allocation. Solvent loading was varied from 4–10 L per dry kg biomass holding all other process assumptions constant. The methanol-case solvent loading is 9 L per dry kg biomass. Negative GWP indicates the potential for greater GHG uptake than emissions over the lignin fraction life cycle. All data shown in Fig. 4 are included in numerical form in Table S12 (ESI†). The sensitivity analysis results for GWP and CED under economic allocation are given in Table S13 (ESI†).

Fig. 5. Economic and environmental comparison of RCF process configurations. Methanol utilizes methanol solvent and hydrogen gas, ethanol uses solvent generated in the biorefinery, hydrogen-free generates hydrogen in situ from hemicellulose, and ethylene glycol uses hydrogen gas and ethylene glycol solvent. (a) A summary of RCF oil composition, yield, and productivity. The composition was based on literature values for similar processes at the bench-scale. Yield and productivity values were outputs of the process model at a 2000 dry metric tons feedstock/day throughput. (b) Capital cost breakdown by area for each process configuration, with the total installed capital cost (TIC) included above the pie charts. (c) Yearly operating costs for each process configuration. “Other” costs are associated with carbohydrate conversion to ethanol downstream of the RCF area. (d) GWP-lignin fraction for each configuration shows categories that contribute to carbon emission and carbon uptake, with net GWP-lignin fraction indicated with a circle. (e) CED-lignin fraction for each configuration shows categories that contribute to energy consumption and energy generation (in the form of excess electricity sold to the grid), with net CED-lignin fraction indicated with a circle. See Fig. 1 and Fig. S3–S5 (ESI†) for diagrams of each configuration, Table 2 for a complete list of assumptions for each case, and Tables S17–S21 (ESI†) for full tabulated data shown in this figure. GWP-lignin fraction and CED-lignin fraction under economic allocation for each configuration are given in Table S22 (ESI†).

Fig. 6. Membrane cost allowance. Sensitivity scan of maximum membrane capital cost allowance per throughput volume in order to achieve a given RCF oil MSP at the fuel selling price target. See Table S23 (ESI†) for tabulated data shown in this figure.

Fig. 7. Non-GHG air pollutant emissions. Overview of (a) CO, (b) NO_X, (c) PM, and (d) VOCs for each of the RCF process design cases and the hypothetical membrane separation case, broken down by the source of the emission. The assumptions and methods used for this analysis are summarized in Materials and methods. The raw tabulated data is included in Tables S24–S28 (ESI†). Bar charts for other air pollutants analyzed in this study (PM10, PM2.5, SO₂, and hazardous air pollutants) are included in Fig. S7 (ESI†).

Fig. 8. Minimum RCF oil selling prices with the U.S. price and global consumption of various chemicals for reference. The U.S. Price and Global Consumption data were sourced from a 2016 report by Biddy et al. and are 3–5 year averages. Tabulated data are summarized in Table S34 (ESI†).

Fig. 9. LCA system boundary diagram, with connections to the background LCI database indicated with italics.