Design and optimization of a bioprocess to simultaneously produce biochemicals and remediate acid mine drainage using indigenous South African grasses

Abstract

Acid mine drainage (AMD) is highly acidic water, with a high dissolved sulfate and metal concentration, that is produced through mining activity. AMD has a negative effect on natural ecosystems into which it flows and negatively impacts national water security. The impacts of AMD are widespread in the Witwatersrand mining area and Mpumalanga coalfields in South Africa. As South Africa is a water-scarce country it is imperative to efficiently remediate AMD and protect natural water resources. One method that has been investigated for AMD remediation is dissimilatory sulfate reduction (DSR), which is catalyzed through sulfate-reducing bacteria (SRB). DSR converts highly acidic sulfuric acid into weaker hydrogen sulfide, which raises the pH and results in the precipitation of metals. SRB require a carbon source that can be supplied in the form of lignocellulosic biomass. The acidity present in the AMD breaks apart the lignocellulosic biomass, releasing xylose into the water, which can then be utilized by the SRB. Lignocellulosic biomass has also been identified as a good source of biomass for the production of biochemicals. One of the main drawbacks of using lignocellulosic biomass as a feedstock is that the presence of the ligno-hemicellulosic matrix prevents cellulase enzymes from accessing the cellulose fibrils, which are broken down to glucose and subsequently fermented to biochemicals. To overcome this the biomass first needs to be pre-treated to break apart the ligno-hemicellulosic matrix. Various pre-treatment methods have been investigated (physical, chemical, physico-chemical, biological) but pre-treatment with sulfuric acid has been identified as the most commercially feasible pre-treatment. As the breakdown of lignocellulose for DSR remediation of AMD is similar to the pre-treatment of lignocellulosic biomass, there is the potential to combine these two processes. Biomass can undergo pre-treatment with AMD, releasing xylose into the AMD which can be utilized by SRB for DSR. The biomass can then undergo enzymatic hydrolysis to release glucose that can be fermented to produce various biochemicals. This study investigated the design, optimization, and feasibility of this simultaneous process. The investigation started with the evaluation of different biorefinery options based on the biorefinery complexity profile (BCP). Indigenous South African grass was found to be the most suitable feedstock due to its abundance in the regions where AMD is generated. The use of xylose as a substrate for SRB, and the production of bioethanol through glucose fermentation was found to be the most feasible biorefinery option. The production of additional high-value chemicals from glucose, and processing of lignin and distillery silage into valuable products were also identified as options to increase the economic feasibility of the biorefinery. Next, the optimal configuration of reactors to pre-treat biomass and perform DSR was investigated, through developing flowsheets for three different reactor configurations. It was found that the rate of hydrolysis of biomass is severely rate-limiting, and the only process which shows any promise for commercial implementation is one in which hydrolysis of biomass is operated at elevated temperatures (100°C), in a stand-alone hydrolysis reactor separate from the DSR reactor. Sensitivity analyses revealed that the design was highly sensitive to the kinetic data used, and the temperature at which the pre-treatment reactor operates. As the design was found to be highly sensitive to the kinetic data used, the rate of dilute sulfuric acid hydrolysis of indigenous South African grass, at low temperatures (<100°C), was investigated. The iii xylan (hemicellulose) was found to have two distinct fractions, which react at significantly different rates. The rate of reaction of one fraction was found to display Arrhenius type temperature dependence (Ea = 155.06 kJ/mol, A0 = 1.65×1019/min), whereas the rate of the reaction of the other fraction was so slow it could be considered negligible. The portion of the fraction with a slower rate of reaction was found to be 50%, which is lower than previously determined (55 – 100%). Following this, the optimal conditions for the production of bioethanol from biomass were investigated. This included investigating the optimal pre-treatment time, the optimal pre-treatment solid loading, and the glucose and ethanol yield in both separate hydrolysis and fermentation (SHF) and simultaneous saccharification and fermentation (SSF), for biomass pre-treated with AMD or H2SO4. Empirical models were also developed to predict the glucose and ethanol concentration over time. In both SHF and SSF pre-treatment using AMD was found to achieve a glucose/ ethanol concentration that was 70 – 80% of that achieved from pre-treatment with H2SO4. The empirical models had a high correlation (r2 = 0.87 – 0.99) to experimental data. An Aspen Plus™ flow sheet was then developed and used to perform a techno-economic evaluation to determine the economic feasibility of a lignocellulosic bioethanol facility that uses AMD pre-treatment of biomass. Both SHF and SSF reactor configurations were considered, and evaluations with H2SO4 pre-treatment were also performed for comparison. Simple estimations for capital and operating costs were used to estimate the economic feasibility of all scenarios. Only the Scenario with H2SO4 pre-treatment and SHF was found to make a profit, although the profit was so small the payback period would be 80.7 years, making the process infeasible. SHF was found to be better suited to the process than SSF for both AMD and H2SO4 pre-treatment. Finally, a techno-economic evaluation was performed on a simultaneous AMD remediation and lignocellulosic bioethanol production facility. As previous studies showed that having separate pre-treatment and DSR, as well as SHF, is more feasible the techno-economic evaluation was based on this. The minimum ethanol selling price (MESP) was calculated using discounted cash flow methods and found to be 3 114 USD/ton ethanol. Although this is higher than recent historical prices sensitivity analyses revealed that there are various opportunities to reduce the MESP substantially. The MESP was especially sensitive to the ethanol yield and the operating temperature of the pre-treatment reactor. An evaluation at ethanol yields that are considered to be achievable and slightly increased pre-treatment temperatures (+20°C) resulted in an MESP of 741 USD/ton. This is within the recent historical prices, and indicates the process could be feasible at these conditions and favorable ethanol prices.