Towards implementation of sustainable bio-succinic acid production

Succinic acid (SA) is a small organic molecule that can be used to produce more than 30 commercially valuable products. Historically a petroleum refinery product, SA has recently been produced on a large-scale through biomass fermentation, covering about half of the market demand. Currently, due to the higher production costs, bio-succinic acid (bio-SA) is only a niche product and the majority of SA on the market is petroleum-based. Novel hybrid techniques and production process optimization have been demonstrated to be potential ways to increase economic competitiveness and environmental sustainability of bio-SA production.

The objective of this thesis was first to study and identify the strengths and weaknesses of existing processes and technologies for bio-SA production and then to generate novel cost-effective and environmentally sustainable processes. Novel potential production processes were generated using two different approaches: first, by combining established technologies with promising laboratory scale technologies; second, through deterministic and stochastic process optimization based on established technologies (technology readiness level - TRL 7-9). The simulation studies highlighted the potential of nanofiltration (NF) membranes in the downstream of bio-SA production. However, a lack of understanding of the separation mechanisms in NF membranes emerged from reviewing the literature.

A literature review was conducted first to identify the following: (1) the most relevant feed-stocks in terms of availability and potential bio-SA yield; (2) pre-treatments techniques to maximize conversion to fermentable substrates and to reduce the quantity of inhibitory compoundsentering the fermenter; (3) the most suitable fermentation technologies and microbial hosts; (4) promising downstream technologies. Next, a comparative process techno-economic analysis of three relevant patented processes and of a novel hybrid and conceptual process were performed. Two of the processes were based on the patents from companies that produce, or have been producing, bio-SA at a commercial scale, Reverdia (Process I) and BioAmber (Process II), while a third process was based on a Patent from Michigan State University (Process III) and includes highly integrated stream recycling and chemicals regeneration. The conceptual process (Process IV) includes a continuous fermentation with immobilized cells combined with in situ bio-SA extraction using an anion exchange membrane (AEM) electrolytic cell. The conceptual process required 50% or lower capital investment and had ~40 to 55% lower manufacturing costs than the other processes. The minimum selling price was estimated to be 1.4 USD per kg of bio-SA, which would make this process potentially competitive with petroleum-based SA, estimated to have a minimum selling price of ~2.0 USD kg^-1. The Reverdia-based process could also be competitive, while the other two processes were not profitable. However, the environmental sustainability of Process IV was only marginally better than petroleum-based SA and, as a conceptual process, suffers from high technical and economic uncertainties.

Through process optimization using established technologies (TRL 7-9), a potentially robust, cost-effective and environmentally sustainable bio-SA process was generated from glycerol. The process is based on neutral pH “aerobic fermentation” hosting Escherichia coli, while the downstream of bio-SA includes membrane separation, crystallization, and especially ion-exchange columns as key technologies.
However, the deterministic and stochastic optimization showed high economic and optimal topology uncertainties. The final estimated bio-SA minimum-selling price was 1.6-1.9 USD kg^-1, which could make the process competitive with the petroleum-based SA process in the short term.

Overall, both bio-SA process simulation studies revealed that low-pH fermentation is economically and environmentally more sustainable than neutral pH fermentation. This is because less buffering chemical needs to be added in the fermenter and therefore less needs to be removed and disposed later in the downstream. However, this should not come at the cost of low bio-SA fermentation yield as seen for some low-pH tolerant yeasts. If yield is low, the economic advantage obtained from reducing the load of chemical reagents could be negated by the higher biomass load that must be processed and the greater plant size needed to handle it. Glycerol and corn stover, which were identified through the critical review, were confirmed to be among the most important feed-stocks for bio-SA production. With regard to the fermentation stage, two important outcomes emerged: i) neutral pH fermentation can still be part of cost-efficient bio-SA production processes, even though it requires high buffering reagent loads; ii) the natural microbial host Actinobacillus. succinogenes has strong potentials to be used for large-scale bio-SA production.

The literature revision, process techno-economic analysis and process optimization studies led to the finding that NF can be used in key bio-SA separation steps like bio-SA concentration and bio-SA separation from other organic acid by-products of fermentation. Chapter 5 examines the results of several laboratory experiments that were conducted with commercial polymeric NF membranes to separate bio-SA from other organic acids in synthetic solutions and fermentation broth. The initial screening of five membranes showed very low (<20%) to no bio-SA rejection for all five tested membranes (NF270, NTR7450, TS40, DL, DK). However, the permeate fluxes decreased up to 25 times with increasing the pH from 2.2 to 7.0. Tests with xylose solutions showed rejection <40% and no rejection differences for solutions with pH values 2.2, 5.0 and 7.0. Thus the performance of NF membranes with bio-SA solutions was considered the result of size exclusion and electromigration-related phenomena. Tests with synthetic solution mixtures (bio-SA concentrations between 3 to 10 g L-1) and A. succinogenes fermentation broth returned bio-SA rejection values up to 94.1 ± 1.5 % and 95.5 ± 3.5 % and purity below 70% and 60% for synthetic solution mixtures and fermentation broth, respectively. The model evaluation using the Donnan-steric pore model with dielectric exclusion (DSPM-DE) showed that 99% of the succinate rejection occurs at the membrane interface and that the most important separation mechanism is dielectric exclusion at the membrane interfaces. The sensitivity analysis showed that, out of 13 parameters affecting mass transport across the membrane, the most important are charge density (𝑋𝑑), membrane pore dielectric constant (^𝜀𝑝𝑜𝑟𝑒) and pore radius (𝑟𝑝). After parameter estimation and model calibration using synthetic solution mixtures of organic acids, successful prediction of organic acids rejection from fermentation broth was achieved with NF270 and DK (12% and 3% prediction error for succinate rejection by NF270 and DK, respectively). Furthermore, the potential contribution of different transport mechanisms was quantified for the first time.

To conclude, this work intends to give new impetus to the production of SA from biomass fermentation while potentially also contributing to advancements in other biorefinery production. The conceptual Process IV can potentially change the market for bio-SA, and laboratory experiments are being conducted to define the more realistic potential of this process. The process resulting from deterministic and stochastic optimization using established technology could also be competitive with petroleum-based SA production. However, the resulting bio-SA minimum selling price is close to that of SA from a petroleum refinery. Advancements in a key technology like nanofiltration membrane could therefore help to separate successful commercialization of bio-SA from petroleum price fluctuations.