(348b) An Optimization-Based Framework for the Assessment of Biomass-to-Fuel Conversion Strategies

Advances in biofuels research have resulted in the development of a large number of conversion technologies that can potentially replace conventional petroleum-based production of fuels. However, it is unclear what type of technologies should be combined and what mix of products should be produced to make biofuel production more attractive. To address this type of research questions, we develop an optimization-based framework for the synthesis and evaluation of existing and emerging biofuel production strategies. In particular, we generate a biomass utilization superstructure that consists of 172 major conversion technologies and the corresponding 125 compounds. Based on this superstructure, we develop a system-level framework for the comparison of biofuel strategies, the identification of the major technology gaps and cost drivers, and the assessment of the impact of technology uncertainty.

First, we estimate a set of parameters such as yields of major products, unit production costs, and energy requirements of each technology. Based on the technology superstructure along with the associated parameters, we develop a linear programming (LP) model that consists of (i) material balances for all compounds, (ii) production capacity constraints, (iii) feedstock availability constraints, and (iv) product demand satisfaction constraints. Using this model, various types of questions (e.g., which feedstock/strategy is best to produce a given product, which strategy/product is best to utilize a given feedstock) can be addressed using different types of assessment criteria (e.g., economic, environmental). In addition to the optimal strategy, we develop a mixed-integer programming (MIP) model to identify alternative strategies.

Furthermore, we assess the uncertainty in the projected production cost of the technologies based on their levels of maturity and complexity. Accordingly, each technology is assigned an uncertainty level, which is then used to determine an interval for the total production cost of each strategy. Also, we carry out a sensitivity analysis on the major parameters to understand the effects of their variations on the objective function value using the interpretation of LP duality that establishes dual (shadow) prices. Shadow prices aids to predict how much improvement in the unit production cost, feedstock price or product selling price is required to make a strategy attractive, and anticipate the impact of introduction of a new technology into the superstructure. 

To illustrate the capabilities of the proposed framework, we study the production of ethanol from hardwood as a case study. Ethanol can be produced from hardwood via hydrolysis (dilute acid, ammonia fiber expansion, hot water), direct or indirect gasification, or pyrolysis. We find that the most cost-effective strategy is the production of ethanol from methanol synthesis via indirect gasification followed by acetic acid production and hydrogenation at $3.50 per gallon of ethanol. Although gasification-based strategies have higher capital and operating costs, their unit production costs are lower than fermentation-based strategies mainly due to their high ethanol yields and byproduct (acetic acid) credits. Sensitivity analyses reveal that the economics of the gasification-based strategies can be improved primarily through processing improvements (e.g., cheaper catalyst), while cheaper feedstocks can considerably reduce the total production cost of the fermentation-based strategies. 

Finally, we present a software tool that allows users with no optimization background to use the proposed optimization-based framework. This tool enables users to provide and update the required data as well as extend the superstructure by introducing new technologies. The results of the optimization can be visualized using a network representation and graphics.