Incorporating agricultural waste-to-energy pathways into biomass product and process network through data-driven nonlinear adaptive robust optimization. (1st August 2019)

• Record Type: Journal Article
• Authors: Nicoletti, Jack; Ning, Chao; You, Fengqi
• Is Part Of: Energy. Volume 180(2019)
• Page Start: 556
• Page End: 571
• Publication Date: 2019-08-01
• Subjects: Adaptive robust optimization -- Return on investment -- Biomass conversion -- Machine learning; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2019.05.096
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: A biomass product and process network that displays how organic waste and other non-traditional biomass feedstocks may be converted into useful bioproducts and biofuels is a necessary addition to the field of biomass conversion and utilization. We develop a processing network of 216 technologies and 172 materials/compounds that contains conversion pathways of agricultural and organic waste biomass sources, such as food peels, animal manure, and grease. To examine the effectiveness and economic feasibility of these conversion pathways, the biomass product and process network is optimized for return on investment. The resulting problem is a data-driven two-stage adaptive robust mixed-integer nonlinear fractional program, which was effectively solved via a tailored optimization algorithm. The proposed approach is applied to two case studies in which traditional agricultural feedstocks are used alongside biological and agricultural waste feedstocks. The selected feedstocks were used to satisfy and, in some cases, even exceed demand for selected products. The optimal pathways have returns on investment of 26.1% and 6.2%, with utilized conversion technologies ranging from hydrocracking to microwave hydrodiffusion. In both cases, we find that profitable processing pathways are utilized at maximum capacities to increase return on investment. Specifically, in the case study where orange peel wastes are used to produce pectin, we find that this pathway is highly profitable at the given market price. The two cases that are run using the proposed model are then compared to additional cases to display differences that arise when uncertainty is not considered and the objective function of the model is changed. Highlights: The expanded biomass product and process network is described. Two-stage data-driven ARO is applied to a biomass product and process network. Case studies, computational results, and future applications are discussed. A novel application of the proposed computational algorithm is presented.