A robust optimization generic model for forest biorefineries design considering uncertainties on biomass growth and product selling prices.

• Ano de defesa: 2022
• Autor(a) principal: Theozzo, Bruno
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Biorrefinaria Florestal; Celulose; Forest biorefinery; MILP; Optimization under uncertainty; Otimização robusta; Otimização sob incerteza; Papel; Pulp and paper; Robust optimization
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/3/3137/tde-20122022-085410/

Biorefining emerges as a potential alternative for fossil-based industries by proposing the use of renewable biomass to produce chemicals, fuels, and energy. Several opportunities arise from a large set of available biorefining products, conversion technologies, sources of biomass, and integration points with other industries. The design of biorefineries that optimally explore this vast opportunity space is not trivial. Several tools have been proposed for this task. Among these, mathematical programming is highlighted as one of the most promising strategies due to its systematic evaluation of a large space of structural alternatives. The state-of-the-art mathematical programming-based frameworks deal with process synthesis challenges along with supply chain challenges under explicit spatial and temporal consideration. However, these frameworks cannot capture some particularities of forest systems, one of the most promising platforms for integrating biorefining operations. Thus, this thesis proposes an optimization generic model for biorefineries design able to account for the specificities of forest systems, under an MILP (Mixed-Integer Linear Problem) formulation for maximizing the operating net present value (NPV) over four interconnected layers of decision: (I) Forest dynamics, (II) Conversion systems, (III) Supply Chain, and (IV) Markets. The model also incorporates uncertainties in biomass productivity and product selling prices under a robust optimization formulation to ensure the operations profitability and feasibility even on the materialization of the worst-case values for the uncertain parameters. The degree of conservatism of the solution is controlled through the description of the uncertainties within the intersection of a box and polyhedral sets. The model is applied to a case study on the design of a eucalyptus biorefinery in Brazil to produce bleached pulp (used for papermaking), lignin (used as a cement additive), and electricity. The case study showed that biomass dynamics played a vital role in the core strategic decisions, with biomass availability and forest distances driving decisions on all other layers. The adequate consideration of the supply networks was also relevant, as some highly productive forest lands become financially attractive in scenarios with reduced logistic costs. The model was demonstrated useful for minimum selling price estimation for products that might not be considered financially attractive under the nominal values for the input parameters. The choice of conservatism degree was demonstrated as an important feature of the model. Under a full conservative approach, the entire biorefinery operation was considered financially unattractive, contrasting with the nominal case scenario that indicates an opportunity of over 136 billion BRL (around 27 billion USD) in net present value. The over-conservatism on forest uncertainties is especially harmful as it proposes a 70% excessive usage of land to ensure a stable wood supply. This excessive land usage might compete with land for food crops. In between, some still conservative designs were proposed that could provide a robustness level consistent with the nature of the uncertain parameters and still benefit from the biorefining opportunities.