GOALI: Infrastructure Investment and Operation Decisions for Biobased Production Networks

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

0933392 Lee The objective of this GOALI project is to develop an optimization model that supports the multi-stage decision-making of infrastructure investment and operation of bio-refineries. These decisions include the selection of fuel conversion technologies, design and expansion of processing network, and the logistics of transportation. In collaboration with Weyerhaeuser NR, the PIs recently developed a preliminary version of such an optimization model. Testing of the model with preliminary data provided by the company yielded some useful insights into the relationship between design structure and various cost elements. However, the current model is able to use only very basic information (e.g., transportation cost, capital cost, operating cost, etc.), is limited in the scope of technologies it considers, and does not consider synergies with existing wood processing infrastructure. In addition, it does not account for uncertainties in performance and costs of major system components and does not support dynamic decision-making needed to address various temporal aspects of the problem. All of these are inherent features of the problem. In this project, the PIs will develop a richer, broader model that considers opportunities for integration with existing forest product processing facilities, uncertainties associated with various silvicultural, technical and economic parameters of the problem, and temporal aspects such as active planning of biomass supplies through forestry management and multi-period expansion of the processing capacities. They will test the models on company-representative data and analyze the results in terms of profits and other measures of system performance, such as the percentage carbon in the forestry resource that is captured in the product, the ton miles travelled, and the percentage utilization of capacity for different market shares of new chemicals and fuels. This information will be integrated into a life cycle inventory of the system to support life cycle assessments for renewable fuel standards. Intellectual Merit: Extension of the current model to remove these limitations requires meeting several intellectual challenges: (1) Synergistic integration with existing wood processing infrastructure requires decisions regarding mass and heat integration schemes with the existing operations to be combined with process and location choices. The industrial collaborator can provide specific facility data for this purpose. (2) To address uncertainties, such as those in the biomass availability, process performance, and final product demand / price, multi-stage stochastic decision problems will be formulated and solved. The industrial collaborator is in a position to provide reasonable structures and estimates of these uncertainties. The multi-stage formulation is also needed to support dynamic decision making. The solution of multi-stage stochastic decision problems can be computationally challenging, and will require development of a tailored solution algorithm. Broader Impact: A systems model of bio-refinery investment optimization can help maximize return on investments, existing and new, made to convert forestry resources into a broader product portfolio. With one of the leading forestry product companies participating in the project, the research outcome is expected to make an immediate impact in the forest product industries. The model can serve as a tool to inform policy makers and other stakeholders what types of policies incentives for investment are needed to spur the economic development based on fuels and chemicals from forestry resources. It will promote a more quantitative, engineering-based, approach to evaluating investments in bio-based fuels and chemicals. The PIs also expect to advance the state of the art of ADP by applying it to an important problem of a scale never tried before. The success of ADP on a problem of major national interest could spur greater interest in this problem solving technique within the process systems community. On the education side, there is a paper science and technology option that could absorb many of the results on the detailed process simulation and integration and a graduate data driven modeling course, which uses a case study approach. In addition, the company will educate its executives of the need for systematic evaluation of bio-refinery options and general capabilities of advanced computational systems.