A process selection optimization model and systems planning methodology for environmentally conscious manufacturing.

摘要

Recent concepts such as Industrial Ecology and Environmentally Conscious Manufacturing take a holistic view of manufacturing systems. This means that all material flows through a facility are important, not just those of the primary products, but also secondary material flows such as process waste streams, catalysts, and worn tools. In this dissertation, we address a particular optimization model for process selection and product routing, while accounting for equipment capacity, process waste streams, process energy usage, and raw cycle time. The decisions in this model are: to select manufacturing process and waste mitigation equipment, to route the flow of products through the resultant set of machines in the facility, and to determine recommended operating parameters, such as temperatures or speeds. This model is unique in that it considers equipment selection, product flows, and process settings simultaneously. Our model is one of the first to consider manufacturing system waste flows and their costs explicitly in a planning model.; We formulate the multi-product process selection and routing decisions as a mixed-integer optimization problem, having the form of a capacitated, multicommodity network design and flow problem with additional side constraints. With this network flow approach, we can include many process alternatives, and can model a variety of manufacturing system configurations that appear in different industrial sectors, including jobshops and re-entrant flows. The structure of the model allows the inclusion of unit process models that quantify waste streams and energy usage as functions of the settings of process and operating parameters. These unit process models introduce nonlinearities into the formulation. The objective function consists of amortized equipment costs, waste treatment costs and penalties, and manufacturing costs. Constraints in the problem include conservation of network flow, multi-commodity capacity limitations, limits on waste mass flows and total process energy usage, and bounds on raw cycle time for particular products. We develop a novel multi-layer decomposition scheme based on Benders' decomposition to solve the problem. We utilize a column generation algorithm to solve the multicommodity network flow sub-problem that results from the decomposition procedure. We also develop heuristics for use in solving the other subproblems.; The central contributions of this dissertation are (i) formulation of the new process selection and product routing optimization model, (ii) development of the multi-layer decomposition solution approach to solve this problem, and (iii) creation of a methodology to analyze and improve manufacturing system environmental impacts, which contributes to Environmentally Conscious Manufacturing.