May 23, 2014, Broida 1640
Christos T. Maravelias
University of Wisconsin - Madison, Biological and Chemical Engineering
We present an overview of recent advances in chemical production scheduling, including a general framework, new modeling and solution methods, and some concepts for the online solution of scheduling problems. First, we develop a framework for the description of scheduling problems in the chemical industries. While building upon ideas used in discrete manufacturing, the proposed framework accounts for features such as material handling restrictions which make chemical production environments different. Second, we present a classification of the various modeling approaches that have been presented in the process systems engineering (PSE) literature and offer some critical insights. Third, we briefly review a series of mixed-integer programming (MIP) models we have developed to address limitations of existing approaches. Fourth, we present an overview of two solution methods that lead to order-of-magnitude speedups, namely, a constraint propagation algorithm for the calculation of parameters used to generate tightening constraints, and a series of reformulations. Fifth, we present a state-space formulation for general scheduling problems. We discuss the modeling of “scheduling disturbances”, as well as stability and recursive feasibility in the context of scheduling. We close with some thoughts on the treatment of scheduling as an online problem with desired closed-loop properties.