This is an online supplement for "Efficient Control for a Multi-Product Quasi-Batch Process via Stochastic Dynamic Programming" where we present complete list of numerical results discussed in Section 6 of the paper. Tables 1 through 4 provides the results for N = 3 case. The first column stands for the number assigned to the data set, which is used as a label for that data set later. The next five columns list the parameter values specific to the data set. The cost values are given in the

more »

... of optimal, traditional and alternative together with the percentage cost and energy/material savings obtained via the optimal policy against the traditional and alternative policies. Note that "energy" labels on the tables stand for "energy/material", which is abbreviated due to space constraints. Tables 5 through 8 present the results for heuristics and optimal solution for the case of N = 5. The numbers in the first column correspond to those in the tables for N = 3 and represent the data set used in the experiment. The CPU time used (seconds) to generate the proposed policies by the exact value iteration algorithm and the heuristics are displayed in addition to the cost values. Computational times for Heuristic 2, Heuristic 3 are excluded due to space constraints as their results are quite similar to those of Heuristic 1. Decomposition of total cost into different components is provided as well as the total system-wide cost values. Note that the penalty cost is not provided for the optimal policy and the Decomposition Heuristic as it can be calculated using the given cost values. Similarly, the switching cost for the naive Heuristics can be calculated using the results for total cost and energy/material cost. Note that for those heuristics, there is no penalty cost listed, as under-processing is not allowed and that type of cost is not incurred. Finally, Tables 9 to 12 list the results for the cases of N = 15 and N = 100, which are discussed in Section 6.2.1 of the paper.