In this thesis study, a new method for batch process planning was developed. It is based on a simulated annealing technique and can solve a variety of batch process planning problems having quite general objectives and various constraints such as intermediate products and compatibility between products and processing units. Especially, this new approach is designed to deal with discrete demands and due dates. Hence, it helps us to find a feasible production plan which can meet all the demands and keep the due dates given by customers. In another point of view, using the algorithm, we can check if a batch process is overutilized under given demands and due dates, which may be an important means for capacity planning of a batch process.
A software was implemented to demonstrate the versatility of this approach. Using the software tool, results for a series of case studies are presented to show the characteristic feature of the planning such as a trade-off between the inventory cost and setup cost and/or the revenue loss during the setup time.
Modern environmental regulations and disposal costs as well as shipping regulations can dramatically increase the incentive for recovery of solvents. In this situation, the same tool can be used to minimize wastes from the solvent recovery system. Several alternatives to reduce the solvent wastes were investigated. When the batch process utilization is low, the waste amount can be reduced to some extent with waste minimization function of the software.
In the next phase of this study, we describe a general strategy to treat uncertainties in batch process scheduling and planning. The strategy is based on a probabilistic approach using simulated annealing and Monte Carlo simulation techniques. First, to deal with uncertainties in product demands, due dates and other parameters which play important roles in the scheduling of batch processes, two algorithms are developed: flexible scheduling and reactive schedule adaptation.
Through the flexible scheduling algorithm we obtain a flexible schedule. To adjust this flexible schedule in response to new information during a production campaign, we use the reactive adaptation algorithm. This algorithm finds a new good solution under the new conditions by applying a combination of different local search methods to the original schedule.
To complete the general strategy, we extended the same approach to flexible planning which incorporates flexibility into the production plan. Using the flexible planning, scheduling and reactive schedule adaptation algorithms, we can deal with a variety of uncertainties involved in batch processing such as the product demands, the due dates, the processing times, etc.
