Heat Treatment for Bearings: Scheduling Strategies and Distributed Decision Support System 1 B.R.S.M. JOTHI and SACHINB. PATKAR Department of Mathematics, Indian Institute of Technology, Bombay, Mumbai - 400 076, INDIA. Abstract: - Scheduling and Production Planning problems are typically large in scale and fairly complex. A decision support system for such an operations research (OR) problem makes a good case study to be implemented using the web enabled, enterprise scale facilities of Java 2 Enterprise Edition (J2EE) and Simple Object Access Protocol (SOAP). We propose mathematical programming models as well as design and analyze certain specialized algorithms for special versions of these problems. In general, we use a powerful grade Mixed Integer Programming (MIP) solver, which we have web-enabled in our distributed decision support system, to solve the MIP models created by our system. Reusing web infrastructure can drastically lower the cost of setting up these application and allows us to reuse all kinds of tools originally built for the web. Keywords: - Scheduling, Apache SOAP, EJB, Web Services, JMatLink, Matlab and FortMP solver. 1 Introduction Scheduling and Production Planning problems are typically large in scale and fairly complex. Schedul- ing problems, concern the allocation of limited re- sources over time to perform some tasks to satisfy certain conditions. Scheduling problems exist almost everywhere in real-world situations, especially in the manufacturing industries. A decision support system for such a operations research (OR) problem makes a good case study to be implemented using the web en- abled, enterprise scale facilities of Java 2 Enterprise Edition (J2EE) and Simple Object Access Protocol (SOAP) [1, 7]. A bearings manufacturer produce a large number of bearings of different kinds. It operates under many constraints, some of which involves the utilization of resources, productivity, heat treatment for bear- ings, downtime of machines and the target price - the price of the product specified by the customer. After the heat treatment process the bearings must 1 Research supported by grant from project WebOPT (Asia IT & C contract ASI/B7-301/97/0126-73) be assembled. Hence, it requires to find out the op- timum number of different bearing types for the heat treatment schedule. We have modeled two Mixed In- teger Programming formulation’s for the Scheduling in Bearing Heat Treatment Plant. In this paper, we investigate a few variations of scheduling problems arising on the shop floor of a heat treatment plant of a bearings manufacturing industry. These problems arose in our collabora- tive efforts with the industry NRB Bearings Limited, Mumbai. We propose mathematical programming (in particular, mixed integer programming (MIP)) models as well as design and analyze certain special- ized algorithms for special versions of these prob- lems. However, in general, we use the powerful, industry-grade MIP solver such as FortMP[4], which we have web-enabled in our distributed decision sup- port system, as the solver in the back-end to solve the MIP models created by our system. (FortMP is an in- dustrial strength large scale optimization solver sys- tem developed by CARISMA, Brunel University as a research tool which is also used for teaching with many industrial applications [4]). 1