G. Mohan Kumar

A simulation based heuristic discrete particle swarm algorithm for generating integrated production-distribution plan

Deciding the strategy for production and distribution in a stochastic demand scenario is important for the manufacturing industries. An integrated production-distribution plan considering regular, overtime and outsourced production costs along with inventory holding, backorder, hiring/laying-off and trip-wise distribution costs is developed for a renowned bearing manufacturing industry producing three types of products at three locations. Demand is assumed to vary uniformly and a novel simulation based heuristic discrete particle swarm optimization (DPSO) algorithm is used for obtaining the best production-distribution plan that serves as a trade-off between holding inventory and backordering products. The algorithm also uses an innovative regeneration type constraint handling method which does not require a penalty operator. In addition to the bearing manufacturing industry data set, two other test data sets are also solved. The simulation based optimization approach gives good approximate solutions for the stochastic demand problems.

A genetic algorithm approach to generate an integrated multiplant aggregate production-distribution plan

Industries adopting multisite manufacturing allocate their forecasted demand from various customers/demand centres to the respective plants based on gross production, inventory holding and distribution costs. But demand allocation (by considering the gross production cost) will not be appropriate, as production in each plant can be carried out through regular, overtime and outsourcing means. Also, this will lead to the allocation of the lions share of demand to a plant whose regular production cost is cheaper, but its overtime/outsourced production costs may be costlier than the regular production cost of other plants. In this paper, an aggregate production-distribution plan considering all the above-mentioned costs is developed for a renowned bearing manufacturing industry in India. The proposed Integer Nonlinear Programming (INLP) model is solved using a genetic algorithm and the results are compared with LINGO 8.0, a popular operations research software. The performance of the genetic algorithm is found to be superior to that of the LINGO 8.0 results.

A discrete PSO approach for generating an integrated multi-plant aggregate production-distribution plan

An aggregate production-distribution model, which minimizes the regular, overtime and outsourced production costs along with inventory holding, backorder, hiring/laying-off and trip-wise distribution costs, is developed for a renowned bearing manufacturing industry in India. The proposed integer non-linear programming INLP problem is NP-hard and hence a discrete particle swarm optimization DPSO algorithm is applied to solve the multi-site, multi-period production-distribution problem. The results obtained using DPSO algorithm are compared with the results of memetic algorithm MA, genetic algorithm GA and simulated annealing SA. DPSO algorithm results are far superior to that of other meta-heuristic approaches for the problem considered.

An integrated multi-plant aggregate production-distribution plan generated using memetic algorithm

Many industries adopt multi-site manufacturing in order to reduce their distribution expenses. Such industries allocate the forecasted demand from various customers/demand centres to their respective plants based on gross production, inventory holding and distribution costs. But demand allocation by considering the gross production cost will not be appropriate as production in each plant can be carried out through regular, overtime and outsourcing means. Also, this will lead to allocation of the lions share of demand to a plant whose regular time production cost is cheaper than the other plants. But overtime/outsourced production cost may be costlier than the regular time production cost of other plants. In this paper, an aggregate production-distribution plan, considering all the above mentioned costs, is developed for a renowned bearing manufacturing industry in India. The proposed integer non-linear programming model is solved using memetic algorithm (MA), a hybrid form of genetic algorithm (GA) which uses simulated annealing (SA) for local search. The results obtained using MA, GA and SA are compared with the solution generated using LINGO 8.0, one of the popular operations research software.

A Simulated Annealing Algorithm for the Generation of Multi-Factory Aggregate Production Plan

Aggregate production plan (APP) aims to attain the highest profit level by utilizing all the resources efficiently. Most of the industries today are generating the APP separately for each plant even though they produce the same family of products in all the plants. APP devised for the plants after allocating the demand for the corresponding plants based on gross production and distribution costs is not optimum. The industry must work towards meeting the forecasted demand by creating a medium-term plan considering all its plants simultaneously. In this paper, APP is generated using simulated annealing (SA) for a multi-factory model, where the same family of products is manufactured at different localities. Data collected from a bearing manufacturing company is fitted to the mathematical model developed in this case. The parameters of the APP, viz., period-wise workforce level, quantity to be produced in regular production, overtime hours, outsourcing, finished goods inventory status and number of products sent from each plant to the demand centers are all normalized towards the objective of cost minimization. * Corresponding Author