The paper suggests an approach for optimizing a material ow system consisting of two work-stations and an intermediate bu??er. The material ow systemmay be a production system, a distribution system or a pollutant-deposit/removal system. The important characteristics are that one of the work-stations is unreliable (random breakdown and repair times), and that the performance function is formulated in average terms. The performance function includes random production gains and losses as well as deterministic investment and maintenance costs. Although, on average, the performance function is smooth with respect to parameters, the sample performance function is discontinuous. The performance function is evaluated analytically under general assumptions on cost function and distributions. Gradients and stochastic estimates of the gradients were calculated using Analytical Perturbation Analysis. Optimization calculations are carried out for an example system.
|Title of host publication||Probabilistic constrained optimization: Methodology and applications|
|Place of Publication||Dordrecht|
|Publisher||Kluwer Academic Publishers|
|Publication status||Published - 2000|