Optimization of production control policies in failure-prone homogenous transfer lines

The production control of homogenous transfer lines with machines that are prone to failure is considered in terms of inventory and backlog costs. Because problem complexity grows with line size, a heuristic method based on the profile of the distribution of buffer capacities in moderate size lines is developed in order to enable the optimization of long lines. A method consisting of an analytical formalism, combined discrete/continuous simulation modeling, design of experiments and response surface methodology is used to optimize a set of transfer lines, with one parameter per machine, for up to seven machines. A profile in the parameter distribution which can be modeled using four-parameters is observed. Consequently, the optimization problem is reduced to four parameters, in turn greatly reducing the required optimization effort. An example of a 20-machine line, optimized at 130 runs, versus 5243 090 runs that would be necessary to solve the 20-parameter problem, is presented to illustrate the usefulness of the parameterized profile.