A note on optimal allocation of work in stochastic assembly systems

We consider the problem of work allocation in stochastic assembly systems so as to maximize throughput. A recent study showed that unbalancing the work allocation in the direction of assigning less work to assembly and more to component fabrication improves the expected throughput rate. The computational results showed that, for certain process time distributions, the expected throughput rate could go up with an increase in the number of components. This result contradicts the design-for-assembly practice of minimizing the number of components. This paper provides analytical insights into the effect of the number of components and process time distributions on the expected throughput rate. We then present a more realistic assembly system where the results resemble the design-for-assembly practice.