Performance assessment of mixed methods for time integration using an assembly-type queueing model

Allocation of integration methods to processors and selection of time-step sizes affect the solution time of a. mixed-method time integration problem. In this paper it is proposed that an assembly-type queueing model be used to quantify these effects by demonstrating that data exchanges among integration methods are similar to product flows in assembly systems. An overview is provided that establishes the analogy between data exchanges and product flows. A basic queueing model is derived for synchronous data transfers and is demonstrated using an example coupled-code: problem.