A comparison of the accuracy of discrete event and discrete time

Many combat and agent-based models use time-step as their simulation time advance mechanism. Since time discretization is known to affect the results when numerically solving differential equations, it stands to reason that it might likewise affect the results of such simulations. This paper demonstrates that is indeed the case. Using simple queueing models, we demonstrate that the size of the time step can have a substantial impact on estimated measures of performance. While large time steps can execute faster than a corresponding discrete event model, there can be substantial errors in the estimates. Conversely, with small time steps the results match both the discrete event measures as well as the analytic values, but can take substantially longer to execute.