for better ways to address this question.) This is a modif cation of a model developed by Bretthauer and Côté (1998) and Bretthauer (2000). The latter does an analytical evaluation and optimization under the assumption that this is a Jackson network of queues, where service times also have exponential distributions (not triangular or other kinds of distributions). The advantage of simulating instead is that the model can be more general and thus possibly more valid; for instance, we do not have to assume exponential service times if the data suggest otherwise. The disadvantage of simulation is that we do not get exact answers, so we must do proper statistical analysis of the output (see Chapter 6). (Thanks to Professor Jim Morris of the University of Wisconsin-Madison for suggesting these papers and this exercise.)