[ewilliams]

A good way to begin is with a very simple black-box model. There will be four black boxes: Preprocessing, Chemistry, Hematology and Microbiology. Check with the client (the management familiar with the laboratory) on all possible paths a specimen can take through the model (for example, Preprocessing, then Hematology, then Chemistry may be one possible path). It seems likely that all possible paths begin at Preprocessing. A simple animation will let you and the client check that the specimens (entities in the model) are moving in valid sequences. Use "first guesses" with basic distributions (constants at first, then simple distributions such as triangular) for time spent in each black box, and tweak until overall results (throughput and time-in-system) look reasonable. To determine which black box most likely can yield the quickest overall process improvements, examine these preliminary results and ask the client. Then refine the model by adding detail to that black box modeled separately (e.g., in computer science terms, make it a "white box").