"Task Trees" — A Hierarchical Structure for Modelling Complex Jobshops

The throughput of clinical laboratories is determined by the "case mix" of the patients treated. Case mix is the mix of analyses determined by the types of patients treated at a hospital. This varies widely, as specific days may have different clinics. The assumptions made in order to use current simulators to model clinical laboratories (invariant case mix, superficial averaging of analyser performance, and indiscriminate scheduling) are no longer acceptable in a decision support system. A novel core engine that implements a "hierarchical" version of the discrete-event simulation approach, "task trees," is presented. This addresses the fundamental problems of "interlocking" between tasks (both within and without jobshops) and the need to model at a detailed level, i.e., scheduling individual tests on analysers. This engine is currently being used to compare many different laboratory analysers in "what if' scenarios.