Dynamic metamodelling in capacity planning

Abstract

Traditional capacity planning modelling techniques focus on the steady-state behaviour of a system. This is because transient behaviour complicates the problem, both conceptually and analytically, and solutions tend to be time consuming. Nevertheless, the transient behaviour of a system is often of equal consequence to that of the steady-state behaviour and requires methods for characterization. Further, today's increasingly powerful personal computers allow complex, dynamic behaviours to be modelled in a timely manner. This paper presents a model for incorporating dynamic behaviour into capacity planning. The model is demonstrated through a case study of a telephone order entry system.

Production planning with load dependent lead times: an update of research

“…This is supported by Bertsimas and Sethuraman (2002) who state that the optimality loss for fluid relaxations decreases as the number of jobs in the system increases. This is also experienced by Stahlman and Cochran (1998) and Riaño et al (2006) where the latter get slightly better approximations when the utilization values are high. Furthermore, their approximations perform less well for decreasing input functions when the system starts to empty due to the problem of transition stated by Missbauer (2006b).…”

“…This is also experienced by Lin and Cochran (1990) and Stahlman and Cochran (1998). The latter describe the transition of a workstation from one state to another and back due to a provisional decrease in capacity.…”

Aggregate order release planning for time-varying demand

“…Odoni and Roth (1983) also derived approximations for the transient behaviour of the M/M/1 model for small t, where (1) is not a good approximation. Lin and Cochran (1990) and Stahlman and Cochran (1998) used similar formulations for modelling the transient behaviour of a work centre, Lin and Cochran (1990) also give approximations for the transient behaviour if ¶ > · for a limited period.…”

Approximation of the non-stationary M(t)/M(t)/c(t)-queue using stationary queueing models: The stationary backlog-carryover approach

“…For example, the arrival rate of new jobs in a machine center may change or resources may become temporarily unavailable, see for example Lin and Cochran (1990) or Stahlman and Cochran (1998). Kolesar et al (1975) and Ingolfsson et al (2002) describe an approach for scheduling police patrol cars using the analysis of a non-stationary queueing system.…”