Jakob Asmundsson
Sangwook Hwang
Reha Uzsoy
Laboratory for Extended Enterprises at Purdue
School of Industrial Engineering
1287 Grissom Hall
Purdue University
West Lafayette, In 47907-1287
ph: (765) 494-0829
fax: (765) 494-5448
uzsoy@ecn.purdue.edu
A fundamental problem in
the domain of production planning has been that of developing computationally
tractable aggregate planning models that accurately capture the nonlinear
relationships between workload and lead times. Most current planning algorithms
either use lead time estimates that are independent of workload, essentially
assuming infinite capacity, or use detailed transaction-level models to
capture shop dynamics, resulting in poor scalability. We give a brief review
of existing models in this area, and propose a nonlinear approach based
on clearing functions that yields extremely promising results. We provide
computational results examining different aspects of such models, such
as the effects of different shop floor scheduling policies and lot sizing
techniques.