You are here

  1. Home
  2. » Forschung
  3. » Preprints
Preprint-No.: <   407   >   Published in: August 2014   PDF-File: IGPM407.pdf
Title:Integrating Release and Dispatch Policies in Production Models Based on Clearing Functions
Authors:Dieter Armbruster, Michael Herty, Xinping Wang, Lindu Zhao
Aggregate production planning for highly reentrant production processes is typically generated by finding optimal release rates based on clearing function models. For production processes with very long cycle times, like in semiconductor production, dispatch policies are used to cover short term fluctuations. We extend the concept of a clearing function to allow control over both, the release rates and priority allocations in re-entrant production. This approach is used to improve the production planning problem using combined release and the allocation dispatch policy. The control parameter for priority allocation, called the push-pull point (PPP), separates the beginning of the factory which employs a push policy from the end of the factory, which uses a pull policy. The extended clearing function model describes the output of the factory as a function of the work in progress (wip) and the position of the PPP. Its qualitative behavior is analyzed. Numerical optimization results are compared to production planning based only on releases. It is found that controlling the PPP significantly reduces the average wip in the system and hence leads to much shorter cycle times.

Note to Practitioners:
This study is focussed on the semiconductor production industry where multiple passes through the same machines (re-entrant production) are common. We show that changing priority rules is essentially a re-allocation of spare production capacity which may lead to significantly reduced lead times. In addition, a control scheme that determines the changes in the priority rules is shown to lead to smaller safety stocks for large demand fluctuations compared to a static priority rule.
Keywords:production planning, dispatch control, partial differential equations, reentrant production
DOI: 10.3934/nhm.2015.10.511
Publication:Networks and Heterogeneous Media (NHM)
Pages: 511 - 526, Volume 10, Issue 3, September 2015