Hybrid Decision Support and Justification Methods for Production System Selection
This article presents different decision support and justification method principle for production system selection and development. Focus is on the analytic approaches, discrete event simulation, queue theories and linear programming, the other approcahes, strategic and economics are only listed, even if they are also of great importance. The benefit of discrete event simulation is discussed and also reasons not to use simulation is shown. The analytic methods, queue theories, linear programming, i.e. optimisation are useful and can be used before or parallel to simulation modelling. Use of different methods is compared and compination of different methods, i.e hybrid methods is discussed. Author is proposing to use hybrid methods, different type of modelling and simulation in production system concept creation and selection. The article is based on the author’s experience in assembly system design and on a literature study.
| Attachment | Size |
|---|---|
| paper81.wmv | 6.29 MB |
Thank you for the comment.
Actually I created the names my self to the figure. They were not in the source: Sheng-Jen (Tony) Hsieh. Hybrid analytic and simulation models for assembly line design and production planning.
I agree embedded might be a wrong word, integrated could be better.
In general my interest is to find analytical methods to supoort simulation studies, hybdrid or integrated methods or replace the simulation (Discrete event simulation) with analytuc methods, like queue theories or optisation (interer, linear programming)
Regards
Juhani
Dear Professor Heilala,
I would favour the analytical approach whenever appropriate as it is amenable to rigorous proof and thus can avoid uncertainty.
However, most real problems are complex and the analytical approach is not always applicable.
Would a suitable hybrid approach be to simplify the problem, apply analytical tools to obtain approximate results and then employ simulation to produce a more detailed solution?
Thank you.
D Pham.
My ideas are similar, with analytical,simpilified model limit solution space, later verify with accurate simulation model. Naturally it depends on the case.
Analytic calculation and some optimisation brings results fast, but due simplicity it does not cover all the complexity. So we need to verify results with some more precise method.
Regards Juhani
Hi,
With regard to Fig. 4, I'd suggest another classification system as follows:
1. parallel system(I & II in your paper)
2. Supporting system(rather that embedded): A support S, S support A(III & IV)
3. Serial: A after S, S after A(V & VI)
What's your opinion on that?
Thanks...