Reducing simulation model complexity by using an adjustable base model for path-based automated material handling systems - a case study in the semiconductor industry

Abstract

Usually simulation studies of automated material handling systems of semiconductor fabs are extremely time consuming. This is due to the high detail of models used for investigations which are partly provided by transportation system suppliers. These models only provide poor possibilities for adjustment and are computationally expensive. The article will address these issues by proposing an adjustable supplierindependent base simulation model. It allows easy building, adjusting, and running simulation models of path-based systems without deeper programming knowledge. A use case of Infineon's Dresden fab revealed simulation results with an accuracy in the same range as the supplier models while disregarding a few details and thus showing significant time savings in modeling and adjusting the system as well as running simulation studies. This can be done by choosing an appropriate level of abstraction.

Model simplification in manufacturing simulation – Review and framework

“…Choice of model detail may be guided by generic model structures. For example, Rank, Hummel, Schmidt, and Schneider (2015) propose an adjustable base model addressing automated material handling systems of semiconductor fabrication plants. Likewise, Duarte, Fowler, Knutson, Gel, and Shunk (2007), and Pehrsson, Frantzén, Aslam, and Ng (2015) suggest to model manufacturing plants and supply chains by a simplified model, concerning a few standard components.…”

Maximizing traffic flow of automated guided vehicles based on Jamology and applications

“…Thus, the rear AGVs must completely stop until all of the front AGVs relinquish permission. We obtain the transfer flow using basic Newtonian mechanics, such as in equation (4).…”

“…This problem has been solved using generic models. 4 However, a good result does not always reflect true best performance because it is a result of trials and errors. Queueing network theory and Markov models are often used as a statistical approach.…”

The correct level of model complexity in semiconductor fab simulation — Lessons learned from practice

“…For daily operative decisions simple guidelines had to be applied. While it is generally agreed that simple results before deadline are better than detailed results arriving too late [14], the authors' experience has shown that, depending on the system's layout and the workload, models with an "optimized" level of detail show a benefit factor in runtime of 10 up to 100 [15].…”

“…Due to highly dynamic system behavior, discrete event simulation models were applied instead of static or analytic approaches. The models were built from a generic model presented in [15]-hence they use the identical operation logic. The major decision to be made was whether a resource-driven or a job-driven model, see e. g. [33], should be established.…”