Integration of distributed heterogeneous simulation models for design of manufacturing systems
Abstract
Modern manufacturing systems are complicated, dynamic organizations of people, information, materials and machines, created one piece at a time but judged on their performance as an integrated whole. The development of comprehensive modelling representations for each of these elements is important to manufacturing system design. But understanding the interactions and dependencies among the many components, within the context of a changing and dynamic system, is equally as important as the detailed consideration of each constituent element. The primary objective of this research is the development of a methodology that allows independently developed models to interoperate and create seamless and comprehensive system perspectives of manufacturing systems, facilities investigation of the interactions and dependencies among the collection of component parts, while revealing the activity of the individual elements. Model Integration Architecture is the abstract specification of the functional and structural components needed to access models, construct model networks and support communications among them for exchange of data and synchronization of their modelling activities. It is divided into two areas of concern: the organizational and structural aspects and the operational dynamics for control and coordination. A second objective was developing an operational engineering prototype system for integrating heterogeneous models in a network of distributed computers. The result of this effort was ENVISION--an ENVironment for Interactive Simulation Integration Over Networks. Existing models are logically encapsulated to isolate their implementation differences from other model network participants. The encapsulated models use message passing to support their participation in the model network configurations. ENVISION provides the mechanisms to access distributed models, link them together, and monitor and control their execution during simulation experiments; model networks may also be saved and used again for subsequent manufacturing system evaluation and design efforts. The requirements for existing models to function within an ENVISION system are illustrated with the conversion of two analytic models, CANQ and Cutree, and the extension of SLAM II, a simulation language. The applicability of model integration and its appropriate use in design of manufacturing systems is illustrated with an example three-model network.
Degree
Ph.D.
Advisors
Solberg, Purdue University.
Subject Area
Industrial engineering|Systems design|Operations research
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.