In this paper, a virtual factory is proposed in the form of a distributed simulation model which provides a tool for performance evaluation of large and complex manufacturing systems. The virtual factory is constructed on a computer network developed for c
factory is large and the size of a shop is set to belarge, as assumed in item 1. This suggests that thefollowing assumption is acceptable: the materialscompleted at Area i in a period will not be used inanother shop, e.g., Area j, in the same period evenif material becomes ready for processing.
3.The capacity of the loading/unloading area at ashop is large, and no blocking of material flow willoccur due to the shortage of the space in which tokeep the materials.
The period in item 1 is named as a Time Bucket inthis study and following terms are defined: BT
Bucket Time (time length of a Time Bucket) TBk
k-th Time Bucket in the simulation run STBk, ETBk
Times in GVT at the beginning of and at the end ofk-th Time Bucket, TBk, respectively. STBk+1 = ETBk= STBk + BT4.3Material Flow and Synchronization Mechanism
When a subsystem such as Area or T-Process is mod-eled in one simulation system, the arrival at and de-parture from the subsystem are represented by gener-ating and absorbing nodes of transactions, often
called Source and Sink nodes in simulation languages.In constructing a virtual factory as a distributed simu-lation system, these nodes in a model of a subsystemare connected, as in Figure 5. We need to notice thedifferent nature of these nodes when subsystems areconnected. Arrival of material will occur at any time;
therefore, an accurate realization of material flow de-pends on the method to represent the arrival of mate-rials. In a distributed simulation, material flow maybe realized by transmitting a message generating anarrival event in the subsystem of destination, and asynchronization mechanism which can respond to thematerial flow is essential to process the arrival eventin a chronological order with no conflict in the sub-system. The conservative and Time Warp mechanismsare proposed for synchronization [9] and are widelyused [10]. The Time Warp mechanism, however, re-quires state saving and exchanging functions for roll-back operations. In simulation languages commer-cially available, a state-saving function is provided torestart a new simulation run from the state saved inother runs using a state exchanging function. Thismeans that these functions in presently available com-mercial simulation languages are not applicable to therollback operation, and that a distributed simulationsystem with a Time Warp mechanism has to be devel-oped by a general programming language.
5.Time Bucket Mechanism
As suggested above, existing approaches for dealingwith synchronization in distributed simulation (e.g.,Time Warp) are not well suited to the needs of a CIMsimulation system (i.e., virtual factory). The adapta-tion of existing programs for the development of thevirtual factory has substantial advantages.
A Time Bucket mechanism was proposed by theauthors [11, 12] as a mechanism for synchronizationwhich circumvents the drawbacks described in thepreceding section. The method is based primarily on

observations of real manufacturing systems.
Figure 5. Connection of subsystems
MARCH 1999
SIMULATION5
system for job-shop manufacturing- A system ...Through the integration of these mechanisms into a...An improved genetic algorithm for the distributed ...
system for job-shop manufacturing- A system ...Through the integration of these mechanisms into a...An improved genetic algorithm for the distributed ...
热门文档