Overview of Simulation-Based Control

The control system developed at Penn State uses discrete event simulation to implement a message-based process control system. Simulation technology is extended to support real-time communication to access remote databases, and pass messages, which are used to control hardware. This makes possible the development of a generic hybrid control capability. The RapidCIM architecture and its associated tools developed at The Pennsylvania State University and Texas A&M are capable of automatically generating much of the software necessary {up to 80 or 90% of a typical application} for automating discrete manufacturing systems. This substantially reduces the cost of developing and integrating such systems, and allows a detailed simulation to be used for both analysis as well as for control.

The RapidCIM architecture was initially designed for discrete manufacturing control, and has been implemented using a real-time simulation control framework. By real-time simulation control, we refer to the use of a simulation software package (Arena RT) for direct control of a real system. Arena RT does not use a traditional event calendar but rather advances in real-time via a series of messages passed to and then back from an execution system when they have been completed. An important benefit of this approach is that much of the software development and debugging can be done by using traditional simulation for modeling shop-floor events. For example, deadlocks and other performance problems can be rapidly detected by running the simulation in the fast mode rather than in real-time. After debugging the overall software, simulated shop-floor events can be progressively replaced by plugging in actual physical equipment to develop a fully integrated shop-floor control system. It should be emphasized that the control and message passing logic are identical in the simulation and the actual shop-floor control system

At the top level, the simulation model is used to keep track of the status of the system and generates all the required messages. The simulation sends commands (messages) to the control software and then receives feedback (messages) from the system. At any given instant, the status of the production facility is reflected in the simulation model since the central controller uses the system status to execute its control logic by generating appropriate messages in real-time. The control logic also necessitates establishing communication with a planner/scheduler and reading orders.