Catherine M. Harmonosky

An improved simulated annealing simulation optimization method for discrete parameter stochastic systems

This paper proposes a new heuristic algorithm for the optimization of a performance measure of a simulation model constrained under a discrete decision space. It is a simulated annealing-based simulation optimization method developed to improve the performance of simulated annealing for discrete variable simulation optimization. This is accomplished by basing portions of the search procedure on inferred statistical knowledge of the system instead of using a strict random search. The proposed method is an asynchronous team-type heuristic that adapts techniques from response surface methodology and simulated annealing.Testing of this method is performed on a detailed simulation model of a semi-conductor manufacturing process consisting of over 40 work-stations with a cost minimization objective. The proposed method is able to obtain superior or equivalent solutions to an established simulated annealing method during each run of the testing experiment.

A multi-factor plant layout methodology

Abstract This paper presents a formulation for the plant layout problem, incorporating more than two input factors that may be either qualitative or quantitative in nature. A heuristic is presented which combines a construction and improvement procedure to quickly develop a good layout from the formulation. The output layout from the heuristic may be used directly as the final layout or it may be used as a good starting layout for any other computerized layout improvement procedure. The heuristic is illustrated with a practical example of a large plant with 36 departments.

Implementation issues using simulation for real-time scheduling, control, and monitoring

Some of the most significant implementation issues of using simulation for real-time scheduling and control decision-making in a computer-integrated manufacturing system are presented. The issues discussed include modeling structure, interfacing to the physical system, saving system status for evaluating alternatives, and recovery at decision points. The discussion is based on experience working with an existing discrete event simulation language applied as a real-time decision tool in a laboratory setting as well as on interchanges with other industrial experts working in this area.<<ETX>>

Simulation-based real-time scheduling: review of recent developments

With increased usage of computer monitoring of the factory has come increased industrial and academic interest in developing more science-based real-time scheduling techniques. Simulation, best known for its application during system design or modification, is also being considered for use on day-to-day scheduling problems. This paper reviews recent developments using simulation as a tool in real-time scheduling.

A real-time methodology for minimizing mean flowtime in FMSs with routing flexibility: Threshold-based alternate routing

This paper presents an efficient continuous real-time routing strategy, namely threshold-based alternate routing (TAR), to minimize mean flowtime of parts in a FMS with routing flexibility. TAR routes parts to alternate machines instead of their primary machines when the benefit in terms of waiting time obtained from routing to an alternate machine exceeds a pre-determined threshold value. This study proposes that the threshold value for each manufacturing system is unique and presents a methodology for determining its unique value. The threshold concept and the performance of TAR in minimizing mean flowtime are tested with extensive experimentation, involving intricate experimental design. TAR provides very significant improvements in system performance measures compared to other real-time rerouting methods and shows that the threshold value is unique and dependent on system parameters for each manufacturing system. The relationship between the threshold value and system parameters has also been determined.

Optimization of systems with multiple performance measures via simulation: Survey and recommendations

Computer simulation is a powerful evaluative tool for the design and configuration of stochastic systems. Methodologies are needed that assist simulation end users in locating high quality design alternatives to the real system through the use of their simulation models. This paper focuses specifically on the simulation optimization problem that involves multiple performance measures. It surveys available methodologies for this problem and discusses notable strengths and weaknesses of each. Suggestions are made for future research that alleviates the common weakness of the reviewed literature.

A simulation optimization method that considers uncertainty and multiple performance measures

Abstract Simulation optimization provides a structured approach to system design and configuration when analytical expressions for input/output relationships are unavailable. This research focuses on the development of a new simulation optimization technique applicable to systems having multiple performance measures. The aim of this research is to incorporate a simulation end user’s preference towards risk and uncertainty into the search process for the best decision alternative. Automation of the optimization procedure is a necessity. Therefore, this paper proposes a simulation optimization method that involves a preference model, specifically adapted for decision making with simulation models. The proposed simulation optimization method is evaluated against two simulation optimization methods with embedded deterministic, multiple criteria decision making strategies. It is shown on average to obtain significantly better solutions in multiple types of experimental settings having normally distributed simulation performance measures.

Simulation in a CIM environment: structure for analysis and real-time control

Simulation has long been recognized as a valuable tool for analyzing manufacturing systems. It is effective for assessing the impact of changing system parameters (e.g. reducing processing time) upon system performance measures, and it can also aid decisions concerning system configuration. At Penn State, simulation is playing an important role in the development of a Computer Integrated Manufacturing Laboratory. Currently, it is being used as an analysis tool studying system design and computer communication issues. Future plans are to use simulation as a real-time scheduling and control tool. Due to this ultimate long-term goal of the simulation model use, the model structure is different than traditional manufacturing applications of simulation. Rather than having events associated with workpiece movement and processing drive the simulation, computer communication events drive the model. This paper first discusses the general considerations involved when applying simulation as an analysis tool and potential real-time control tool in the CIM environment. The paper then discusses these analysis and real-time control issues in detail using the Penn State CIM Lab application as an illustrative example.

The system generator concept for FMS research and analysis

SUMMARY In the research area of FMS, much past work has concentrated upon unique systems analysing production control strategies. In some cases, no physical system was available as reference, necessitating a variety of system assumptions by researchers. It has been very difficult to compare control strategy effectiveness among applications or among researchers due to many different system assumptions and a general lack of standardization among system definitions. This paper discusses development of a system generator procedure, which would greatly enhance generic control strategy research by allowing rapid generation of test system environments. The basic issues and considerations involved in development of a system generator are discussed. Considering these issues, an implementation of a system generator procedure is presented along with an example of output system representations for given inputs to the generator procedure.

A real-time methodology for minimizing mean flowtime in FMSs with machine breakdowns: threshold-based selective rerouting

This paper presents the threshold-based selective rerouting (TSR) heuristic to minimize mean flowtime of parts in a flexible manufacturing system (FMS) with machine failures. Most of the available methodologies considering machine failures are either off-line methods or propose different versions of rerouting all the jobs to alternate machines. The objective of this methodology is to develop an efficient real-time rerouting strategy, which can be easily applicable in case of machine failures and is able to adapt to the dynamic features of an FMS. Parts in the queue of a failed machine are selectively rerouted to alternate machines when the benefit in terms of waiting time obtained from rerouting exceeds a specific pre-determined threshold value. The threshold concept and the performance of TSR in minimizing mean flowtime are tested with extensive simulation experiments. It has been shown that TSR provides significant improvements in system performance measures compared to other real-time rerouting methods and that the threshold value is dependent on system parameters. The relationship between the threshold value and system parameters has also been determined.