Antoni Guasch

Optimization of Logistic and Manufacturing Systems through Simulation: A Colored Petri Net-Based Methodology

Simulation models have proved to be useful for examining the performance of different system configurations and/or alternative operating procedures for complex logistic or manufacturing systems. However, when applying simulation techniques to increase the performance of those systems, several limitations arise due to their inability to evaluate more than a fraction of the immense range of options available. Simulation-optimization is one of the most popular approaches to improve the use of simulation models as a tool to obtain the best (optimal or quasi-optimal) decision variable values that minimize a certain objective function. However, despite the success of several simulation-optimization packages, many technical barriers still remain. The authors describe a new approach to integrate evaluation (simulation) methods with search methods (optimization) based on not only simulation results but also information from the simulation model.

Deadlock-Free Scheduling Method for Flexible Manufacturing Systems Based on Timed Colored Petri Nets and Anytime Heuristic Search

This paper addresses the deadlock (DL)-free scheduling problem of flexible manufacturing systems (FMS) characterized by resource sharing, limited buffer capacity, routing flexibility, and the availability of material handling systems. The FMS scheduling problem is formulated using timed colored Petri net (TCPN) modeling where each operation has a certain number of preconditions, an estimated duration, and a set of postconditions. Based on the reachability analysis of TCPN modeling, we propose a new anytime heuristic search algorithm which finds optimal or near-optimal DL-free schedules with respect to makespan as the performance criterion. The methodology tackles the time-constrained problem of very demanding systems (high diversity production and make-to-order) in which computational time is a critical factor to produce optimal schedules that are DL-free. In such a rapidly changing environment and under tight customer due-dates, producing optimal schedules becomes intractable given the time limitations and the NP-hard nature of scheduling problems. The proposed anytime search algorithm combines breadth-first iterative deepening A* with suboptimal breadth-first heuristic search and backtracking. It guarantees that the search produces the best solution obtained so far within the allotted computation time and provides better solutions when given more time. The effectiveness of the approach is evaluated on a comprehensive benchmark set of DL-prone FMS examples and the computational results show the superiority of the proposed approach over the previous works.

A Methodology for Solving Logistic Optimization Problems through Simulation

In this paper we present a methodological approach designed to automate the decision-making in logistic systems, with deterministic time, by solving optimization problems. The Colored Petri Net (CPN) formalism has been used as a base to develop a methodology that integrates the features of operational research, artificial intelligence and simulation fields. At the same time, it combines the modeling of discrete event systems with simulation, analysis and system optimization, transforming a conceptual model into a simulation model, and a decision problem into a search problem. The use of the CPN formalism has allowed the integration of all of these different research fields into a unique decision support tool.

CSP GENERATION FROM PETRI-NETS MODELS

Abstract The use of traditional production planning techniques is constrained by large numbers of decision variables, uncertainty in demand and time production, and non-deterministic system behaviour, characteristics intrinsic in manufacturing. The aim of this paper is to present a methodology that combines the modelling power of petri-nets (PN) to represent both manufacturing architecture and production logistics, together with the optimisation performance given by constraint programming (CP). While PN can represent the entirety of any system, CP is effective in solving large problems, especially in area of planning. The foundations to generate a Constraint Satisfaction Problem (CSP) from a PN are given.

PN to CSP Methodology: Improved Bounds

Traditional production planning techniques are constrained by large numbers ofdecision variables, uncertainty in demand and time production, and non-deterministic system behaviour (intrinsic characteristics in manufacturing). This paper presents an improvement to a methodology in the area ofKno wledge Based Systems (KBS) which generates automatically Constraint Satisfaction Problems (CSP), using Petri-nets (PN) to model the problem and Constraint Programming (CP) in the solution. The methodology combines the modelling power of PN to represent both manufacturing architecture and production logistics, together with the optimisation performance given by CP. While PN can represent a whole production system, CP is effective in solving large problems, especially in the area ofplanning. The improvement raises from the design of a more complete algorithm to calculate the transitions firings bounds, and hence to remove useless problem variables.

Elements of Statistics for Simulation

This chapter presents statistical concepts and definitions that are used when designing a simulation model. We start, in the first instance, by considering a conceptual model, then the need to verify the initial data for the model, followed, if necessary, by the data that can be adjusted to some probability distribution, where validating this adjustment also involves statistical concepts. Later, once the results are in, we consider the experiments that need to be done, as well as the replications, ending up with the analysis of the data obtained.

Introduction to Digital Simulation

A computer simulation is an attempt to model a process from a real or hypothetical system by means of a computer program in order to observe, analyze and improve its behavior. In more practical terms, simulation can be used to forecast the future behavior of a system and determine what can be done to influence this behavior.