William G. Ferrell

Dynamic rescheduling that simultaneously considers efficiency and stability

Dynamic job shop scheduling is a frequently occurring and highly relevant problem in practice. Previous research suggests that periodic rescheduling improves classical measures of efficiency; however, this strategy has the undesirable effect of compromising stability and this lack of stability can render even the most efficient rescheduling strategy useless on the shop floor. In this research, a rescheduling methodology is proposed that uses a multiobjective performance measures that contain both efficiency and stability measures. Schedules are generated at each rescheduling point using a genetic local search algorithm that allows efficiency and stability to be balanced in a way that is appropriate for each situation. The methodology is tested on a simulated job shop to determine the impact of the key parameters on the performance measures.

Integrated production and distribution scheduling with a perishable product

This research focuses on the practical problem of a perishable product that must be produced and distributed before it becomes unusable but at minimum cost. The problem has some features of the integrated production and distribution scheduling problem in that we seek to determine the fleet size and the trucks’ routes subject to a planning horizon constraint. In particular, this research differs because the product has a limited lifetime, the total demand must be satisfied within a planning horizon, multiple trucks can be used, and the production schedule and the distribution sequence are considered. A mixed integer programming model is formulated to solve the problem and, then, heuristics based on evolutionary algorithms are provided to resolve the models.

A hybrid random-key genetic algorithm for a symmetric travelling salesman problem

This paper describes a methodology that finds approximate and sometimes optimal solutions to the symmetric Travelling Salesman Problem (TSP) using a hybrid approach that combines a Random-Key Genetic Algorithm (RKGA) with a local search procedure. The random keys representation ensures that feasible tours are constructed during the application of genetic operators, whereas the genetic algorithm approach with local search efficiently generates optimal or near-optimal solutions. The results of experiments are provided that use examples taken from a well-known online library to confirm the quality of the proposed algorithm.

Evaluating simple scheduling rules in a mixed shop environment

The layout and flow of real manufacturing shops are most often neither pure job-shop nor pure flow-shop. This paper presents the results of a simulation-based study in which some of the most common and simple scheduling rules are tested in mixed shops containing jobs with both job-shop and flow-shop routing. Specifically, the rules that are used are first in first out, shortest processing time, modified shortest imminent operation, non-decreasing slack, and modified slack. Two performance measures are considered, a modified mean flow time to assess the amount of time that jobs spend waiting for processing and the cost of earliness and tardiness to evaluate how closely the rules complete jobs relative to their due date. The data is analyzed using both graphical and hypothesis tests so that not only conclusions based on general trends could be drawn, but also assessed with statistical rigor. Using these techniques, the results are somewhat mixed but, in general, indicate that the shortest processing time rule is preferred when the mean flow time is the performance measure and modified slack is preferred when cost is the performance measure. By adding additional insight from the actual simulation runs to these conclusions, we recommend that practitioners can likely get good performance by using either of these rules as the basis for a heuristic that adds logic to avoid extreme conditions.

Quality assurance and stage dynamics in multi-stage manufacturing. Part II

SUMMARY This is the second in a two-part report on control engineering considerations in quality assurance. The first paper introduced some basic concepts in the measurement of the ‘loss’ due to poor product quality, and approached the issue of parameter optimization of the control system through the modem of manipulating the PID constants, using an experimental approach coupled with simulation as the methodology to achieve the desired objective. This paper approaches the problem of parameter optimization via mathematical programming, and then addresses the issue of setting the standard of system performance through Markov decision processes.

Virtual sign animated pedagogic agents to support computer education for deaf learners

In Thailand, there is a desperate need to improve the educational opportunities for deaf and hearing-impaired university-aged students. The research described herein is targeted at two aspects of this issue: 1) design and test a virtual sign animated instructional tool for the Thai sign language, and 2) investigate how instructors of university classes can most effectively teach deaf students and assess their performance, particularly by using this tool. Currently, I have completed a thorough review of the literature, constructed a 3-D human model that includes rudimentary facial expressions, and devised a framework to investigate the instructional process.

Production planning and control in a JIT environment

Abstract In addressing planning, design, and control issues of a production system, the solution of a normative model provides answers to planning issues at the aggregate level but fails to capture the time-dependent behavior of the system. This paper illustrates the usefulness of using both mathematical programming and simulation modelling to investigate the material flow characteristics of a “Just in time” system with part quality requirements.

An interactive memetic algorithm for production and manufacturing problems modelled as a multi-objective travelling salesman problem

In this paper, a preference-based, interactive memetic random-key genetic algorithm (PIMRKGA) is developed and used to find (weakly) Pareto optimal solutions to manufacturing and production problems that can be modelled as a symmetric multi-objective travelling salesman problem. Since there are a large number of solutions to these kinds of problems, to reduce the computational effort and to provide more desirable and meaningful solutions to the decision maker, this research focuses on using interactive input from the user to explore the most desirable parts of the efficient frontier instead of trying to reproduce the entire frontier. Here, users define their preferences by selecting among five classes of objective functions and by specifying weighting coefficients, bounds, and optional upper bounds on indifference tradeoffs. This structure is married with the memetic algorithm – a random-key genetic algorithm hybridised by local search. The resulting methodology is an iterative process that continues until the decision maker is satisfied with the solution. The paper concludes with case studies utilising different scenarios to illustrate possible manufacturing and production related implementations of the methodology.

Integration of interim quality status information in discrete-part process control modelling

Abstract This paper deals with the modelling of a class of discrete-part production control problems that explicitly joins product quality with optimal production control actions. A multistage production system is the basis for modelling, each stage consisting of a processing aspect and a rework/repair aspect. It is assumed that the “quality state” of a part can be measured with certainty at several points within each stage and that control actions can be exercised over each aspect. The state transition function depends on the control action taken and follows a predetermined distribution. Within this general framework, two related problems are studied: (1) independence both stage-to-stage and between states and (2) independence only between states. Both problems can be formulated in a {0, 1} nonlinear mathematical programming format, and numerical examples of each are provided.

US-Mexico Industrial Engineering Education Partnership: Clemson University and University of Sonora

This paper outlines an international engineering program between the Industrial Engineering Departments of Clemson University and University of Sonora focused on integrating education and research with industrial collaboration. The paper provides a brief rationale for the need to develop such a program, overall partnership model, potential impact, and assessment to support the successful evaluation of the program.