Reuven Karni

A Heuristic Algorithm for the Multi-Item Lot-Sizing Problem with Capacity Constraints

Abstract A heuristic algorithm is presented for solving the scheduling of several items in parallel under capacity constraints with setup and carrying costs. The method is based upon finding a lower bound solution for these costs, securing the feasibility of the solution, and improving the feasible solution so obtained until no further improvements can be made. Comparison of the performance of the proposed heuristic algorithm to that of an exact mixed-integer programming model showed that best solution costs found by the heuristic deviated on an average by 1% from the optimal values, while the computing time was on an average 1/140 of that required by the exact method.

A comparative study of multiattribute decision making methodologies

Three “real life” cases are considered in this paper to apply and compare the rankings obtained by the Analytic Hierarchy Process (AHP) and other Multicriteria Decision Making (MCDM) techniques such as Simple Additive Weighting (SAW), ELECTRE and Weighted Linear Assignment Method (WLAM). The results indicated that the AHP, SAW, and ELECTRE rankings do not differ significantly, however, the WLAM tends to exhibit more disagreement. However, because of the limited nature of this study, we do not suggest this as a general conclusion.

An engineering tool for the conceptual design of service systems

2 The system concept 67 2.1 General definition of a system 67 2.2 Distinctive characteristics of a service system 67 2.3 A system as an hierarchy of classifications and actualizations 68 2.4 The metamodel construct 68 2.5 A provisional metamodel for service systems 70

Multi-item, multi-level lot sizing with an aggregate capacity constraint

Abstract This paper presents a heuristic approach to finding the least cost MRP schedule when end products and their components are produced under a single aggregate capacity constraint. A multi-phase algorithm first obtains a feasible solution. It then proceeds to develop better cost solutions by making controlled changes to the initial feasible schedule. The best cost solutions found by the algorithm differ, on the average, by less than 1% from the corresponding best known solutions. For small problems, the average computation time is about 1 60 - th of that requireed by an exact integer programming model.

Multicriterion issues in energy policymaking

Abstract A unique opportunity has been provided, enabling us to be able to study policy problems actually faced by decisionmakers in the area of energy resources, and to develop a framework to help them to formulate policy alternatives, express preferences and obtain feedback on the outcomes of these preferences. Several MCDM methods were used to examine whether policies and their outcomes were clear-cut, or whether certain risks would be involved when preferring one policy over another. When decisionmakers were asked to express preferences by direct and indirect approaches, different rankings were obtained, probably as a result of their uncertainty about actually being able to achieve desired goals, and their implicit tradeoff between the desirability of the policy alternative and the likelihood that it could be implemented. A picture of this tradeoff provided a realistic basis upon which policy decisions could be made.

Capacity requirements planning— a systematization

Abstract The central problem of manufacturing planning is to reconcile clue dates, derived from customer delivery timetables, with schedule dates, developed from production capabilities. Conventional capacity planning techniques utilize the technological sequencing of material flow in order to build up production schedules. As they ignore interactions between work stations, and override due date requirements, these techniques are often unsatisfactory. Wight and Belt have suggested a new approach to the problem, and this paper presents a systematic methodology based on their ideas, to characterize and analyse the flow of work through a work station, and relate this flow to the nominal capacity of the station. Operation of the station is measured by work in process, delay and underload (operation below nominal capacity); flow between stations is measured by queue length and lead times (process plus wait time). Performance is evaluated by the degree of underload and overload planned for the station—the degre...

A Tool for the Conceptual Design of Production and Operations Systems

This paper proposes a conceptual design tool, based upon inferential design theory. It has been specifically developed for the design of production and operations systems, but its use can be extended to other engineering areas, such as mechanical and structural systems. Inferential design theory and its foundation in the inferential theory of learning are briefly outlined. Both theories are based on the idea of using specialised knowledge operators in learning and design, termed knowledge transmutations and design knowledge transmutations respectively. The 24 transmutations existing in the two theories are outlined, and a further 12 design-specific transmutations are proposed. These have been developed as a result of our research. A conceptual design process is proposed, in which design knowledge transmutations are used. A software tool for design, CREDO, is also described and an example of its use in the generation of design concepts for an after-sales service facility is presented. The conclusions discuss the initial methodological experience of using CREDO to generate design concepts. They are based on the introductory use of CREDO at Technion in Israel for teaching purposes. Directions for further research are also provided.

Architecture of an expert system for ergonomics analysis and design

Abstract The architecture of a computerized ergonomics expert system (ERGOEX) for ameliorating the conditions of workers at a workplace is described. The purpose of this computer-assisted tool is to transform sets of worker data and working requirements into a specification or description of an artifact, which is capable of producing ergonomic functions. The ergonomic advisor has been developed so as to reach out to both expert and non-expert users, who can thereby take advantage of this novel approach to workplace design without requiring sophisticated computing equipment. Turbo-Pascal has been chosen as language, rather than an expert system shell, or a logic-based language. In realizing the aims of “guidance” and “user friendliness” we have adopted the principle that the expert system must be able to interface with both novice and expert users. The analysis function is represented as hierarchy, workplace structure, environment, and evaluation and failure analysis at the highest level, and further breakdowns at lower levels. The design process is divided into data input, generation of recommendations, critiquing, and referral to sources for basic supplementary information. Relevance to industry The article describes an expert advisory system to be used by Industrial Engineers, Ergonomists, factory personnel and medical staff for analysis, evaluation and design of a workplace. The methodology has been developed for both expert and non-expert practitioners, who can operate the system using an interactive man–computer dialogue.

Accelerated enterprise process modeling through a formalized functional typology

An enterprise process model encompasses a set of business processes implemented or to be implemented in the enterprise. As such, it expresses the requirements of the organization and thus constitutes a compulsory prerequisite for the successful implementation of process-based IT systems such as ERP, SCM and CRM. However, there is a lack of an enabling science to guide the generation of an individualized process model for a particular enterprise. Conceptually, content based enterprise process modeling – itemizing the processes carried out within the enterprise – is based on the assumption of similarity between enterprises that operate within a given industrial sector, so that a generic model should be applicable, with some customization, to all enterprises within that sector. Our approach is based upon the premise that enterprises are characterized by their functionalities, rather than by their end products or technologies. We thus propose a method which enables the functionality of a specific enterprise to be defined; and from this definition a unique enterprise process model can be generated to constitute a statement of the business processes of the enterprise.

Dynamic Algorithms for Input-Output Planning of Work-Station Loading

Abstract The central problem in manufacturing planning is to reconcile due dates with schedule dates. Conventional (static) scheduling methods utilize expected process times, constant work station capacities, and technological sequencing to develop schedule dates, which rarely match planned due dates. A different approach, input-output control, is based upon throughput times (wait plus process), variable work station capacity, and lead times (from input to output) to try to match scheduled output dates with due dates. This paper presents several dynamic programming and transportation-type algorithms for determining optimal output level profiles under restrictions of permissible work station capacity levels and wait queue sizes.