4 Presentations

• 
  04:30 PM - 04:52 PM

  Comparison of Single-Sourcing (with Order Splitting) and Dual-sourcing in the Presence of Stochastic Processing Times

  • Niloy Mukherjee, presenter, Virginia Tech
  • Subhash C. Sarin, Professor, Virginia Tech

  The strategy of ordering material for the production of a lot of identical, discrete items (products) from two vendors (called dual-sourcing) has been wellstudied in the literature. This strategy has been shown to result in a shorter lead time than that obtained by ordering material from either vendor alone. However, this strategy induces additional ordering cost. In this paper, we study an alternative strategy of sourcing the material from a single supplier but by permitting its delivery in two partial shipments (called single-sourcing with order splitting). Such a splitting of a lot into sublots has been studied in the deterministic scheduling area pertaining to economic lot sizing and lot streaming. However, in this paper, we study it in a stochastic environment by allowing processing times to be stochastic. Besides affording a lower ordering cost, we show that single-sourcing (with order splitting) incurs lower inventory and offers a better stockout risk behavior than that incurred and offered, respectively, by dual-sourcing.

• 
  04:52 PM - 05:14 PM

  A Strategic Capacity Allocation Model for Aligning Product Family Portfolios with Supply Chain Manufacturing Features

  • Sylverin Kemmoe, CRCGM
  • Pierre-Alban Pernot, presenter, LIMOS
  • Nikolay Tchernev, Université d'Auvergne

  Because of customization trend, firms must design products in order to regroup these into product families according to parts commonalities; this opera-tion resulting in reduction of loss of capacity and in raw material wages. However in order to be effective prod-uct family definition must be realized in a long term planning process considering supply chain features and inversely. In this paper, we propose a model allowing supply chain manager to measure impact of product family definition on demand satisfaction and supply chain design and operational costs. Confronting differ-ent scenarios, then it is possible to decide how define product family and how design supply chain. Among elements that we take account for, we can find resources constraints (minimum amount, purchase costs and transfer) but also capacity lost, capacity limitation, capabilities and inventory costs.

• 
  05:14 PM - 05:36 PM

  The Segmentation Method for Long Line Optimization

  • Chuan Shi, Massachusetts Institute of Technology
  • Stanley B. Gershwin, presenter, Massachusetts Institute of Technology

  In this paper, we present a segmentation method for long line optimization, in which we maximize profit (a function of production rate, buffer sizes, and average inventory). Instead of optimizing the original long line, the segmentation method divides it into several short lines, optimizes these short lines separately, and combines the optimal buffer distributions to find an approximately optimal buffer distribution of the original line. This method reduces the computer time for long line optimization dramatically. Both heuristic explanations and numerical experiments are provided to show the accuracy and speed of the method.

• 
  05:36 PM - 05:58 PM

  Performance Evaluation of Stochastic Production Systems : Discretization and Decomposition

  • Jean-Sebastien Tancrez, presenter, Université catholique de Louvain

  The performance evaluation of stochastic production systems is crucial to support managers decisions as well as challenging for researchers. In this paper, we propose a new methodology to analyze production systems with general assumptions: assembly/disassembly systems, general processing time distributions and finite storages spaces. The general distributions are first discretized by probability mass fitting, and the transformed system is then analytically modelled by decomposition. The system is decomposed into two station subsystems and the processing time distributions of the virtual stations are iteratively modified to approximate the impact of the rest of the network, adding estimations of the blocking and starving distributions. Decomposition allows to analyze large systems in a reasonable computational time (unlike exact models), and with good accuracy. Computational experiments show that the relative error is on the order of one percent, and less with buffer sizes larger than two. Moreover, as it allows a fine approximation of the blocking and starving time distributions, PMF seems to bring an improvement in the application of the decomposition technique.