Informs Annual Meeting 2017

TC57

INFORMS Houston – 2017

6 - Vendor Selection in Hospitals for Consumables and Non-consumables

order pickers. We develop closed-form expected travel time models for the class- based storage assignment policy. Also, we develop optimization models to find the optimal boundary of the classes as well as allocation of the SKUs to classes. The objective of the models is to minimize the expected travel time subject to different

Bhavin J. Shah, Associate Professor, Indian Institute of Management, Indore, Faculty Office # C-206, First Floor,, Prabandh Shikhar,, Indore - Madhya Pradesh, 453556, India, bhavinj@iimidr.ac.in, Hasmukh Gajjar, Peeyush Pandey We study various parameters governing vendor selection in hospitals for consumables and non-consumables category. We come up with unique membership functions for each category to select the vendor.

operating policies, demand profiles, and shape factors. 2 - Model Based Approach to Smart Warehousing

Leon McGinnis, Georgia Institute of Technology, Isye Dept, 755 Ferst Drive, Atlanta, GA, 30332-0205, United States, leon.mcginnis@isye.gatech.edu, Camilo Bernal-Nisperuza, Francisco Perez Hellec, Mohamed El Tonbari

A formal warehouse system model integrates multiple viewpoints, and supports design, planning and control to achieve “smart warehousing.” It also provides a basis for automating the generation of a range of analysis models to support design decision making. The use case is a highly automated facility for dispensing drugs. 3 - Integrated Cross-docking and Assignment Arpan Rijal, Erasmus University, Rotterdam, Netherlands, rijal@rsm.nl, René B.M. De Koster, Marco Bijvank Companies use cross-docking to simultaneously minimize transport and inventory holding cost. We study scheduling and dock assignment of inbound and outbound trucks at an LTL cross-dock. We develop an ALNS-based heuristic that beats state-of-the-art sequential truck scheduling and assignment methods both in solution quality and speed and that works for real-life instances. Tests on instances based on operations of a Dutch retailer reveal that the integrated model can vastly reduce internal transport and late departures. 4 - An Analysis of Storage Policies on a Multi-dock Unit-load Warehouse with Contour-line Shaped Storage Regions Mahmut Tutam, PhD Candidate, University of Arkansas, 1359 N Leverett Avenue, Fayetteville, AR, 72703, United States, mtutam@uark.edu, John A.White In this paper, the performance of a unit-load warehouse is studied when storage regions can be any shape under class-based storage policies. Specifically, unit- loads are assigned to storage regions on a priority basis. Storage regions are either rectangle shaped or contour-line shaped. Expected distance traveled with rectangle-shaped storage regions are compared with their counterpart contour- line based storage regions. Minimizing expected round-trip distance, the best rectangle-shaped storage regions are determined for different locations of docks 5 - Robotic Warehouses; State-of-the-art and Research Opportunities

TC57

362D Stochastic Optimization Sponsored: Simulation Sponsored Session

Chair: Raghu Pasupathy, Purdue University, pasupath@purdue.edu 1 - Exact and Inexact Subsampled Newton Methods for Optimization Raghu Bollapragada, Northwestern University, Evanston, IL, United States, raghu.bollapragada@u.northwestern.edu, Richard Byrd, Jorge Nocedal, Albert Solomon Berahas In this talk, We study the solution of stochastic optimization problems in which approximations to the gradient and Hessian are obtained through subsampling. We first consider Newton-like methods that employ these approximations and discuss how to coordinate the accuracy in the gradient and Hessian to yield a superlinear rate of convergence in expectation. We also consider inexact Newton methods and investigate what is the most efficient linear solver in terms of computational complexity. 2 - Optimizing the Design of a Latin Hypercube Sampling Estimator Alexander Zolan, University of Texas-Austin, 204 E Dean Keeton We present a method of reducing the variance of a Latin hypercube sampling (LHS) estimator by using an approximation of the estimator’s variance and formulating a nonlinear program. We solve this problem exactly via a dynamic program for each random component, using a discrete set of candidate stratum boundary points. We present conditions for this technique to guarantee variance reduction, compared to LHS with equal-probability strata. We assess the impact of our sampling method on bias in optimal values obtained via solving stochastic programs, versus common techniques in the literature. 3 - The Adaptive Sampling Line Search Method for Unconstrained Stochastic Optimization Raghu Pasupathy, Purdue University, West Lafayette, IN, 47906, United States, pasupath@purdue.edu, Fatemeh S. Hashemi, Michael R. Taaffe We present the adaptive sampling line search method for solving unconstrained stochastic optimization problems where the objective function can only be estimated using an inexact oracle, e.g., Monte Carlo, QMC, or numerical quadrature. We will discuss two flavors (fixed-step and backtracking search) of the proposed method along with convergence calculations and global complexity results. The complexity results are especially illuminating when compared with classical results on the complexity of line search methods in presence of an exact oracle. Street, Stop C2200, Austin, TX, 78712, United States, alex.zolan@utexas.edu, John Hasenbein, David Morton 362E Smart Facilities Sponsored: TSL, Facility Logistics Sponsored Session Chair: Margarit Khachatryan, PhD, Monsanto, Monsanto, MO, United States, margarit.khachatryan@monsanto.com 1 - Multiple In-the-aisle Pick Positions AS/RS with Class-based Storage Faraz Ramtin, Carian Consulting Group, South Plainfield, NJ, United States, faraz.ramtin@knights.ucf.edu, Jennifer A. Pazour An automated storage and retrieval system with multiple in-the-aisle pick positions (MIAPP-AS/RS) is a semi-automated case-level order fulfillment technology. The put-away to the storage locations and replenishment to pick positions are performed by the crane and the in-the-aisle case-picking is done by TC58

René B.M. De Koster, professor of Logistics and Operations Management, Erasmus University-Rotterdam, Rotterdam, Netherlands, rkoster@rsm.nl

The new generation of warehouses will be fully robotized. Managers can select from different competitive robotic techniques to fulfill the customer orders. We develop analytical performance models and use them to optimally size the systems for costs, throughput and response time, and for comparison between the most popular robotic techniques. However, so far many new systems and techniques have not received any research attention.

TC59

362F Joint Session RAS/Practice: Yard Perfomance Modeling Sponsored: Railway Applications Sponsored Session Chair: Michael F. Gorman, University of Dayton, Dayton, OH, 45469, Michael F. Gorman, University of Dayton, 300 College Park, Dayton, OH, 45469, United States, michael.gorman@udayton.edu In practice, it is a challenge to estimate yard capacity. Yard simulations attempt to describe yard operations to allow evaluation of changes in policy at a yard. Unfortunately, yard simulations are highly specific to the layout of a specific yard, and are not reusable. I develop a general purpose, reusable yard simulation model that gives good high-level estimates of yard performance and can be applied to many different yards. Model input requirements are minimal. The model provides insight int receiving and departure track use, bowl inventory levels, and car, block and train throughput. To my knowledge, no other multipurpose yard simulation is found in the literature or in use at railroads. United States, michael.gorman@udayton.edu 1 - A Multipurpose Yard Simulation Model

356