Informs 2015

197 / 552

INFORMS Philadelphia – 2015
195
5 - Data Mining for Problem-Specific State Space Design in Routing
Applications
Dirk Mattfeld, Technische Universität Braunschweig,
Braunschweig, 38106, Germany, ,
Ninja Soeffker, Marlin Ulmer
We consider a dynamic routing problem, where a vehicle serves customer
requests arriving stochastically over time. Due to a time limit, not every request
can be served. Rejections are possible to maximize the overall number of served
requests. For anticipation of future requests, we apply value function
approximation (VFA). For improvement of the approximation process, we
combine VFA with data mining operations to derive a problem specific VFA-state
space based on the observed problem states.
MB68
68-Room 201B, CC
Green Vehicle Routing
Sponsor: Transportation, Science and Logistics
Sponsored Session
Chair: Mesut Yavuz, Associate Professor, University of Alabama, Alston
Hall Box 870226, Tuscaloosa, AL, 35487, United States of America,
1 - Pathways to Green Logistics: Past, Present and Future of the
Green Vehicle Routing Problem (GVRP)
Sevgi Erdogan, Faculty Research Associate, University of
Maryland-NCSG, 1112 J Preinkert Field House, College Park, MD,
20742, United States of America,
This talk will give a brief background to the GVRP and a review of the variants
and extensions to the problem as well as techniques to their modeling and
solution. A close analysis of the state of literature will be given. Pathways for
future research will be discussed.
2 - The Electric Fleet Size and Mix Vehicle Routing Problem with
Time Windows and Recharging Stations
Richard Hartl, Professor, University of Vienna, Oskar-
Morgenstern-Platz 1, Vienna, Austria, ,
Gerhard Hierman, Jakob Puchinger, Stefan Ropke
When routing electrical vehicles, limited battery capacity makes detours to
recharging stations necessary. We introduce the E-FSMVRPTW to model decisions
regarding the fleet composition and the actual routing including the choice of
recharging times and locations. We propose a branch-and-price method as well as
an ALNS with an embedded local search and labelling procedure. The
effectiveness of the proposed approach is shown on a newly created set of
benchmark instances and existing benchmarks.
3 - An Iterated Beam Search Algorithm for the Green Vehicle Routing
Problem in Service Fleets
Mesut Yavuz, Associate Professor, University of Alabama,
Alston Hall Box 870226, Tuscaloosa, AL, 35487,
United States of America,
We present a novel Iterated Beam Search (IBS) solution to the Green Vehicle
Routing Problem (GVRP) in Service Fleets. The problem allows internal and
external refueling and aims to minimize the total travel distance of a
homogeneous fleet. Two mathematical formulations (vehicle flows and set
partitioning) are built and a lower bound is obtained for each. Two heuristics
(savings and insertion) are adopted for upper bounding. We also present
preliminary results of a computational experiment.
4 - Fleet Sizing and Scheduling for Mixed Fleets with Alternative
Fuel Vehicles
Ismail Capar, Texas A&M University, College Station, TX,
United States of America,
We present a mixed-integer formulation for a new type of fleet sizing and
scheduling problem. The formulation consider special need for refueling of
alternative fuel vehicles due to their limited range and/or limited refueling
infrastructure, such as availability of charging stations for electric vehicles. We
provide results of numerical analysis together with managerial insights.
MB69
69-Room 201C, CC
Facility Logistics IV
Sponsor: TSL/Facility Logistics
Sponsored Session
Chair: Debjit Roy, Associate Professor, Indian Institute of Management
Ahmedabad, Vastrapur, Ahmedabad, 380015, India,
1 - Robust Supply Chain Design and Operation under Uncertainty
Deniz Tursun, Postdoctoral Research Associate, University of
Illinois Urbana Champaign, 3308 Sharp Drive, Champaign, IL,
61822, United States of America,
Robust supply chain design and operation under uncertainty problems lead to
confluence of integer and continuous variables, which call for Mixed-Integer
Nonlinear Programming (MINLP) algorithms. We consider a comprehensive
random projection algorithm for a subclass of MINLP’s, where the objective and
constraints are defined by convex functions and integrality restrictions are
imposed on a subset of the decision variables.
2 - Batching Decisions in Stock-to-Picker Order Picking
Debjit Roy, Associate Professor, Indian Institute of Management
Ahmedabad, Vastrapur, Ahmedabad, 380015, India,
Vibhuti Dhingra, Jennifer Pazour
We develop new analytical models to analyze performance of static vs. dynamic
batching policies in stock-to-picker order pick systems. In particular, we analyze
the effect of batch size, variability in item inter-arrival times, and item
commonality on order pick performance.
3 - Branch-and-Price for the Capacitated Mobile Facility
Location Problem
S. Raghavan, Professor, Smith School of Business & Institute for
Systems Research, University of Maryland, College Park, MD,
20742, United States of America, ,
Mustafa Sahin, Sibel Salman
The Capacitated Mobile Facility Location Problem (CMFLP) is a combinatorial
optimization problem with applications in supply chain operations and
distribution of medical services. We propose two Mixed Integer Programming
formulations for the CMFLP and discuss a branch-and-price algorithm for a set
partitioning formulation, where the linear programming relaxation is solved with
a column generation procedure. We demonstrate the quality of the algorithm on
instances adapted from the literature.
4 - Store Fulfillment for Online Orders: A Planning Model in a
Collaborative Store Environment
Ming Ni, SUNY Buffalo, 326 Bell Hall, University at Buffalo,
Amherst, NY, 14260, United States of America,
, Arun Hampapur, Qing He, Xuan Liu
This study on online order fulfillment aims to identify the seasonal planning
dimensions from local retailing outlets perspective. It develops optimization
models and heuristic algorithms which solve order assignment and fleet sizing
problems to construct the supply chain plan. The numerical examples are derived
from same day delivery from a real-world retailer store network.
MB70
70-Room 202A, CC
Joint Session RAS/TSL/AAS: Real-Time Decision
Support Practice
Sponsor: Railway Applications
Sponsored Session
Chair: Ravindra Ahuja, President, Optym, 7600 NW 5th Place,
Gainesville, FL, 32607, United States of America,
1 - Simulation-guided Optimization Algorithms for Real-time
Train Scheduling
Pedram Sahba, Senior Systems Engineer, Optym, 7600 NW 5th
Place, Gainesville, FL, 32607, United States of America,
Ravindra Ahuja, Abbas Bozorgirad
In this presentation, we will describe several algorithms for real-time train
scheduling (also known as meet-pass planning) using simulation, mixed integer
programming and network optimization techniques including their
computational results. These algorithms are in production at a railroad in
Australia and we will give a demonstration of the system using these algorithms.
We will also share our lessons about how these algorithms evolved, what worked,
and what did not work.
MB70