2016 INFORMS Annual Meeting Program

WD64

INFORMS Nashville – 2016

WD62 Cumberland 4- Omni Optimization in Crew Planning and Crew Leave Planning Sponsored: Aviation Applications Sponsored Session

2 - Reliable Sensor Deployment For Object Positioning And Surveillance In A Two Dimensional Space Siyang Xie, U of Illinois at Urbana-Champaign, Urbana, IL, United States, sxie13@illinois.edu, Kun An, Yanfeng Ouyang This paper formulates a mixed-integer non-linear mathematical model for the reliable sensor deployment problem considering site-dependent sensor failures, where at least three sensors are required to work together to locate an object in a two-dimensional plane. The non-linear program is first linearized and customized Lagrangian relaxation algorithm is then developed to effectively solve the model. Numerical examples are presented. 3 - Facility Location Planning Under Uncertainty In Disaster Management Emilia Grass, Institute for Operations Research and Information Systems (ORIS), e.grass@tuhh.de Establishing relief facilities and the pre-positioning of first aid supplies before the occurrence of natural disasters is one of the most important preparation strategies in disaster management. Such location and storage decisions have to be made under a high level of uncertainty since the magnitude, time and location of disasters are not known in advance. In this talk, two-stage stochastic programs are presented which are particularly valuable for these situations. This is due to their ability to model uncertainties and to take into account possible implications of location decisions on relief item distribution in the aftermath of a disaster. 4 - Comparison Of Various Chance Constraints On The Inventory Modulated Capacitated Location Problem Kayse Lee Maass, University of Michigan, Ann Arbor, MI, United States, leekayse@umich.edu, Mark Stephen Daskin, Siqian Shen Using diverse data instances, we compare three approaches to incorporating chance constraints into a stochastic capacitated facility location problem in which processing facilities are able to accept demands in excess of the capacity constraint for short periods of time. Each of the formulations simultaneously assesses a penalty cost for each unit of unprocessed demand and imposes limits on the amount of unprocessed demand allowed. We show how the different approaches affect the optimal solution in terms of the facility locations, demand allocations, amount of unprocessed demand, and overall cost. WD64 Cumberland 6- Omni Theory and Application of the Analytic Network Process Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Orrin Cooper, University of Memphis, Memphis, TN, United States, olcooper@memphis.edu 1 - Improving Coherency In The Anp With A Clustering Algorithm Orrin Cooper, University of Memphis, olcooper@memphis.edu, Idil Yavuz When incoherent priority vectors in an ANP Supermatrix are identified it can be costly to elicit new pairwise comparisons. The proposed method can save decision makers valuable time and effort by using the information and relationships that already exist in the weighted Supermatrix. There is also useful information in the linking estimates that were already calculated and used to measure the coherency of the Supermatrix. A dynamic clustering method is used to automatically identify a cluster of coherent linking estimates from which a new coherent priority vector can be calculated and used to replace an incoherent priority vector. 2 - Being Consistent With Ahp Consistency: Issues And Applications Enrique Mu, Carlow University, emu@carlow.edu Consistency has been a cornerstone of AHP theory and applications. The rule of thumb has been that any pairwise comparison matrix for which the consistency ratio (CR) is less or equal than 0.1 will yield an eigenvector which will diverge very little from the calculated eigenvector for the ideal case of CR=0. However, putting aside the theoretical need for a low inconsistency, we will discuss here what are the potential practical applications of this. Can we use it to screen out of the decision-making process, “irrational” decision-makers? What else can consistency tell us about the decision-maker or the decision-making process? In which situations may consistency take a greater degree of importance?

Chair: Norbert Lingaya, Kronos Incorporated, 3535 Queen Mary Rd, Suite 500, Montreal, QC, H3V 1H8, Canada, nlingaya@kronos.com 1 - Bulk Annual Leave Slot Generator: A Two-phase Approach Luc Charest, Operations Research Specialist, AD OPT, A Kronos Division, 3535 chemin Queen-Mary Ouest, bureau 500, Montréal, QC, H3V 1H8, Canada, luc.charest@kronos.com AD OPT has enriched its Man Power Planning tool, namely “Altitude Insight”, with a new component: the Leave Solver. This new component solves the problem of generating, awarding and assigning Long Service Leave and Bulk Annual Leave slots to crew member. With this presentation, we cover how the two-phase approach of the Leave Slot Generator is effective at generating slots in order to meet the first optimization requirement of leveling the gap related to the operational requirement while attaining the second objective of controlling the distribution and variety of generated slots. 2 - A Satisfaction Distribution Approach For Airline Crew Rostering Problems Babacar Thiongane, AD OPT, A Kronos Division, 3535 Queen-Mary, Bureau 650, Montreal, QC, H3V1H8, Canada, babacar.thiongane@kronos.com The satisfaction distribution problem in airline crew rostering aims to have a fair distribution of the bids satisfaction among crewmembers. We propose a new approach to solve this problem that also provides good quality of global satisfaction. 3 - Integral Column Generation Guy Desaulniers, GERAD, guy.desaulniers@gerad.ca Guy Desaulniers, Polytechnique Montreal & GERAD, Montreal, QC, Canada, guy.desaulniers@gerad.ca We introduce an integral column generation method, which is an adaptation of the integral simplex using decomposition algorithm to the column generation context. The new method finds an improved integer solution at each column generation iteration until reaching an optimal solution. We present results for some crew scheduling problems for which optimal solutions are often obtained in few iterations. 4 - Monthly Crew Pairing With 40 000 Flights Francois Soumis, GERAD, francois.soumis@gerad.ca, François Lessard, Mohammed Saddoune The crew pairing problem is modelled as a set-partitioning problem solved by columns generation. The Dynamic Constraints Aggregation speed-up the master problem and permit to solve a weekly window of 10 000 flights in few hours. The rolling horizon with weekly windows produces solutions improver by up to 5%. WD63 Cumberland 5- Omni Probabilistic Location Models Sponsored: Location Analysis Sponsored Session Chair: Kayse Lee Maass, University of Michigan, 1205 Beal Avenue Ann Arbor, Westland, MI, 48109-2117, United States, Leekayse@umich.edu 1 - Robust Defibrillator Placement Under Cardiac Arrest Location Uncertainty The placement of automated external defibrillators (AED) in public locations allows bystanders of cardiac arrest to administer treatment prior to the arrival of emergency medical responders. A key challenge in AED positioning is that cardiac arrest locations are unknown in advance. We address this by formulating the AED problem as a distributionally robust facility location model with uncertainty in the locations of the demand points. Numerical results demonstrate that hedging against demand location uncertainty has the potential to improve cardiac arrest survival outcomes by mitigating the risk of long response times. Auyon Siddiq, UC Berkeley, Berkeley, CA, United States, auyon@berkeley.edu, Timothy Chan, Zuo-Jun (Max) Shen

475