Informs Annual Meeting 2017

WB77

INFORMS Houston – 2017

WB75

arbitrary mixed-integer linear programming models, and show that the resulting method can be naturally incorporated into existing solvers. Preliminary results suggest that the method can e ectively reduce solution times on hard instances with respect to a state-of-the-art commercial solver. 2 - N-step Cutset Inequalities for Multi-module Capacitated Network Design Problem Haochen Luo, Texas A&M.University, 3131 TAMU, College Station, TX, 77840, United States, hcluo@tamu.edu, Kiavash Kianfar We present our results of studying the multi-module cutset polyhedron, a relaxation of the multi-module capcitated network design (MMND) polyhedron. We give a new class of cutset inequalities for MMND based on the n-step MIR concept, referred to as the n-step cutset inequalities. We show that many cutset- based inequalities introduced in the literature are special cases of our inequalities. In addition, n-step cutset inequalities introduce new classes of strong inequalities for MMND which are facet-defining for the multi-module cutset relaxation of the problem. Our computational experiments demonstrate that our inequalities are very effective in solving MMND instances. 3 - Online Estimation of the Size of the Branch and Bound Tree in MIP Solvers We present an online method that estimates the final size of the branch-and- bound tree in Mixed-Integer Programming solvers. This in turn allows to estimate the total running time. The method combines an old sampling method due to Knuth (1975) and recent work on branching by Le Bodic and Nemhauser (2017). This method is implemented in the MIP solver SCIP and its results are displayed as an extra column. 4 - Facets for Single-module and Multi-module Capacitated Lot-sizing Problems Manish Bansal, Virginia Tech., 227 Durham Hall, 1145 Perry Street, Blacksburg, VA, 24060, United States, bansal@vt.edu We provide sufficient conditions under which the (k,l,S,I) inequalities of Pochet and Wolsey (1993), the mixed (k,l,S,I) inequalities, derived using mixing procedure of Gunluk and Pochet (2001), and the paired (k,l,S,I) inequalities, derived using sequential pairing procedure of Guan et al. (2007), are facet- defining for the single-module capacitated lot-sizing problem without backlogging. We also provide conditions under which the inequalities derived using the sequential pairing and the n-mixing procedure of Sanjeevi and Kianfar (2012) are facet-defining for the multi-module capacitated lot-sizing problem without backlogging. 372F Optimization, Integer Programming Contributed Session Chair: Richard Forrester, Dickinson College, Carlisle, PA, United States, forrestr@dickinson.edu 1 - Nonlinear Model for Failure Cost Assessment of Drone Delivery Network Maryam Torabbeigi, University of Houston, 4655 Wild Indigo St, apt 255, Houston, TX, 77027, United States, mtorabbeigi@uh.edu, Gino J. Lim A delivery network is considered in which some products are delivered to the customers by drones. It is possible that UAV fails during the flight and therefore, some part of the products will be lost. The amount of lost demand depends on UAV failure function and failure location. A new mix integer nonlinear programming (MINLP) model is proposed to minimize the failure cost. The minimum required number of drones is also calculated based on flight time or load capacity or both of them. By solving the model, the path of each UAV, the assigned customers to them, and also the arrival time at each customer and the probability of failure before reaching them are determined while the network failure cost is minimized. 2 - On the Performance of NLP Solvers within Global MNILP Solvers Benjamin Mueller, Zuse Institute Berlin, Takustr 7, Berlin, 14195, Germany, benjamin.mueller@zib.de Solving mixed-integer nonlinear programs (MINLPs) to global optimality efficiently requires fast solvers for continuous sub-problems. These appear in, e.g., primal heuristics, convex relaxations, and optimization-based bound tightening. Two of the best performing algorithms for these sub-problems are Sequential Quadratic Programming (SQP) and Interior Point Methods. In this talk we present the impact of different SQP and Interior Point implementations on the most important MINLP solver components that solve a sequence of similar NLPs. We use the constraint integer programming framework SCIP for our computational studies on instances of the MINLPLib2 instance library. Pierre Le Bodic, Monash University, Clayton, Australia, pierre.lebodic@monash.edu, Gleb Belov, Samuel Esler, Dylan Fernando, George L.Nemhauser WB77

372D Optimization Methodology Contributed Session Chair: Bisheng Du, Ningbo University, Zhejiang, China, dubisheng@nbu.edu.cn 1 - Socially Conscious Product Development: Impact of Costs, Market Growth, and Changes to Brand Image Perception Hannan Sadjady Naeeni, PhD Candidate in Supply Chain Management, University of Houston, Houston, TX, 77096, United States, hsadjady@bauer.uh.edu, Nickolas Freeman, Powell Robinson We propose a game-theoretic duopoly model to identify conditions where one, none, or both of the firms make commitments to socially responsible products at equilibrium. Our analysis shows how the firms’ commitment to socially responsible products depends on the marginal cost differences and degree of market growth, as well as the indirect effect of the product’s social attribute on the consumer’s perception of brand image. 2 - Recovery Optimization Decision-making for Sustainable Manufacturing in the Presence of Demand Uncertainty Kai Meng, Nanjing University of Aeronautics and Astronautics, Nanjing, 210016, China, nuaamk@126.com Kai Meng, Jiangsu Key Laboratory of Precision and Micro- Manufacturing Technology, Nanjing, China, nuaamk@126.com, Peihuang Lou, Peihuang Lou, Xianghui (Richard) Peng, Victor R. Prybutok This research examines recovery optimization in a hybrid recovery system with demand uncertainty. Both dismantling and remanufacturing are addressed. The proposed multi-objective optimization model provides the optimal recovery yield and strategies for a batch of product returns to ensure sustainability. We develop an evolutionary algorithm solution and investigate how demand impacts the recovery decisions. 3 - Retailer Strategies to Encourage Reduced Packaging Adoption: A Threshold Perspective Olga Pak, University of South Carolina, 105 Carolina Ridge Drive, Columbia, SC, 29229, United States, opak@email.sc.edu A majority of CPG manufacturers continue to resist reduction of product packaging because such packaging is associated with several marketing/product positioning risks, like diminished product visibility and lower perceived value. We explore the tools retailers can use to mitigate such concerns, and, thus, convince manufacturers to make their packaging more compact and less wasteful, which has both operational efficiency and sustainability implications. 4 - Flexible Refund on Consumer Return for Customized Products Bisheng Du, Associate Professor, Ningbo University, School of Business, Ningbo University., Fenghua Rd 818, Jiangbei District, Zhejiang, 315211, China, dubisheng@nbu.edu.cn, Guiping Li The mismatch between consumer’s growing demand for customized products and supply side’s increasing leftover inventory, bring many new business models in the real practice, for instance, C2M (Customer to Manufactory). In this study, we analytically study the flexible refund policy instead of full refund (or no refund) for the customer return with respect to their modularity levels. Our intension is to analyze the analytical performance of C2M system and the customer utilities. 372E Theoretical and Computational Advances in Mixed Integer Programming Sponsored: Optimization, Integer and Discrete Optimization Sponsored Session Chair: Manish Bansal, Virginia Tech., Blacksburg, VA, 24060, United States, bansal@vt.edu 1 - Maximizing Reduced-cost-based Fixing Andre Augusto Cire, University of Toronto Scarborough, Department of Management, UTSC, 1265 Military Trail, Toronto, ON, M1C-1A4, Canada, acire@utsc.utoronto.ca, Louis-Martin Rousseau, Omid Sanei Reduced-cost-based fixing is a key computational methodology that cuts out part of the solution space that does not lead to an optimal solution. This technique is, however, dependent on the dual values available at the moment of pruning. We investigate the value of picking a set of dual values which maximizes the amount of fixing that is possible. We test this new variable-fixing methodology for WB76

501