Informs Annual Meeting 2017

POSTER COMPETITION

INFORMS Houston – 2017

5 - Learning in Discrete Optimization Elias B.Khalil, PhD Student, Georgia Tech, North Avenue, Atlanta, GA, 30332, United States, lyes@gatech.edu The branch-and-bound algorithm for solving Mixed Integer Programs (MIP) involves decisions that are currently addressed heuristically. Instead, we propose to use machine learning (ML) to make better-informed, input-specific decisions in MIP branch-and-bound. For branching and primal heuristic selection, our ML approaches significantly improve solver performance on heterogeneous MIPLIB instances and certain homogeneous families of instances. For some classical combinatorial problems, we propose a deep graph embedding framework for learning powerful greedy heuristics, and show that the learned heuristics are competitive with existing heuristics for VertexCover, MaxCut and TSP. 6 - An Investigation on Aggregate Production Planning with Flexible Requirement Profiles (FRP) with Stochastic Uncertainty Setareh Torabzadeh, The University of North Carolina-Charlotte, 9101 Olmsted Dr Apt6, Charlotte, NC, 28262, United States, storabza@uncc.edu, Ertunga C. Ozelkan This paper investigates production planning problem, using a flexible approach, called: Flexible Requirement Profile, enforcing flexible bounds on production levels in different periods to improve the stability of the production plans. Due to uncertainty of data in the future periods, stochastic programming approach is utilized. We investigate how stochastic FRP aggregate production planning compares with the deterministic FRP aggregate production planning and deterministic and stochastic aggregate production planning. For this aim, we develop chance constraint FRP aggregate production planning model. Results show the FRP models result in better stability and comparable costs. 7 - A Multi-objective Maintenance Strategy through Component Reassignment Jiangbin Zhao, Northwestern Polytechnical University, 554 Mail Box, No.127 Youyi Western Road, Xi’an, 710072, China, zhaojiangbin@mail.nwpu.edu.cn, Shubin Si, Zhiqiang Cai System reliability can be improved by corrective maintenance or pro-active replacement, which could be expensive or uneconomical. Reassigning components is a better method to improve the system reliability for linear- consecutive-k-out-of-n systems. We investigate a multi-objective maintenance strategy by seeking the reassignment with the lowest maintenance cost and highest system reliability. A multi-objective optimization model is formulated. A genetic algorithm based on the maintenance cost and system reliability is developed to search for the Pareto solution. The effectiveness is further compared with the popular non-dominated sorting genetic algorithm. 8 - Condition Monitoring of Vessel’s Main Engine: A Case Study Young-Mok Bae, M.S.Candidate, Pohang University of Science and Technology, Nam-gu Chungam-ro 77, Engineering bld 4, Pohang, 37673, Korea, Republic of, ymbae@postech.ac.kr, Min Jun Kim, Kwang-Jae Kim, Chi-Hyuck Jun, Sang-Su Byeon, Kae-Myoung Park Condition monitoring of a vessel’s main engine is a crucial aspect of maintenance since main engine failures may lead to major accident. Recently the advancement of Information and Communications Technologies(ICT) has allowed massive amounts of vessel’s condition related data to be collected. The data used in this study comes from a main engine in an actual vessel, collected from recent operations. This study reviews various variables related to the condition of the main engine, and detects abnormalities and their trend. This study is expected to provide a basis for preventive maintenance and proactive management of vessel’s main engines. 9 - Minimizing Levelized Cost of Onsite Generation: Towards a Zero Energy Supply Chain An Pham, Texas State University, San Marcos, TX, 78666, United States, a_p276@txstate.edu, Tongdan Jin, Clara Novoa, Cecilia Temponi We formulate a linear optimization model to investigate whether it is feasible to operate a net-zero carbon supply chain via 100 percent of onsite wind and solar generation. The study aims to allocate renewable technology, placement and capacity in a three-tier supply chain network with the goal of minimizing the levelized energy cost. The zero-carbon supply chain model is tested on a network comprised of 20 factories, 2 distribution centers, and 30 retail stores under various wind and weather conditions. We also compare the cost effectiveness of onsite generation assuming net metering and variable feed-in-tariff policies, and interesting managerial insights are derived accordingly.

10 - Modeling and Analysis of Sustainable Supply Chain Dynamics Gang Wang, University of Massachusetts Dartmouth, 285 Old Westport Rd, Room 214, North Dartmouth, MA, 02747, United States, gwang1@umassd.edu, Angappa A.Gunasekaran This paper studies the dynamic behavior of the interaction between sustainable supply chains and the environment. A mathematical model based on the nonlinear dynamic system is presented to describe the dynamics of the impact of supply chains on the environment while achieving sustainable supply chains. To better understand the dynamic mechanism of this proposed system, performance analysis is conducted with respect to three parameters: (a) design production capacity; (b) environmental cost; and (c) demand rate. Analytical results validate the dynamic interaction of supply chains with the environment and justify the environmental and economic significance of supply chain sustainability. 11 - Budgeting Dredging Projects to Improve Resilience of the Inland Waterway Network Khatereh Ahadi, PhD Candidate, University of Arkansas, 4207 Bell Engineering Center, Fayetteville, AR, 72701, United States, kahadi@uark.edu We consider the problem of selecting a budget-limited subset of maintenance actions to maximize the expected tonnage of commodities that can be transported through the inland waterway system. Uncertainty in our model is the budget required at each district for emergency dredging arises from unpredictable conditions. Thus, our main decision is allocating budget to each district to cover the cost of emergency dredging. Then, with the remaining budget routine jobs can be completed. We model this problem as a stochastic programming model, develop solution approaches, and analyze computational results. 12 - Robust Synthetic Control Muhammad J. Amjad, Massachusetts Institute of Technology, 235 Albany Street, Suite 3121A, Cambridge, MA, 02139, United States, mamjad@mit.edu, Devavrat Shah, Dennis Shen We present a robust generalization of synthetic control that features a distinguishing data de-noising process via singular value thresholding. Consequently, our method is robust in multiple facets: it automatically identifies a good subset of donors, functions without extraneous covariates (vital to existing methods), and handles missing data (never been addressed). To our knowledge, we provide the first theoretical finite sample analysis for a more general model than previously considered in literature, and show our estimator is asymptotically consistent. We also introduce a Bayesian framework that allows practitioners to quantitatively measure the uncertainty of their estimates. 13 - Sustainable Electricity Markets for Restructured Electricity Industry Mohammad Rasouli, University of Michigan, 404 Wildwood Avenue, Apt 1, Ann Arbor, MI, 48104, United States, rasouli@umich.edu We consider emerging electricity technological changes and design electricity markets that are reliable, sustainable and price efficient. We start by modeling an oligopoly of electricity generation companies which interact over the electricity network for a long horizon and have uncertainties with respect to the future. Then we propose efficient capacity, carbon permit, transmission right, and electricity spot markets. 14 - Smart Hospitals: Application and Evaluation of ‘Internet of Healthcare Things (IoHT)’ in a Hospital Unit Michael W. Carter, University of Toronto, 5 King’s College Road, Toronto, ON, M5S.3G8, Canada, carter@mie.utoronto.ca, Tahera Yesmin Internet of healthcare things (IoHT) has become an emerging technology which can substantially change the mode of care delivery and therefore improve the care provided. Many researchers have exhibited the working methodology and application of IoHT in various aspects of patient care. However, very few researches have evaluated the outcomes. This research demonstrates the effects of IoHT in one of the hospitals of Ontario, Canada. With the help of various tools of data mining, statistics and industrial engineering this study measures the impact of IoHT. Thus findings of the research indicate the effectiveness of the intervention and hence hold the potential for decision making in care improvement.

373