Informs Annual Meeting 2017

MB65

INFORMS Houston – 2017

MB63

2 - B2B Marketing Campaign Recommendation: A Social and Temporal Perspective Jingyuan Yang, Rutgers University, Newark, NJ, United States, jingyyan@rutgers.edu The buying processes of Business to Business (B2B) customers involve series of various marketing campaigns providing multi-faceted product/service information. To better guide B2B customers through dynamic and complex buying processes, we propose a unified view of social and temporal modeling for B2B marketing campaign recommendation. Specifically, we first construct temporal graph as the knowledge representation of each B2B customer’s buying process. The recommender system is then built in a low-rank graph reconstruction framework using probabilistic graphical models. We also incorporate the business community relationship as the social factor to further improve overall performance. 3 - Diagnostics and Treatment of Chronic Disease with Artificial Intelligence Bo Jin, Dalian University of Technology, Dailian, China, jinbo.dlut@gmail.com Parkinson’s disease (PD) is a chronic disease that develops over years and varies dramatically in its clinical manifestations and overall prognosis. It is challenging to compute similarity due to patient data collected from longitudinal clinical trials often consists of multiple modalities and presents temporal dynamics. To provide scalable solutions that can accurately identify similarity from the complex patient data, we propose a deep model that directly learns patient similarity from longitudinal and multi-modal patient data with an RNN architecture that can encode the similarity between two sequences and dynamically match temporal patterns inpatient data. Chair: Mesut Yavuz, University of Alabama, myavuz@cba.ua.edu 1 - Managing Electric Vehicle Charging Station Congestion under Uncertainty MD Abdul Quddus, Mississippi State University, 319 North Jackson Street, Apt 3E, Starkville, MS, 39759, United States, mq90@msstate.edu, Mohammad Marufuzzaman, John Usher, Mesut Yavuz This study incorporates a non-linear cost term in the objective function to capture the congestion effect at charging stations. We propose a hybrid decomposition algorithm that combines Constraint Generation algorithm with a Sample Average Approximation with an enhanced Progressive Hedging algorithm. A case study is conducted by considering road network of Washington, D.C. area. The results of the analysis indicate that incorporating congestion factor encourage larger capacity charging stations opening decisions, increase the number of batteries storing decisions, and higher congestion cost decreases the number of charging stations opening decisions. 2 - Green Location-routing Problem This talk will introduce the Green Location-Routing Problem, an extension of the classical Location-Routing Problem that explicitly accounts for fuel consumption and emissions, the amount of which is measured by a widely used comprehensive modal emission model. The objective is to minimize a total cost function comprising depot, fuel and emission costs. The talk will present a mixed integer programming formulation of the problem with a set of valid inequalities, and describe solution approaches. Computational results on realistic data sets will be presented. 3 - An Integrated Fleet Management Model Introducing Alternative Fuel Trucks into Existing Diesel Long-haul Fleets Ilke Bakir, Georgia Institute of Technology, H. Milton Stewart School of ISyE, Atlanta, GA, 30332, United States, ilkebakir@gatech.edu, Alan Erera We address the challenge of smoothly introducing alternative fuel trucks (AFTs) into an existing long-haul trucking fleet while making necessary structural changes and maintaining feasible operations during the transition. We develop an integrated fleet management model, which incorporates the decisions for (i) opening new maintenance/fueling facilities, and (ii) assigning trucks to travel routes for ensuring demand satisfaction, into a fleet replacement framework. Additionally, in order to obtain optimal (or good feasible) solutions for realistically sized problem instances, we propose a Benders’ decomposition framework and a local search based heuristic. Okan Dukkanci, Bilkent University, Ankara, Turkey, okan.dukkanci@bilkent.edu.tr, Bahar Kara, Tolga Bektas MB65 370F Green Vehicle Routing Sponsored: TSL, Freight Transportation & Logistics Sponsored Session

370D Energy and Climate VI Invited: Energy and Climate Invited Session Chair: Murat Erkoc, University of Miami, Hialeah, FL, 33015, United States, merkoc@miami.edu 1 - A Multi-objective Energy Management Model for Operation Planning in Smart Grids The efficient utilization of distributed generation resources and demand side management (DSM) play a crucial role in satisfying and controlling electricity demand in smart grids. DSM programs make the operation planning problem more difficult with the additional objective of customer stakeholder. Here, we model the operation planning problem considering three conflicting objectives, minimizing the total operation cost and GHG emission, and maximizing the customers’ satisfaction. The multi-objective model is implemented on a generic smart grid and solved using augmented epsilon contraint method. 2 - A Multi-objective Mathematical Model for Integrated Electricity and Natural Gas Pipeline Network Problems Ahmet Akgun, Wichita State University, Wichita, KS, United States, axakgun@shockers.wichita.edu, Mehmet Bayram Yildirim A multi-period multi-objective mathematical model for integration of electricity expansion planning and natural gas pipeline network problem is proposed. The proposed model is formulated as MINLP and it optimizes the costs, greenhouse gas emissions and foreign dependency for both electricity and natural gas pipeline networks simultaneously. This paper aims to determine what kind of generation units to be built, where and when to build generation units. The proposed model is tested on a 22-node electricity and 24-node Belgium natural gas network to determine the optimum investment strategy. 3 - A Zero Carbon Multi-site Production Model under Stochastic Demand and Intermittent Power An Pham, Graduate Research Assistant, Texas State University, San Marcos, TX, United States, a_p276@txstate.edu, Tongdan Jin, Jin Qin, Fei Sun We investigate whether is it economically viable and technically feasible to operate zero carbon manufacturing, transportation and warehouse using onsite wind and solar power. A big data analytics method is developed to characterize the renewables yield based on 11-year meteorological data. We show that zero carbon is achievable if the wind capacity factor is 0.3 or the photovoltaic capacity factor reaches 0.4. Our study facilitates manufacturing industries to become a “prosumser” who actively participate in demand responses as a green energy producer under smart grid paradigm. 4 - Load Shifting and Energy Storage in the Smart Grid Murat Erkoc, Associate Professor, University of Miami, 1251 Memorial Drive, Coral Gables, FL, 33146, United States, merkoc@miami.edu, Eyad M.Al-Ahmadi We study load-shifting and energy storage in an electricity market composed of a single energy provider and multiple customers via a Stackelberg game setting. The provider acts as the leader and decides on price discounts across a finite time horizon. The consumers follow and decide if and how they shift their consumption from their nominal demand. In terms of energy storage, we consider and compare two cases. In the first case, the consumers make the energy storage decisions. In the second, the energy provider manages the client storage devices at consumer sites. We investigate the joint impact of price discounts and storage option on player incentives and peak-to-average ratios. Haluk Damgacioglu, PhD Candidate, University of Miami, 1251 Memorial Drive McArthur Engr. Bldg. Room 280, Coral Gables, FL, 33156, United States, haluk.damgacioglu@miami.edu, Nurcin Celik

MB64

370E Big Data Analytics Sponsored: Data Mining Sponsored Session

Chair: Wenjun Zhou, University of Tennessee Knoxville, 916 Volunteer Boulevard, Knoxville, TN, 37996, United States, wzhou4@utk.edu 1 - Crowdsourcing Xi Zhang, jackyzhang@tju.edu.cn Abstract not Avialable

187