Informs Annual Meeting 2017

MD44

INFORMS Houston – 2017

3 - Mobile Pharmacies for Reducing Supply Chain Inequities in Low Resource Settings Biplab Bhattacharya, University at Buffalo, 82 Springville Ave, Amherst, NY, 14226, United States, biplabsu@buffalo.edu, Rajan Batta, Li Lin Needs of medical drug supplies are not sufficiently addressed by an equal distribution approach. Equity, different from equality, provides a realistic perspective to identify medical drug demand. A measure is developed that incorporates factors affecting inequity. A mobile-pharmacy location-tour model uses that inequity measure as a criterion to choose communities to serve. The model formulation consists of two parts; the first part involves location selection and frequency determination; the second part is a tour based integer programming formulation. Lastly, a modified Gini coefficient that takes into account demand is used to observe changes in inequity. 4 - Values-based Performance Measures in Humanitarian Relief Michael Veatch, Gordon College, Wenham, MA, 01984, United States, mike.veatch@gordon.edu Models and assessments of humanitarian relief efforts need to incorporate additional factors, beyond quantity or cost of relief delivered, in performance measures. These factors include humanitarian principles, specific values of faith- based organizations, donor priorities, and fundraising considerations. We review some performance measures that have been proposed and report on our efforts to construct empirically-based performance measures for disaster response that incorporate values of the organization. 360C Marketing Contributed Session Chair: Yifei Luo, University of Science and Technology of China, Hefei,Anhui, China, ustclyf@foxmail.com 1 - Buyer-induced and Seller-induced Learning when Product Quality is Uncertain Lu Zhang, School of Management, University of Science and Technology of China, Hefei, China, zlu@mail.ustc.edu.cn, Chuanyong Xu, Yong Zha As product uncertainty can be reduced by consumer learning in different ways, we investigate the impact of learning mechanisms on seller’s pricing decisions. We build a two-period pricing model to explore buyer-induced learning and seller- induced learning. We find that under buyer-induced learning the learning rate target increases with consumer learning efficiency, leading to lower prices. Moreover, under seller-induced learning, the learning rate target is not continuous in seller’s education efficiency, implying a shift in pricing strategy. 2 - Coordination through Cooperative Advertising in a Two Period Consumer Electronics Supply Chain Xiaohang Yue, University of Wisconsin-Milwaukee, Milwaukee, WI, United States, xyue@uwm.edu, Qingyun Xu, Yi He Firms in the consumer electronics industry frequently launch new styles of their products, which leads to a “two-period” phenomenon of their product sales. Only a few published articles have considered two-period models in cooperative advertising. This paper investigates co-op advertising strategies in a two-period supply chain consisting of single manufacturer and single retailer. Utilizing the game theory, we consider two different scenarios: decentralized scenario with a cooperative advertising program and integrated scenario. Aside from these scenarios, we propose a supply chain contract to coordinate this supply chain system. 3 - The Addition of Social Attributes and Product Differentiation Qingyuan Zhu, University of Science and Technology of China, 1952 S.Orchard Street, Apt A, Urbana, IL, 61801-6196, United States, zqyustc@mail.ustc.edu.cn, Jiong Sun, Yunchuan Liu This paper studies the addition of a social attribute onto a product line and its effects on product differentiation decisions. We show that the firm benefits more from adding a social attribute onto a high-quality product than onto a low-quality product, and adding a social attribute results in a non-monotonic impact of quality differentiation on profits and, hence, it alters the firm’s product differentiation decision. MD44

AM hubs which grows to 44 AM hubs as demand increases. It was also found that transportation cost was not a significant factor in the hybrid-AM supply chain. Findings from this study will help both AM companies and traditional manufacturers to determine establishing and growing metal hybrid-AM supply chain. 2 - Random Planar Graphs to Quantify the Evolution of Surface Topography During Polishing Operations Ashif S.Iquebal, Texas A&M.University, 3120 Texas A&M.University, Apt 1002, College Station, TX, 77843, United States, ashif_22@tamu.edu We present a random planar graph representation to model the topography of the surface where the nodes represent the individual asperities, and the edges represent the propensity of the neighboring asperities to form a bridge. We provide theoretical bounds on spectral quantifiers of the graphs by invoking the packing density of hard spheres, that connote the various stages, including the end, of polishing. 3 - Process Control for Quality in Custom Additive Manufacturing While additive manufacturing is a promising approach for customized products, because of variability each product could have different parameters. Thus to avoid defects which further increase the already long production times, we explore adaptive control strategies to set parameters on the fly using sensor-based measurements. To that end, we develop stochastic control methods with the objective of minimizing the expected cycle times. 4 - Multivariate Calibration of Computer Simulation Models for Additive Manufacturing Alaa Elwany, Texas A&M.University, College Station, TX, United States, elwany@tamu.edu, Mohamad Mahmoudi, Gustavo Tapia, Brian Franco, Ji Ma, Ibrahim Karaman, Raymundo Arroyave Computer simulation models are widely used to predict various additive manufacturing (AM) processes. Simulating the melt pool particularly for metal AM processes is the key step for understanding the microstructure and properties of the parts. We present a two-stage Bayesian approach to conduct multivariate calibration of simulation model parameters such that the predictions agree with experimental measurements. First, multivariate Gaussian processes are used to construct a surrogate of the original simulation model. Next, experimental measurements are obtained using a commercial AM machine and titanium alloy powder to compute the posterior distributions of the calibration parameters. 360B Humanitarian Logistics in Public Sector OR Sponsored: Public Sector OR Sponsored Session Chair: Sung Hoon Chung, Binghamton University, Binghamton, NY, 13902, United States, schung@binghamton.edu 1 - An Algorithm for Efficient Evacuation Planning in Congested Multi-class Time-expanded Networks Changhyup Oh, POSTECH, Pohang, 37673, Korea, Republic of, changhyup.oh@postech.ac.kr, Min Hee Kim, Byung-In Kim, Jang Won Choi, Young Myung Ko In this study, we present an efficient heuristic algorithm that generates an evacuation plan close to the optimal one in time-expanded multi-class networks. The proposed algorithm significantly reduces computational time compared with other approaches based on mixed integer programming (MIP) models. We construct several networks from real world data including a multiplex cinema and a subway station as well as a generated network for intuition. Numerical results show that the proposed algorithm produces high quality evacuation plans with short computational time and is scalable in terms of population and network size. 2 - Drone Assisted Last Mile Delivery for Disaster Relief Operations Jinkun Lee, Binghamton University, Binghamton, NY, United States, jinlee@binghamton.edu, Sung Hoon Chung, Duaa Serhan, Sang Won Yoon We consider the use of unmanned aerial vehicles, also known as drones, as an assistant to conventional vehicle routing for disaster relief operations. The advantage can be summarized in two folds: 1) faster delivery of critical supplies and 2) expanded delivery areas. A mixed integer programming problem is formulated and multiple objectives, such as minimizing population-weighted arrival times, minimizing total traveling times, and minimizing the latest arrival time, are considered. As the model gets much more complicated, a hybrid heuristic method, utilizing adaptive neighborhood search, insertion algorithm, and tabu search, is proposed. We present case study examples. MD43 Neil Diaz, Texas A&M.University, College Station, TX, United States, nodiaz@tamu.edu, Jin Xu, Harshit Chawla, Satish Bukkapatnam, Natarajan Gautam

251