Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD73

3 - Entry of Pseudo Sharing Siyi Wang, University of Science & Technology of China, Hefei, China Pseudo sharing—-the phenomenon of professional agents participating in the sharing market—-has caused a huge uproar among the affected traditional markets and governments. It exists in different sharing platforms that involved in different mechanisms. We develop a game-theoretic model to exam the sharing platforms’ interaction with pseudo sharers, as well as the genuine one. Our equilibrium results suggest that, to some extent at least, it enhances the matching rate when the clutter of genuine sharing supply and demand caused most occasional transaction failed. 4 - When Should Retailers Integrate Online and Offline Channels Under Competition? Jinpeng Xu, Xidian University, School of Economics and Management, Xi’an, 710126, China, Yufei Huang This paper considers two multichannel retailers selling a product to consumers with different channel preferences. Retailers can choose to operate their online and offline channels separately, with different prices and unit costs in different channels. They can also choose to integrate the two channels, thereby achieving the same price and unit cost in different channels. We use a game-theoretical model to study when one or two retailers should integrate their channels in competition. 5 - Product Quality Decision and Return Channel Choice with Uncertain Demand Buqing Ma, Doctor Student, University of Science and Technology of China, Baohe District, hefei, Anhui province, China, Hefei, 230035, China, Yunchuan Liu The practice of accepting product returns from consumers directly is understudied in previous literature where consumers return products to manufacturers through retailers. In this paper, we study a manufacturer’s return channel choice and product quality decision with demand uncertainty, and find some interesting relationship between product qualities, return channel choice, and demand uncertainty. n WD73 West Bldg 211B Semiconductor Industry Contributed Session Chair: Maryam Anvar, Applied Materials, San Jose, CA, 95125, United States 1 - Minimizing Robotic Arm Travelling Distance for LED Sorting Problem Jo-Ying Chang, National Chiao Tung University, 1001 University Road, Hsinchu, 300, Taiwan, Lo-Min Su, Sheng-I Chen The LED sorting problem is to determine a sequence to transfer dies from a wafer to bins in different quality levels. The objective is to minimize traveling distance of a robotic arm that is used for classifying LED dies. This problem can be seen as the hierarchical travelling salesman problem. We propose a two-stage algorithm, where the sequence of picking same quality level dies is solved in the first stage, and the shortest distance to connect adjacent tours is determined in the second stage. The results demonstrate the capable of the proposed algorithm to solved real-world cases. 2 - Modeling and Solving LED Dies Picking Problems Sheng I. Chen, National Chiao Tung University, 1001 University Road, HsinChu, 30010, Taiwan, W.L. Pearn, Yen Che Tseng, Wei Ting Hsu We consider LED dies picking problem as a graph representation, where each die refers to a node and the objective is to find a shortest walk covered all nodes in a fully connected graph. To reduce the problem size, we transform adjacent nodes located at the same column as a required arc and each pair of vertexes connected two different columns as a non-required arc. The optimization problem is formulated as a rural postman problem (RPP). We propose an algorithm for solving the RPP. Computational results compare both exact and approximate solutions using general die distributions. 3 - On Mathematical Models of Optimal Video Memory Design Yiwen Xu, Assistant Professor, North Dakota State University, 1410 14th Avenue North, Room 202 Civil & Industrial Engineering, Fargo, ND, 58102, United States The big video data size today imposes huge pressure on storage. Designing memory is a challenging problem due to design constraints, multiple memory bitcell design options, and layout integration under different memory technologies. We develop mathematical models for optimizing embedded video memory design without applying a traditional time-consuming and laborious ASIC design process. The problems are formulated as nonlinear programs and integer linear programs. Different SRAM designs and hybrid SRAM and DRAM designs are considered in our models. The numerical studies show that by applying our proposed method the average mean square error of the video storage can be greatly reduced.

4 - A Branch-and-price Algorithm for Solving Multi-site Production Planning Problem in TFT-LCD Manufacturing Yen-Che Tseng, National Chiao Tung University, Hsinchu, Taiwan, Hsin-Lan Tseng, Sheng-I Chen We study the production planning problem with multiple products and multiple plants for a TFT-LCD manufacturer. A MIP model is formulated to determine production quantity at each plant with the considerations of inventory, capacity, and materials. This problem can be viewed as an extension of generalized assignment problem (GAP), where each customer order corresponds to a task and each factory refers to an agent of a typical GAP problem. A branch-and-price algorithm is implemented for solving large-scale problem instances. We develop a parallel computational framework for solving pricing problems, as well as discover dominance relations between scenarios to enhance runtime. 5 - A Parallel Search Algorithm for Solving LED Sorter Scheduling Problem Wei-Ting Hsu, National Chiao Tung University, Hsinchu, Taiwan, Chang-Feng Li, Sheng-I Chen This study considers the assignment of wafers to identical machines in LED sorting. A mixed integer programming (MIP) model is presented with the objective to minimize the completion time of sorting and merging processes. We reformulate the MIP as upper-bound feasibility problems and each feasibility problem is evaluated using a parallel environment. To enhance run time, we further propose a hybrid method including reducing the problem size by predetermining decision variables, recovering infeasibility, and applying valid inequalities. Finally, a bi-sectional search scheme is adapted for exploring problem instances that exceeded the runtime limit. 6 - Queue Time Simulation Maryam Anvar, Industrial Engineer, Applied Materials, Santa Clara, CA, United States A typical wafer process flow contains up to 1,000 individual steps. Some of these steps have time-sensitive processes which limit the queue time between consecutive operations. Wafers that exceed the time limit will either be scrapped or will experience significant yield loss. In this paper, we use a series of simulation experiments to study the effects of these time limits on wafer scraps and tool capacity to understand and quantify the relationship between tool capacity, lot cycle times through these steps, and the associated rate of time limit violations. n WD77 West Bldg 213A Practice- Modeling for Complex Systems I Contributed Session Chair: George C. Mytalas, CUNY, New York, NY, 11209, United States 1 - Building Total Cost of Ownership Model for Power Battery Consumer Chunyan Duan, Tongji University, Room 1114, Tongji Building A, Siping Road 1500, Shanghai, 200092, China Chunyan Duan, University of Arkansas, Fayetteville, AR, 72701, United States, Jianxin You, Hu-Chen Liu, Luning Shao, Mengmeng Shan Power battery’s economic performance is the main influence on the economic performance of the electric vehicle. We are modeling power battery consumer’s total cost ownership by applying the mathematical modeling method, and exploring the factors of the power battery consumer’s total cost ownership and its relationship between these factors under the fast charging mode and the slow charging mode. 2 - Continuous-time Principal-agent Model in Regulating Disruptive Technology Qi Luo, University of Michigan-Ann Arbor, Ann Arbor, MI, 48109, United States, Romesh Saigal, Zhibin Chen Innovative technologies can achieve a faster penetration and therefore disrupt an existing market. For example, self-driving technology may have massive impact on the transportation network, and consequently the government needs to regulate the market by financial subsidies. We study the decentralized decision- making problem as a principal-agent model in continuous time. The market dynamics is described by the diffusion of innovation and the problem can be converted to solving a Hamilton-Jacobi-Bellman (HJB) equation. Similar applications include the regulations of new drugs, electric vehicles and wireless networks.

518