Informs Annual Meeting 2017

MC68

INFORMS Houston – 2017

MC66

MC67

371A Electric Vehicle Charging Sponsored: Transportation Science & Logistics Sponsored Session Chair: Eleftheria Kontou, National Renewable Energy Laboratory, Golden, CO, United States, ria.kontou@nrel.gov 1 - Data-driven Planning of PHEV Taxi Charging Stations in Urban Environments- A Case in Central Area of Beijing Ziyang Guo, Sparkzone Research Group, Room6, yard 7, Fenzi Hutong, Xidan North, Street, Xicheng District, Beijing, China, Beijing, China, 1872371084@qq.com, Yinghao Jia, Yinghao Jia, Yide Zhao Plug-in hybrid electric taxis (PHETs) contribute to energy and environmental challenges. In this study, we provide a spatial and temporal PHET charging demand forecasting method based on one-month Global Positioning System(GPS)-based taxi travel data in Beijing. Then, a mixed integer linear programming (MILP) model is formulated to plan PHET charging stations which minimizes investment and operation costs of all the PHET charging stations and takes into account the service radius of charging stations, charging demand satisfaction and rational occupation rates of chargers. At last, the plan is carried out numerically through simulation and the analysis is complemented according to the results 3 - Risk-averse Joint Capacity Evaluation of PV Generation and Electric Vehicle Charging Stations in Distribution Networks Yu Xin, Sparkzone Research Group, Beijing, 100084, China, lewis_xy@hotmail.com, Yide Zhao, Yinghao Jia, Yinghao Jia Increasing penetration of distribution generation(DG) and electric vehicles (EVs) calls for an effective way to estimate the achievable capacity connected to the distribution systems, but uncertainties of DG outputs and EV charging loads make it challengeable. This study provides a joint capacity evaluation method for photovoltaic (PV) generation and EV charging stations. The method is formulated as adistributionally robust joint chance constrained programming model. And the worst-case conditional value at risk approximation and an iterative algorithm based on semi definite program (SDP) are used. Finally, the test is carried out numerically on IEEE 33-bus radial distribution system. 4 - Integrated Two-level Inventory Problems: Applications to Electric Vehicle and Drone Battery Management We examine a new class of integrated two-level inventory problems (ITLIP) with applications in drone and electric vehicle battery management. In this class of problems, the first level inventory corresponds to battery capacity and the second- level the battery charge. The two-levels of inventory are integrated in which the first-level is depleted (capacity) as a direct result of the replenishment and satisfaction of demand for the second-level (battery charge, discharge, and use). We formulate a Markov Decision Process model for an ITLIP problem to determine optimal policies for inventory management and equipment replacement for an electric vehicle battery swap station application. 6 - Incentive Schemes for Maximizing the Benefits from Battery Electric Vehicle Adoption Eleftheria Kontou, Postdoctoral Researcher, National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO, 80401, United States, ria.kontou@nrel.gov, Yafeng Yin Battery electric vehicles (BEVs) promise elimination of tank-to-wheel emissions and, thus, are widely recognized as a means towards reaching greenhouse gas abatement goals from the passenger transportation sector. Such vehicles are promoted by local and national governments by allocating financial incentives to directly subsidize their sales, or through funding charging stations. This work aims to answer the following research question: what is the optimal magnitude and timeframe of BEV subsidies and chargers investment allocation that the government should plan for to maximize the monetized social benefits from the household vehicle electrification? Amin Asadi, University of Arkansas, Fayetteville, AR, United States, asadi@email.uark.edu, Sarah G.Nurre

371B Technometrics Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Daniel Apley, Northwestern University, Evanston, IL, 60208- 3119, United States, apley@northwestern.edu 1 - Modelling Regression Quantile Processes using Monotone B-splines Yuan Yuan, Research Scientist, IBM.Research Singapore, Singapore, Singapore, idayuan@sg.ibm.com, Nan Chen, Shiyu Zhou Quantile regression studies the covariate effects on the entire response distribution by fitting quantile regression models at multiple different quantiles or even fitting the entire regression quantile process. However, estimating the regression quantile process is inherently difficult because the induced conditional quantile function needs to be monotone at all covariate values. We proposed a regression quantile process estimation method based on monotone B-splines, which can easily ensure the validity of the regression quantile process, and offers a concise framework for variable selection and adaptive complexity control. We demonstrated its use and effectiveness in real problems. 2 - Pairwise Estimation of Multivariate Gaussian Process Models with Replicated Observations: Application to Multivariate Profile Monitoring Qiang Zhou, University of Arizona, Tucson, AZ, United States, q.zhou@arizona.edu, Yongxiang Li, Li Zeng Existing profile monitoring methods almost exclusively deal with univariate profiles, but multivariate profiles are increasingly encountered. In this paper, we advocate the use of multivariate Gaussian process model for such data, which offers a natural way to accommodate both within-profile and between-profile correlations. To mitigate the high computation, a pairwise estimation strategy is adopted and its asymptotic properties are investigated. 3 - Additive Gaussian Process for Computer Models with Qualitative and Quantitative Factors Xinwei Deng, Department of Statistics, Virginia Tech, 406 Hutcheson Hall, Blacksburg, VA, 24061, United States, xdeng@vt.edu, Devon Lin, K-W. Liu, R. K. Rowe Computer experiments with qualitative and quantitative factors occur frequently in various applications. In this work, we propose an additive Gaussian process model for computer experiments with qualitative and quantitative factors. The proposed method considers an additive correlation structure for qualitative factors, and assumes that the correlation function for each qualitative factor and the correlation function of quantitative factors are multiplicative. The merits of the proposed method are illustrated by several numerical examples and a real data application. 371C Knowledge-driven Analytics for Data-rich Environments Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Wenmeng Tian, Virginia Tech, Blacksburg, VA, 24060-6507, United States, tian0414@vt.edu 1 - Physics-based Attacks Detection in Machining Cyber-physical Systems using In-situ Sensors as Side Channels Mohammed Shafae, Virginia Tech, 1145 Perry Street, Virginia Te, Durham Hall, Room 250, Blacksburg, VA, 24061, United States, shafae1@vt.edu Physics-Based Attacks Detection in Machining Cyber-Physical Systems Using In- Situ Sensors as Side Channels 2 - Knowledge Based Data Monitoring in Advanced and Additive Manufacturing Bianca Maria Colosimo, Politecnico di Milano, Via La Masa, 1, Milan, I-20156, Italy, biancamaria.colosimo@polimi.it The contribution describes different applications where knowledge of the manufacturing signature can aid quality optimization, monitoring and control. Examples in foam production and additive manufacturing are presented. MC68

227