Informs Annual Meeting 2017

SA62

INFORMS Houston – 2017

SA62

SA63

370C Advanced Maintenance Models Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Yisha Xiang, Lamar University, Beaumont, TX, United States, yxiang@lamar.edu 1 - Bayesian Degradation Based Reliability Growth with Uncertain Effectiveness of Corrections Cesar Ruiz Torres, University of Arkansas, Fayetteville, AR, 72701, United States, caruizto@email.uark.edu, Haitao Liao, Edward A. Pohl Modern products continue to be made increasingly complex meanwhile are still expected to be more reliable. To meet the increasing reliability requirements, reliability growth tests must be conducted in a fast and effective way. We propose a Bayesian reliability growth method to estimate and improve the reliability of a product based on multiple degradation processes of the product. We assume that each degradation process can be modeled by an Inverse-Gaussian process, but the initial degradation and the parameters of corrective actions are unknown. A Copula approach is utilized to model the dependencies between the multiple failure modes. A simple case study is provided to demonstrate the model. 2 - Reliability Analysis of Repairable Systems with Incomplete Failure Time Data Wujun Si, Wayne State University, 4815 Fourth Street, Room 2167, Detroit, MI, 48202, United States, wujun.si@wayne.edu, Qingyu Yang, Leslie Monplaisir, Yong Chen In many real situations, the failure time data of a repairable system may not be fully observed. Most existing reliability studies concerned with missing failure time data assume the system repair is either perfect or minimal. In this paper, we conduct reliability analysis of repairable systems subject to missing failure time data when the repair is general. A maximum likelihood estimation method is developed for the model parameter estimation. Simulation and case studies are conducted to illustrate the developed methods. 3 - Spare Parts Prediction Considering Uncertain Installed Base: A Review and Future Study Tongdan Jin, Texas State University, 601 University Drive, Ingram School of Engineering, San Marcos, TX, 78666, United States, tj17@txstate.edu, Jose F.Espiritu, Heidi Taboada Maintenance and spares inventory model are often studied based on a static installed base. When a new product is released, the installed base is likely to increase due to the market growth. As a result, the demand for spare parts turns to be nonstationary. This talk discusses the recent advances in spare parts provisioning in a non-stationary setting. We introduce superposition renewal theory, review spares forecasting model, characterize lead-time demand, and discuss the future research opportunities. 4 - Condition Based Maintenance for Wind Turbines Considering Complex Conditions Zhigang Will Tian, University of Alberta, 5-8J Mechanical Engineering Building, Department of Mechanical Engineering, Edmonton, AB, T6G 2G8, Canada, ztian@ualberta.ca, Fangfang Ding, Hao Xu Condition based maintenance strategy aims to develop predictive maintenance schedules based on equipment health conditions. Wind turbines are subject to varying load due to changing wind speed and directions. In this study, wind turbine component health conditions are predicted considering complex varying external conditions, and condition based maintenance schedules are developed based on the predicted future health conditions. 5 - Opportunistic Preventive Maintenance Scheduling for a Multi-unit Series System Yisha Xiang, Lamar University, 2626 Cherry Engineering Building, Beaumont, TX, United States, yxiang@lamar.edu, Zhicheng Zhu, Bo Zeng In a complex multi-unit system, opportunistic maintenance is an effective way to reduce system maintenance cost. The center problem of opportunistic maintenance is to decide which component should be maintained in a decision point under the uncertainty of the components’ life time. In this research, we consider a stochastic opportunistic preventive maintenance for a multi-unit series system, and develop a two-stage deterministic equivalent problem for approximation. Numerical example is given to illustrate the proposed model. Our results show that potential saving can be achieved from optimal opportunistic preventive maintenance.

370D State-of-the-Art Renewable Energy Modeling and Analysis Invited: Energy and Climate Invited Session Chair: John Bistline, Electric Power Research Institute, Electric Power Research Institute, Jbistline@epri.com Co-Chair: Delavane Diaz, Electric Power Research Institute, Electric Power Research Institute, Washington, DC, 20005, United States, ddiaz@epri.com 1 - Advanced Methods for Modeling the Integration of Variable Resource Renewable Energy Technologies in the Electric Power Sector Kelly Eurek, National Renewable Energy Laboratory, Golden, CO, United States, Kelly.Eurek@nrel.gov Because renewable power is a rapidly growing component of the electricity system, robust representations of renewable technologies should be included in capacity-expansion models. This is a challenge because modeling the electricity system—and, in particular, modeling renewable integration within that system— is a complex endeavor. This presentation highlights the major challenges of incorporating renewable technologies into capacity-expansion models and shows examples of how NREL’s Regional Energy Deployment System (ReEDS) model address those challenges. 2 - The Economic Geography of Variable Renewable Energy John Bistline, Electric Power Research Institute, Los Angeles, CA, United States, john.bistline@gmail.com Regional differences in electric sector capacity mixes, wind and solar resource potentials, and markets have important implications for resource planning and policy design, especially for variable renewable energy. Despite the proliferation of state-based policies, capacity planning models have only recently incorporated the temporal and spatial resolution necessary to evaluate economic and environmental effects from such regional market linkages. This talk will summarize novel treatments of these areas in EPRI’s US-REGEN energy-economic model and will describe analyses of renewables-focused policies, which underscore the importance of capturing the spatial dependence of impacts. 3 - Representing Energy Storage in Power System Models with Variable Renewable Energy Audun Botterud, MIT, Argonne National Laboratory, Cambridge, MA, 02140, United States, audunb@mit.edu We discuss the representation of energy storage in modeling of power systems and electricity markets with increasing shares of renewable energy. In particular, we focus on modeling of electrochemical batteries with improved representation of degradation, losses, and power limits. We present examples of short- and long- term analysis of energy storage applications in the power grid, including energy arbitrage, unit commitment, and generation expansion planning. 370E Data Mining Best Paper Award I Sponsored: Data Mining Sponsored Session Chair: Tong Wang, University of Iowa, Pappajohn Business Building, 21 East Market Street, Iowa City, IA, 52245, United States, tong-wang@uiowa.edu 1 - Learning Cost-Effective and Interpretable Treatment Regimes Himabindu Lakkaraju, Stanford University, Escondido Village, 119 Quillen Court, Stanford, CA, 94305, United States, himalv@cs.stanford.edu Decision makers, such as doctors, make crucial decisions such as recommending treatments to patients on a daily basis. These decisions typically involve weighing the potential benefits of taking an action against the costs involved. In this work, we aim to automate this task of learning cost-effective, interpretable and actionable treatment regimes from observational data. We formulate this as a problem of learning a decision list - a sequence of if-then-else rules - that maps characteristics of subjects (eg., diagnostic test results of patients) to treatments. We propose a novel objective to construct this decision list which maximizes outcomes for the population, and minimizes overall costs. Since we do not SA64

24