Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

SC63

Switching Problems Burak Kocuk, Sabanci University, Istanbul, Turkey, Santanu Subhas Dey, Andy Sun

n SC60 West Bldg 102B Joint Session HAS/Practice Curated: Applications of Operations Research to Organ Transplantation Sponsored: Health Applications Sponsored Session Chair: Mariel Sofia Lavieri, University of Michigan, Ann Arbor, MI, 48109-2117, United States Co-Chair: Wesley Marrero, University of Michigan, 1205 Beal Ave, Ann Arbor, MI, 48109, United States 1 - Optimizing Kidney Sharing to Reduce Geographic Disparity Sommer Gentry, US Naval Academy, Mathematics Department, 572-C Holloway Road, Annapolis, MD, 21402, United States, Jack Robinson We use parametric linear programming to investigate tradeoffs in decreasing geographic disparity in kidney transplants. We optimize the sharing of kidney supply between neighboring Donation Service Areas (DSAs) to make the number of kidneys transplanted per incident waitlist candidate similar among DSAs. We adjust the percentage of kidneys recovered in a particular DSA that must be transplanted in that same DSA, to reduce geographic disparity while limiting disruption to existing transplant programs which might otherwise oppose the plan. Surprisingly, even if DSAs retain 90% of their original supply, we can significantly improve the kidney transplant rates in the most underserved areas. 2 - Achieving Fairness in Allocating Uncertain Resources with Applications to Kidney Transplantation Sanjay Mehrotra, Northwestern University, Chaithanya Bandi A novel model for fair allocation of uncertain resources becoming available over time is presented. Demand exceeds supply. Supply distribution is constrained by certain logistical requirements. Our goal is to achieve a fair distribution of supply. Fairness objective is stochastic and fractional. We show that a parametric convex optimization problem can be used to identify an optimal solution. We give parametric tractable re-formulations of our model using polyhedral and conic ambiguity sets, motivated from the concepts of robust queues. Numerical results on the performance of the model using data from the US kidney allocation system, which motivated the problem, will be presented. 3 - Cherrypicking Kidneys and Patients: Incentives in Transplant Centers In 2007 the Centers for Medicare and Medicaid Services implemented a set of regulations for transplant centers called conditions of participation (COP). These rules evaluate transplant centers based on one-year patient and organ survival rates post transplantation. According to the COP, centers performing worse than the COP expectation face losing their certification to transplant. Using actual offer and transplant data, we empirically analyze some potential unintended consequences of these regulations such that more risk averse centers choose healthier patients and higher quality organs to transplant. We also discuss some recommendations to policymakers for improving the COP. 4 - The Health and Economic Impact of Presumed Consent Organ Donation on Patients with End-stage Renal Disease David W. Hutton, University of Michigan, 1015 Fairmount Drive, Ann Arbor, MI, 48105, United States, Huey-Fen Chen, Hayatt Ali, Wesley Javier Marrero Colon, Neehar Parikh, Mariel Sofia Lavieri We evaluated the impact of opt-out organ donation policies using a Markov model of end-stage renal disease patients in the US. We evaluated policy increasing donation rates by 5%, 25%, or 50% versus the status quo. An opt-out, presumed consent policy would save $18.3 billion, and gain 137,029 Quality- Adjusted Life-Years over the lifetime of an open cohort of patients on dialysis and on the waitlist for kidney transplantation. When the presumed consent policy only increased donation by 5%, the cost saved was $4.0 billion, and 29,478 Quality-Adjusted Life-Years were gained. The cost savings were $32.8 billion and 254,515 Quality-Adjusted Life-Years were saved if the policy increased rates by 50%. n SC61 West Bldg 102C Recent Advancement on Computational Methods in Energy Emerging Topic: Energy and Climate Emerging Topic Session Chair: Feng Qiu, Argonne National Laboratory, Lemont, IL, 60439, United States 1 - Node-based Valid Inequalities for Optimal Transmission Mazhar Arikan, Associate Professor, University of Kansas, Lawrence, KS, United States, Baris Ata, Rodney P. Parker

Although the benefits of line switching are evident in terms of reducing operational cost and improving system reliability, finding the optimal power network configuration is a challenging task due to the combinatorial nature of the underlying optimization problem. In this work, we carry out a polyhedral study related to a certain “node-based relaxation of the Optimal Transmission Switching Problem. We construct an integral extended formulation of this substructure and propose a practical approach to obtain valid inequalities. Finally, we present the results of our computational experiments on difficult test cases from the literature. 2 - Alternative Formulation of Transmission Constraints for Unit Commitment Problems Alinson S. Xavier, Argonne National Laboratory, Lemont, IL, United States, Feng Qiu, Santanu Subhas Dey When solving the Security-Constrained Unit Commitment Problem on large-scale transmission networks, one of the most complicating factors is handling transmission and security constraints. The most common formulation used in the industry, based on Injection Shift Factors (ISF), yields very dense constraints, which can cause significant performance issues. In this work, we propose an alternative formulation for transmission constraints, based on network decomposition. Experiments on realistic instances show that this formulation is considerably sparser and can lead to improved running times. 3 - Enumerating Multiple Local Optima for the Optimal Power Flow Problem Bernard Lesieutre, University of Wisconsin Madison, WI, United States, Daniel K. Molzahn, Dan Wu The non-convex AC optimal power flow problem generally admits multiple local minima and it can be beneficial to examine the differences between these solutions. We present a deterministic method to trace between optimal solutions. The method relies on recasting the original equations in quadratic form to a more specific geometrically elliptical form, and then applying the Fritz John conditions for optimality. A path tracing algorithm is recursively applied to locate new critical points from previously known solutions. We present results for several challenging test systems. 4 - Machine Learning for Expediting Security Constraint Unit Commitment Solution Feng Qiu, Argonne National Laboratory, 9700 S. Cass Ave, Building 202, Room C205B, Lemont, IL, 60439, United States, Alinson Xavier, Shabbir Ahmed Security constrained unit commitment (SCUC) is one of the fundamental optimization problems in power system operations and market clearing. The computational performance of SCUC is crucial for system reliability and market efficiency. We propose to use machine learning to expedite the solution process of SCUC using mixed-integer programming solvers. n SC63 West Bldg 103B Joint Session DM/Practice Curated: Data Analytics of Complex Data and its Applications Sponsored: Data Mining Sponsored Session Chair: Wanlu Gu, University of Arizona, Tucson, AZ, United States Co-Chair: Neng Fan, University of Arizona, Tucson, AZ, United States 1 - Parametric Approach for Developing Quick Quotes in Fabrication Ravi Suman, University of Wisconsin-Madison, Madison, WI, 53726, United States, Omar Habib, Ananth Krishnamurthy, Sushanta Sahu We describe a methodology to develop rapid and accurate quotes for fabrication jobs. First, fabrication hours are estimated using parametric equations and part drawings. Next, multistep regression model is used to minimize the input features, speed up the estimation process, and derive accurate estimates of pre- weld and post-weld operations. Finally, model coefficients, exponents, and hyperparameters are set using training data. Validation studies using industry data shows that the methodology reduces quoting times by 50% while providing estimates that are within acceptable industry range.

83