INFORMS 2021 Program Book

INFORMS Anaheim 2021

TD42

3 - Finding the Optimal Screening Policy for Chronic Kidney Disease among Diabetics using a POMDP Framework Chou-Chun Wu, University of Southern California, Los Angeles, CA, 90007-4221, United States, Sze-chuan Suen The US CDC estimates that up to 90% of those with chronic kidney disease are undiagnosed, resulting in 30+ million unmanaged cases. To encourage timely diagnosis of at-risk patients, we develop screening guidelines stratified by age, proteinuria status, and prior test history among diabetics by race and gender. To do this, we adopt a Partially Observed Markov Decision Process (POMDP) framework to identify the optimal action (screen or wait) to take every three months from ages 30-85 that maximizes a patient’s discounted lifetime net monetary benefit. We draw some of our POMDP model inputs from a microsimulation which estimates disease progression, lifetime quality-adjusted life years, and medical expenses. We find that the optimal policy recommends more frequent screening in all race and gender groups compared with the annual screening recommended in the status quo. TD40 CC Room 211B In Person: OR/MS in Industry Practice I General Session Chair: Daniela Aguilera, Sr. Manager Inventory Strategy and Optimization, AEO Inc, Pittsburgh, PA, United States 1 - Omni-channel Inventory Placement for Regional Fulfillment Paulie Anne Williams, American Eagle Outfitters, PA, United States As e-commerce business grows at an unprecedented rate, it is increasingly important for retailers to increase delivery speed to customers while minimizing shipping costs by balancing inventory in a multi-node network. At American Eagle Outfitters (AEO), we have found that optimal inventory placement across the network is challenging due to high variability of demand in the fashion industry, which reduces the effectiveness of even state-of-the-art predictive models. We will discuss how we decide assortment and stock levels in each node, as well as how we constantly re-balance inventory to respond to unpredictable demand and ensure regional fulfillment in our network. 2 - Prime Radiant: A System for Evaluating EVTOL Configurations and Vertiport Networks Mike D. Prince, Archer Aviation, Seattle, WA, 76244, United States Archer’s mission is to advance the benefits of sustainable urban air mobility (UAM). Archer is creating the world’s first electric airline that moves people throughout the world’s cities in a quick, safe, sustainable, and cost-effective manner.Prime Radiant is a suite of in-house developed tools used to inform key strategic decisions related to Archer’s business operating model. In this session we will discuss two core optimization models built for this purpose — one for determining vertiport network design in a given city and a fleet routing optimization model used to evaluate vehicle size and configuration. 3 - How Inventory Segmentation and Being Agile Adjusting Inventory In 2020, American Eagle Outfitters (AEO) initiated an Inventory Productivity program. One of the pillars involved analyzing Inventory segmentation and evaluating legacy Inventory policies. The initiative shows how applying universal Inventory principles can be easily suited to any environment and help your supply chain organization remain adaptable and increasing working capital efficiency. Implementation challenges, change management, market changes, data issues as part of the roadmap. *How implementing Inventory Segmentation can help to determine proper Inventory policies *Using data to leverage Inventory optimization and adjust Inventory policies in a fast-changing RTL and E-comm environment *Monitor your Safety Stock Inputs to ensure proper days of inventory and optimize service levels. *Measuring and tracking Inventory productivity Policies Can Leverage Your Supply Chain Performance Daniela Aguilera, Sr. Manager Inventory Strategy and Optimization, AEO Inc, Pittsburgh, PA, United States

TD41 CC Room 212A In Person: Electrical Markets Contributed Session Chair: Santiago Maiz, CIUDAD REAL, 13071, Spain 1 - Computation of Convex Hull Prices using Dantzig-wolfe Decomposition

Panagiotis Andrianesis, Boston University, Brookline, MA, United States, Dimitris Bertsimas, Michael C. Caramanis, William W. Hogan

Several US ISOs have recently considered Extended Locational Marginal Prices as approximation to Convex Hull (CH) prices, mainly because determining exact CH prices is computationally challenging, while providing little intuition about the price formation rationale. We describe the CH price estimation problem by relying on Dantzig-Wolfe decomposition and Column Generation as a tractable, highly parallelizable, and exact method, with finite convergence, which provides intuition on the underlying price formation rationale. We provide several stylized examples and realistic ISO-scale datasets to support scalability and validate proof- of-concept. 2 - Variable Renewable Generation Participation in U.S. Ancillary Services Markets James Hyungkwan Kim, Lawrence Berkeley National Laboratory, Berkeley, CA, United States, Fredrich Kahrl, Andrew Mills Rising penetrations of variable renewable generation (VRG) are reducing VRG value and creating new challenges for system operators. Enabling VRG participation in ancillary services (AS) markets could provide additional revenue and allow system operators to access lower-cost integration solutions. Using profit-maximizing dispatch against 2015-2019 energy and AS prices in all seven U.S. ISOs/RTOs, we found that the average incremental value of AS market participation to hybrid (storage-paired) VRG owners is significantly higher than for standalone VRG. The value to system operators can be high, suggesting the need to consider expanding eligibility to participate in AS markets. 3 - Expansion Planning of a Price-maker Virtual Power Plant in Energy And Reserve Markets Santiago Maiz, Universidad de Castilla-La Mancha, Ciudad Real, Spain, Raquel García-Bertrand, Luis Baringo We address the expansion planning problem of a virtual power plant (VPP) considering the possibility of building new assets such as conventional, renewable, and storage units. The VPP is modeled as a price-maker player that participates in energy and reserve markets altering the prices of these markets to its own benefit. Uncertainties in production levels of renewable units and up/down reserve deployment requests are addressed using a stochastic programming approach. Numerical results show the influence of the behavior of the VPP in the expansion decisions. TD42 CC Room 212B In Person: Energy Policy and Planning Contributed Session Chair: Carlos Olivos, Auburn University, Auburn, AL, 36830, United States 1 - Co2 Infrastructure Planning for Fossil- and Bio-energy with Carbon Capture and Storage Emma JAGU, IFP School, Rueil-Malmaison, France, Olivier Massol BioEnergy with Carbon Capture and Storage (BECCS) is a critical technology to limit global warming. However, its up-scaling requires the installation of a costly CO2 transportation infrastructure, which will likely be shared between BECCS plants and fossil Carbon Capture and Storage (CCS) plants. We examine the conditions for the deployment of such an infrastructure using an adapted cooperative game theoretic framework. We then apply this model to a contemporary project in Sweden. Our results support pragmatic policy recommendations to organize the deployment of the BECCS technology. 2 - Sensitivity Analysis of the Market Penetration in China’s Passenger Vehicle Market Through Monte Carlo Method Mohamed Ali Saafi, Lab Scientist, Aramco Services Company, Novi, MI, United States, Shiqi Ou, Zhenhong Lin, Xin He This study uses the python version of the New Energy and Oil Consumption Credits (NEOCC) model a tool integrated consumer discrete choice and optimization methods to quantify the impact of fuel price, battery cost, markup and fast-charging power on the electric vehicle market success as well as the industry profit in 2020-2050. Through integrating the Monte Carlo Simulation, it tests the robustness of the NEOCC model, and highlights the parameters that could affect the market penetration projection. The results show that markup affects the market the most, while the market becomes more sensitive to the fuel price and battery cost after 2035 which is explained by less policy constraints.

137