INFORMS 2021 Program Book

INFORMS Anaheim 2021

POSTER

2 - The Effect of Stockout Based Substitution on Fill Rates Kevin Sweeney, Sam Houston State University, Huntsville, TX, United States, Heidi P. Celebi, Alan Pritchard, Phil Evers In this research we construct a series of decision trees and a simulation model to examine the impact of stockout based customer switching on different fill rate metrics in a continuous review inventory system. Using a variety of target service levels and customer substitution probabilities, we find that the decision trees occasionally misestimate expected fill rates when compared to the simulation model, and that this misestimation is more likely and of larger magnitude in scenarios where both substitute item service levels and customer willingness to switch to a substitute item are high. Implications for managing retail inventory are discussed. 3 - Optimizing Inventory Placement in a Two-echelon Distribution System with Fulfillment-time-dependent Demand Yue Wang, Texas A&M University, College Station, TX, United States, Joseph Geunes, Xiaofeng Nie We study a two-echelon system where a fulfillment center supplies local distribution centers (LDCs) within committed resupply leadtime, and LDCs serve end customers within a committed delivery time (CDT). Expected system demand depends on price, CDT, and the number of LDCs (N), which are determined by a profit maximizing model. We characterize optimal solution structures and consider models for characterizing demand growth as N increases. The results provide insights on how operational scale and constraints influence strategic stock placement and distribution system structure. 4 - Designing Stockless Production Systems in the Presence of Manufacturing Disruptions Antonio Arreola-Risa, Texas A&M University, College Station, TX, United States, Jordi Fortuny-Santos, Carla Vintró-Sánchez In stockless production systems, an item is first demanded and then produced. We study the design of production systems with heterogenous items. Items demand is a Poisson process and unit production times are generally distributed random variables. Production experiences random disruptions of random duration. The objective is minimization of the long-run average holding and back-ordering costs. We establish conditions on the design variables process velocity and variability for which stockless operation is optimal. SC46 CC Room 213D In Person: Health Care, Strategy and Policy Contributed Session Chair: Shubham Akshat, University of Maryland, College Park, MD, 20742-1862, United States 1 - Vancomycin Dosing in Critically Ill Patients: A Machine Learning Approach Mohammad Samie Tootooni, Assistant Professor, Loyola University Chicago, Maywood, IL, United States, Erin Barreto, Kianoush Kashani, Kalyan Pasupathy As a nephrotoxic medication, both sub- and supra-therapeutic vancomycin trough concentrations have consequences. We aimed to identify the key predictive factors for the vancomycin steady-state trough level and their relative contribution and estimate the risk of a steady-state trough outside the goal range. Our models were tested via the left-out set in predicting sub-therapeutic (ROC: 0.85, Specificity: 0.53, and Sensitivity: 0.94) and supra-therapeutic (ROC: 0.83, Specificity: 0.47, and Sensitivity: 0.94) categories, respectively. We also developed an on-demand recommendation engine which offers the optimal dosing regimen for each individual. 2 - Developing a Novel Exact Model of Zoning Optimization for Marine Spatial Planning Mohadese Basirati, IM T. Atlantique, Lab-STICC, UM R. CNR S. 6285, Brest F-29238, France, Brest, France, Patrick Meyer, Romain Billot Marine spatial planning (MSP) as an efficient planning tool simplifies decisions on the sustainable use of marine resources. Determining an optimal zone for one marine user, considering the other users’ activities, represents one challenge in MSP. We propose modeling the problem as an Exact Multi-Objective Integer Linear Program. We developed the raster data model to maximize the interest of the zone dedicated to a single actor and to maximize its spatial compactness. We are studying two approaches for resolution: first, a weighted sum and second, an improved augmented version of the -constraint method, AUGMECON2. We validate the model by performing experiments on artificially generated data.

3 - Does Broader Sharing Improve Patient Outcomes? Analysis of Share 35 Liver Allocation Policy Shubham Akshat, University of Maryland, College Park, MD, United States, Liye Ma, Subramanian Raghavan Broader sharing of organs is believed to mitigate geographic disparity in access to liver transplants. We build a structural model to study the impact of Share 35 policy, a variant of broader sharing introduced in 2013, on behavior changes and on patients’ welfare. We find that Share 35 policy helped in reducing the geographic disparity. The sicker patients benefited from policy and became selective in accepting organs, however there was heterogeneity in behavior change across geographies in lesser sick patients. Collectively, not all geographies benefited from Share 35 policy. We conclude that the current acuity circles policy would result in lower patient welfare than the previous Share 35 policy.

Poster CC – Exhibit Hall B, Foyer In Person Poster Session Poster Session

1 - A Dynamic Programming Model for Joint Optimization of Electric Drayage Trucks Operations and Charging Stations Planning at Ports Xuanke Wu, University of South Carolina, Columbia, SC, United States, Yunteng Zhang, Yuche Chen Port electrification is a promising strategy to achieve sustainability at ports, but its success depends on coordination of infrastructure planning and operation. This paper fills the knowledge gap by proposing a jointly optimization framework to co-optimize infrastructure decisions and operational scheduling to achieve the minimum system cost. The scheduling decision is modeled as a dynamic programming problem with sequential decision-making. We incorporate spatial and temporal heterogeneities of charging and driving costs of different truck trips.We implement our model on an empirical study to fulfill 5% of daily Twenty-foot Equivalent Unit containers in Port of LA and Long Beach. 2 - A Robust Optimization Approach for Robust Explicit Model Predictive Control The adoption of explicit model predictive control (MPC) for a process control application includes uncertainty, which primarily stems from plant-model mismatch. Robust optimization has been utilized to solve such problems. However, an open challenge is their solution for a linear quadratic regulator formulation that avoids dynamic programming. We present an algorithm which reformulates the explicit MPC problem to its robust counterpart. Furthermore, linear transformations are employed to preserve the linearity of the feasible space. Finally, the robust solution of the problem is derived by solving a multiparametric optimization problem and its benefits are exhibited through an example. 3 - Selection, Scheduling of Project Portfolios under Profit Uncertainty and Limited Available Scientists by using Adaptive Robust Optimization Hedieh Ashrafi, Southern Methodist University, Dallas, TX, United States, Aurelie Thiele We present a model for the selection and scheduling of R&D projects with several phases. The initial problem concerns single-phase projects containing development costs and uncertain commercialization profits. The goal of this model is to maximize the net present value under limited scientists’ availability and uncertain profit. Then, we provided the adaptive robust optimization model tackling multi-phase projects in which the new information regarding the previous phases was revealed during the time horizon. We compared the performance of our proposed approach in terms of running time and optimality gap in experiments with static robust optimization benchmarks. 4 - Mothers’ Satisfaction from Childbirth Service in Israel Iris G. Moryossef, Hadassah Academic College, Jerusalem, Israel Keren Orchen Mother’s satisfaction during childbirth influences well being as a mother and relationship with the baby. The study covers more than 300 Ultra-Orthodox Jewish and Non-Religious Jewish Mothers to be in Israel emphasize the importance of Personal Interaction with the mother as a significant factor for childbirth satisfaction for both segments’ satisfaction during childbirth. The more responsiveness, empathy and engaged of the mother during childbirth, her satisfaction and safety increase. The surrounding atmosphere hygiene and aesthetic of the room was significant only for religious mothers. Results justify hospital’s efforts in service provider. Iosif Pappas, Texas A&M University, College Station, TX, United States, Nikolaos A. Diangelakis, Richard Oberdieck, Efstratios Pistikopoulos

29