INFORMS 2021 Program Book

INFORMS Anaheim 2021

TE36

2 - A Stochastic Prepositioning Model for Distribution of Disaster Supplies Considering Lateral Transshipment Zhijie Sasha Dong, Texas State University, San Marcos, TX, 78666- 4684, United States, Yusheng Wang, Shaolong Hu This work focuses on addressing uncertainties in disasters when considering lateral transshipment opportunities for pre-positioning relief supplies. To deal with uncertain demands the problem is formulated as a two-stage stochastic programming model, which decides simultaneously on the locations of relief facilities and the allocations of relief supplies to demand nodes. Meanwhile, different damage levels caused by disasters are considered and reflected by a survival rate of usable stocked relief items. Multiple types of supplies with various priorities, values and spaces are explored. A real-world case study based on the Gulf Coast region of the United States is conducted to illustrate the application of the developed model. By comparison with the direct shipment solution, the lateral transshipment solution is demonstrated to be more cost-effective and flexible. The sensitivity analysis of out-of-stock penalty cost and maximum travel distance provides managerial insights for relief agencies. 3 - A Policy and Infrastructure Evaluation Model of Commodity Flows Through Inland Waterway Ports Sanjeev Bhurtyal, University of Arkansas, Fayetteville, AR, United States, Sarah Vavrik Hernandez, Sandra D. Eksioglu, Manzi Yves The purpose of the study is to formulate the two-stage stochastic optimization model to determine which inland waterway ports to invest in to provide expanded commodity-specific handling capacity given changes to commodity- specific demand. Calibration and validation of the two-stage optimization model is carried out in Arkansas River. To overcome the computational burden from large size mixed-integer linear programming model, Benders decomposition algorithm is used. The results from the model serve to prioritize inland waterway port infrastructure and equipment capacity expansion investment decisions under scenarios of commodity growth/decline. 4 - Load Plan Scheduling Problem Mike Hewitt, Loyola University Chicago, Glen Ellyn, IL, 60137- 5246, United States, Fabien Lehuede We introduce a new optimization problem, the Load Plan Scheduling Problem that is relevant to Less-than-truckload freight transportation carriers. This problem seeks to determine a schedule for a given set of shipment paths that minimizes transportation costs by achieving high levels of consolidation. We present different integer programming formulations of this problem. One formulation is based on a time-space network. We illustrate how the size of that network can be reduced through a filtering procedure and propose multiple classes of valid inequalities. Another formulation does not involve a time-space network but instead is an extended formulation, which we propose solving with a branch-and-price solution approach. We report on the effectiveness of these techniques with a computational study.

TE33 CC Room 209A In Person: Online Resource Allocation: New Models and Algorithms General Session Chair: Vineet Goyal, Columbia University, New York, NY, 10027, United States 1 - Online Resource Allocation With Time-flexible Customers Evan Yao, Massachusetts Institute of Technology, Canton, MO, United States We study an online resource allocation model where certain arriving agents are time flexible, meaning that they are willing to wait a short period of time to receive the resource. For flexible agents, we must make an immediate and irrevocable commitment to accept them, but how exactly we allocate resources to satisfy their demand can be made in an offline manner after we have seen more of the input sequence. When there are 2 or 3 types of agents, we present algorithms that achieve the maximum possible competitive ratio, while for 4+ types, we present a simple algorithm which achieves at least 80% of the maximum competitive ratio. 2 - Spatial Elasticity Bobby Nyotta, UCLA Anderson School of Management, 25369 Avenida Ronada, Los Angeles, CA, 91355-3203, United States, Fernanda Bravo, Keith Chen Using transactions data from a popular downtown neighborhood in a large metropolitan city’s mobile phone application for parking payments, we analyze customer behavior from a natural pricing experiment to estimate the “spatial elasticity,” a measure of how individuals quantify the cost of walking an additional mile, in an urban mobility setting.We find that customers require approximately $81 to walk an additional mile to their intended destination. The results are robust against several varying assumptions and when considering factors such as weather and time of day.Our estimates can be used in ride-sharing, bike-sharing, e-scooter-sharing settings to incentive users to end their trips at key locations to either ensure future availability or reduce congestion. 3 - Discrete Choice via Sequential Search Aydin Alptekinoglu, Pennsylvania State University, University Park, PA, 16802-3603, United States, Natalia Kosilova This work considers the sequential search process of the consumer and derives the resulting choice probabilities. While the optimal search strategy was characterized by Weitzman (1979), to the best of our knowledge there is no work deriving the choice probabilities that result from the optimal search strategy. TE35 CC Room 210A In Person: Freight Transportation I General Session Chair: Zhijie Sasha Dong, Texas State University, San Marcos, TX, 78666-4684, United States Co-Chair: Mike Hewitt, Loyola University Chicago, Glen Ellyn, IL, 60137-5246, United States 1 - The Middle Mile Consolidation Network Design Problem with Fixed Origins And Destinations: A Time-constrained Continuous Rate Model Lacy Greening, Georgia Institute of Technology, 878 Peachtree St Ne Apt 716, Atlanta, GA, 30309-4469, United States, Alan Erera The focus of the talk is on continuous rate load planning for large-scale middle mile order fulfillment of time-sensitive bulky items. We will demonstrate how to explicitly incorporate service time requirements within a flat network and how to solve realistically-sized problems using an IP-based local search heuristic.

TE36 CC Room 210B In Person: Analytics in eBusiness General Session

Chair: Bryce Mclaughlin, Palo Alto, CA, 94306, United States 1 - Service Quality and Wage Differentiation in Two-sided Ridesharing Platforms Haozhao Zhang, University of Texas at Dallas, Richardson, TX, 75080-3021, United States, Chenglong Zhang, Srinivasan Raghunathan We examine the quality differentiation strategy for a two-sided platform that matches drivers with riders. The riders have different valuations for waiting time, which is one measure of service quality. Unlike product markets, the ridesharing platform faces a self-scheduled supply and they may also be strategic in accepting a ride request. In the presence of strategic drivers, offering differential wages to control driver supply leads to adverse effects from driver-side cannibalization. An increase in the driver-side cannibalization diminishes the platform’s incentive to practice wage differentiation as well as quality-differentiated services.

147