Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TE05

2 - Impact of Inconvenience and Liquidity Constraints on the Usage of Off-grid Solutions: Evidence from Rwanda Bhavani Shanker Uppari, INSEAD, 1 Ayer Rajah Avenue, Singapore, 138676, Singapore, Serguei Netessine, Ioana Popescu, Rowan Clarke, Manuel Barron, Martine Visser One-fifth of the world’s population does not have access to electricity. Solar-based solutions are unaffordable in these markets due to consumers’ poverty. There are alternative business models relying on rechargeable light bulbs (sold at a subsidized price) which require regular payments for recharges. We investigate the viability of these recharge-based models under poverty. In collaboration with a firm in Rwanda, we collected the bulb usage data from randomized experiments wherein the price and the bulb capacity were varied. We also build a structural model that incorporates the light consumption dynamics, and use it to evaluate theoretically-preferred changes to the existing model. 3 - Monitoring in Public Sector Supply Chains and the Role of Technology Maya Ganesh, Indian School of Business, Hyderabad, India, Sarang Deo, Sripad K. Devalkar We examine the impact of installation of point-of-sale (POS) devices at fair-price- shops (FPS) on leakages. Using a quasi-experimental difference-in-difference approach along with the technique of propensity score matching, we estimate an average monthly reduction of 44 kgs in leakage of rice per FPS. Our results suggest that investing in technology-based monitoring mechanisms can improve performance of public sector supply chains and create economic value even without changing the contracting between channel players. We further conduct a simulation study to estimate the potential value of information of using the real time information provided by POS devices for better decision making. 4 - Rent-to-Own Business Models in Developing Economies: An Empirical Analysis Jose A. Guajardo, University of California-Berkeley, Haas School of Business, 545 Student Services Bldg, Berkeley, CA, 94720-1900, United States Rent-to-Own business models have become popular for the diffusion of energy products in developing economies, as they provide the flexibility to make incremental payments that eventually lead to product ownership by low-income consumers. I empirically analyze consumer behavior in rent-to-own environments in the context of distribution of solar lamps in Sub-Saharan Africa, characterizing different factors that influence operational performance. n TE04 North Bldg 122A Advances in Theory and Practice of Location Science Sponsored: Optimization/Integer and Discrete Optimization Sponsored Session Chair: Manish Bansal, Virginia Tech., Blacksburg, VA, 24060, United States 1 - Locating Drone Platforms for Delivering Medical Supplies with Time Costs Pitu B. Mirchandani, Arizona State University, Schol of Computing, Informatics and, Decision System, Tempe, AZ, 85287, United States, Monica Gentili, Alessandro Agnetis In this paper, we consider a drone delivery problem that could potentially create a significant improvement in terms of cost and time required for providing essential medical supplies. In particular, we consider first the case when we study the problem of sending perishable items in a single day to demand points where cost of delivery depends not only on the time of delivery but also on a soft delivery due date. Other cases are studied where delivery may take place over several days and when platforms can be moved from day to day. 2 - Planar Maximum Coverage Location Problem with Partial Coverage and General Spatial Representation of Demand and Service Zones Manish Bansal, Virginia Tech., 227 Durham Hall, 1145 Perry Street, Blacksburg, VA, 24060, United States We present a new generalization of the planar maximum coverage location problem in which demand and service zones are represented by spatial objects such as circles, polygons, etc., and are allowed to be located anywhere in a continuous plane. We allow partial coverage in its true sense, i.e., covering only part of a demand zone is allowed and the coverage accrued in the objective function as a result of this is proportional to the demand of the covered area only. We present a greedy algorithm and a pseudo-greedy algorithm for it, and showcase that the solution value corresponding to the greedy solution is within a factor of 1- 1/e of the optimal solution value where e is the base of natural logarithm.

3 - Design and Operation of Renewable Hybrid Energy Systems Alexandra M. Newman, Colorado School of Mines, 1500 Illinois

St, Golden, CO, 80401, United States, Alex Zolan, Michael S. Scioletti, Mark Husted, Gavin Goodall

Renewable energy technologies, specifically, solar photovoltaic cells, combined with battery storage and diesel generators, form a hybrid system capable of independently powering remote locations, i.e., those isolated from larger grids. If sized correctly, hybrid systems reduce fuel consumption compared to diesel generator-only alternatives. We present an optimization model for establishing a hybrid power design and dispatch strategy, solution techniques for this mixed- integer, nonlinear model, and insights and extensions.

n TE05 North Bldg 122B

Joint Session OPT/Practice Curated: Recent Developments and Applications using Julia for Optimization Sponsored: Optimization/Computational Optimization and Software Sponsored Session Chair: Joey Huchette, MIT, Cambridge, MA, 02139, United States 1 - Infrastructure Optimization in Julia Carleton Coffrin, Los Alamos National Laboratory, Los Alamos National Laboratory, los Alamos, NM, United States The significant penetration of distributed renewable energy resources and increasingly dynamic fuel prices present significant challenges for the design and operation of critical infrastructure systems. To help explore how optimization algorithms can address these emerging challenges this work presents a collection of Julia packages for infrastructure optimization. These packages provide a variety of tools for, reproducing state-of-the-art results, developing new problem formulations, and benchmarking novel algorithms using standardized data formats and test cases. 2 - Mixed-integer Programming in Julia: From Formulations to Modeling to Algorithms Joey Huchette, MIT, 77 Massachusetts Avenue, Cambridge, MA, 02139, United States In this talk we show how a variety of tools in the Julia programming language can be used throughout the optimization pipeline. First, we present how computational tools can be used to build MIP formulations: by guiding intuition, but also in a completely automated fashion. Next, we present modeling tools for piecewise linear functions that showcase how a high-level interface can help make advanced techniques more accessible to researchers and practitioners. Finally, we show how to use Julia to customize modern MIP solvers to implement advanced and experimental MIP algorithms. 3 - POD.jl: A Global, Mixed-integer Nonlinear Programming Solver in Julia Kaarthik Sundar, Bryan, TX, 77801, United States Non-convex, mixed-integer nonlinear programs (MINLPs) are hard optimization problems to solve to global optimality. State-of-the-art global solvers, such as BARON, Couenne, and SCIP, often handle MINLPs using the spatial branch-and- bound approach in combination with various other tools. However, there has recently been a great deal of interest in MILP-based approaches for solving MINLPs that are based on piecewise convex relaxations. In this work, we present a MILP-based global solver for MINLPs, named POD, where the name represents the three key solution techniques to solve MINLPs: Piecewise convex relaxations (P), Outer-approximation (O), and Dynamic partitioning (D). 4 - Modeling and Solving Mixed-integer Nonlinear Optimization Problems in Julia Christopher D. Coey, Massachusetts Institute of Technology, Cambridge, MA, 02139, United States, Juan Pablo Vielma We use real-world decision problems to illustrate the convenient modeling power of JuMP (github.com/JuliaOpt/JuMP.jl) for mixed-integer nonlinear optimization. We demonstrate how to access a variety of solvers, including our next-generation mixed-integer conic solver Pajarito (github.com/JuliaOpt/Pajarito.jl). Pajarito is written in Julia and uses a primal-dual continuous conic solver (such as Mosek) and a mixed-integer linear solver (such as CPLEX) to perform conic-certificate- based outer approximation. 5 - Mixed-integer Convex Representability Juan Pablo Vielma, Massachusetts Institute of Technology, 77 Massachusetts Avenue, E62-561, Cambridge, MA, 02139, United States, Miles Lubin, Ilias Zadik We consider the question of which nonconvex sets can be represented exactly as the feasible sets of mixed-integer convex optimization problems (MICP). We first show a complete characterization for the case when the number of possible integer assignments is finite. We then further study the characterization for the more general case of unbounded integer variables and introduce a simple necessary condition for representability. We illustrate these characterizations through various examples of sets that can or cannot be modeled as MICP.

381