Informs Annual Meeting 2017

SC62

INFORMS Houston – 2017

SC63

4 - Approximate Dynamic Programming-based Policies for Donation Collection and Distribution Robert Alexander Cook, University of Alabama, Tuscaloosa, AL, 35487, United States, Racook1@crimson.ua.edu This study introduces Approximate Dynamic Programming (ADP) techniques for solving a Markov Decision Problem (MDP) in which we distribute stochastically- arriving donations to disaster survivors. Donations accumulate over time at collection sites and are periodically transported to relief centers where they are distributed to beneficiaries. The MDP model minimizes expected unsatisfied demand during a finite horizon. 370C Capacity and Inventory Logistics Sponsored: TSL, Freight Transportation & Logistics Sponsored Session Chair: Satya Sarvani Malladi, Georgia Institute of Technology, Atlanta, GA, 30329, United States, mss@gatech.edu 1 - SKU Rationalization with Substitutable Demands in a Packaged Gas Supply Chain Ethan Malinowski, University at Buffalo (SUNY), 2014 Delaware Avenue, Apt 5, Buffalo, NY, 14216, United States, ejmalino@buffalo.edu, Mark Henry Karwan, Lei Sun, Jose M. Pinto Effective SKU rationalization is advantageous when applied to businesses with high variance and a variety of potentially substitutable product offerings. Advantages include lower production costs, reduced inventory, and system-wide reductions in transportation costs. We develop two MILP formulations for an industrial gas business with over ten thousand SKU’s. One model considers maintaining current customers’ product choices and the other introduces notions of swapping and equivalence among SKU’s. Both models minimize production and transportation costs while simultaneously reducing SKU variance throughout the system. Computational results and policy implications are presented. 2 - Looking Upstream: Optimal Policies for a Class of Capacitated Multi-stage Inventory Systems Alexandar Angelus, The University of Texas at Dallas, Jindal School of Management, 800 West Campbell Road SM.30, Richardson, TX, 75080, United States, alexandar.angelus@gmail.com, Wanshan Zhu We consider a multi-stage, multi-period inventory system with stochastic demand and capacity constraints at each stage. For a class of such systems characterized by having the smallest capacity at the most downstream stage and system utilization above a certain threshold utilization, we identify the structure of the optimal policy. We find the explicit functional form of the order-up-to levels, and show that they depend (only) on upstream echelon inventories. We establish that, above the threshold utilization, this form of the optimal policy achieves the decomposition of the multidimensional objective cost function for the system into single-dimensional convex optimization subproblems. 3 - Exploration of Flexibility Schemes for Time Window Management in E-fulfillment Charlotte Köhler, Viadrina University Frankfurt (Oder), Frankfurt (Oder), Germany, koehler@europa-uni.de, Jan Fabian Ehmke, Ann Melissa Campbell When accepting customer requests, online retailers can build tentative delivery routes and check if each customer can be accommodated feasibly with the remaining delivery capacity. Extending customer acceptance mechanisms from the literature, we consider the remaining logistics capacity, the time of the request relative to the booking horizon, and the customer location in the decision of how many different time windows and what time window widths to offer to the customer. We investigate the impact of these ideas on the number of served customers given the demand structure of order data from an e-grocer in Berlin, Germany. 4 - Dynamic Mobile Capacity Logistics and Multi-location Inventory Control Satya Sarvani Malladi, Georgia Institute of Technology, 2209 Briarcliff Rd NE, Apt 17, Atlanta, GA, 30329, United States, mss@gatech.edu, Alan Erera, Chelsea C. White We investigate the logistics of distributed production-inventory systems that allow mobile production capacity to be shared dynamically. We establish the value addition over fixed systems and present well-performing heuristics to solve the problem efficiently. SC62

370D Energy and Climate 3 Invited: Energy and Climate Invited Session Chair: Yongyang Cai, Ohio State University, Ohio State University, cai.619@osu.edu 1 - Adding Quantity Certainty to a Carbon Tax: the Role of a Tax Adjustment Mechanism for Policy Pre-Commitment Hafstead Marc, Hafstead@rff.org 2 - External Impacts of Local Energy Policy: the Case of Renewable Portfolio Standards Alex Hollingsworth, Indiana University, Bloomington, IN, United States, hollinal@indiana.edu, Ivan Rudik Renewable portfolio standards are state policies that require in state electricity providers to procure a minimum percentage of electricity sales from renewable sources. We show how RPSs induce out of state emissions reductions through state trade of the credits used for RPS compliance. When one state passes an RPS, it increases demand for credits sold by firms in other states. We find evidence that increasing a state’s RPS decreases coal and increases wind generation in outside states. We perform a welfare simulation to evaluate benefits of reductions in local coal fired pollutants, a 1 percentage point increase a state’s RPS results in up to $100 million in benefits towards the United States. 3 - Build Today, Regret Tomorrow? Infrastructure and Climate Policy Yongyang Cai, Ohio State University, Columbus, OH, United States, cai.619@osu.edu, Elizabeth Baldwin, Elizabeth Baldwin, Karlygash Kuralbayeva The timing of optimal policy to combat climate change is controversial. We consider this question in the light of irreversible “dirty” and “clean” investments. This leads to a “Reverse Green Paradox”: the knowledge of an increasing carbon tax will reduce investments in assets that pollute, and so reduce emissions in the short term. This contrasts with the well-known effects of such policy on the suppliers of fossil fuels. So stranded assets play opposing roles, depending on whether these assets supply or demand fossil fuels. Additionally, we show that it is optimal to begin deployment of clean technologies early, if their cost decreases with their cumulative deployment. 4 - Optimal Clean Energy R&D Investments under Uncertainty Giacomo Marangoni, FEEM, Corso Magenta 63, Milano, 20123, Italy, giacomo.marangoni@feem.it, Gauthier de Maere d’Aertrycke, Valentina Bosetti The improvement of clean energy technologies asks for risky efforts in research and development (R&D). When evaluating optimal R&D portfolios, accounting for uncertainty while maintaining the rich set of information produced by complex climate-economy models can lead to intractable problems. We avoid succumbing to computation complexity thanks to Approximate Dynamic Programming. We exemplify the methodology by computing the optimal allocation of near-term innovation investments in four key clean energy technologies. We employ an integrated assessment model (WITCH) with a fairly rich description of the energy-economy nexus, and experts’ beliefs on future costs to quantify uncertainty. 5 - Promoting Clean Technologies Through Consumption and Production Subsidies Soheil Shayegh, Fondazione Eni Enrico Mattei (FEEM), 149 Madison Avenue, Midland Park, NJ, 07432, United States, soheilsh@gmail.com New clean energy technologies typically cost more than established technologies. In order to help clean technologies compete in energy markets, governments can provide two classes of deployment subsidies to different recipients: production subsidies, which lower production costs, or consumption subsidies, which reduce the final price to consumers. Here, we use demand curves to quantify the impacts of these classes of subsidies on technology deployment, social welfare, and market structure. We identify a critical market share in which social welfare is equal under both subsidy schemes. At market shares smaller than the critical value, the benefits of competition favor production subsidies.

90

Made with FlippingBook flipbook maker