Informs Annual Meeting 2017

TD81

INFORMS Houston – 2017

TD78B

How to defend the world’s cities from natural hazards is the subject of intense debate. Allegedly, some adaptive measures increase vulnerability to unforeseen events. Here, we develop a cellular automaton model to quantitatively examine the effect of alternative adaptive strategies in reducing a community’s risk over time when subject to repeated natural hazards (tsunamis). We find that while hard measures (sea walls) protect against expected events, they increase vulnerability to future, larger events. Actions to avert development from risk zones appear more effective. These results demonstrate hard-adaptive measures can result in maladaptation, thus challenging current practice. 2 - How Accessible Are Our Cities? A Cross-sectional Analysis of Access to Green Space Tim G. Williams, Industrial and Operations Engineering, University of Michigan, Ann Arbor, MI, 48104, United States, tgw@umich.edu, Tom Logan, Connie Zuo, Kevin Liberman, Seth Guikema Adapting our cities requires us to resolve the contentious issue of social justice in service access. Green spaces provide a wide range of benefits, and their access has been subject to much analysis. However, the generalizability of many existing conclusions is limited by both scope and resolution of analysis. Here, we evaluate access to green space in multiple US cities to: a) Investigate whether systemic inequity exists across US cities; and b) Assess the sensitivity of these conclusions to measures of amenity quality. This large-scale, fine-resolution analysis incorporates uncertainty and enables us to draw enhanced insights to answer these questions, essential to designing livable cities. 3 - Using Machine Learning to Predict Building Level Waste Generation and Recycling Rates Constantine Kontokosta, ckontokosta@nyu.edu, Boyeong Hong New York City generates approximately 3.2 million tons of waste each year at a cost of over $1.5 billion. This paper develops a machine learning model to predict building-level waste generation at daily and weekly timescales for over 800,000 residential properties, and uses the results to evaluate the impact of incentive and regulatory measures designed to change household behavior. We use NYC Department of Sanitation daily waste collection data over eight years at the individual truck level, which account for more than 5,000,000 observations, and integrate these with weather, socioeconomic, and land use characteristics. We Shane Jensen, Wharton School, University of Pennsylvania, Philadelphia, PA, United States, stjensen@wharton.upenn.edu Urban data analysis has been recently improved through publicly available high resolution data, allowing us to empirically investigate urban design principles of the past half century. We are currently using safety as a measure of the health of a neighborhood and exploring how it relates to local neighborhood features, including land use zoning, business locations and activity as well as economic measures and population density. I will discuss matching analyses of the relationship between vibrancy and safety and spatial modeling of the change in safety over the past decade. Philadelphia is an interesting case study for this work with its recent population growth and substantial urban development. 382A Dynamic Optimization: Theory and Applications Sponsored: Optimization, Optimization Under Uncertainty Sponsored Session Chair: Vineet Goyal, Columbia University, Columbia University, New York, NY, 10027, United States, vgoyal@ieor.columbia.edu 1 - Improvement of Linear Decision Rules in Robust Optimization by Lifted Uncertainty Sets Frans de Ruiter, Tilburg University, H.R.Holststraat 48, Tilburg, 4103 VB, Netherlands, f.j.c.t.deruiter@uvt.nl We design a new class of nonlinear decision rules for robust optimization and approximate this class of decision rules using lifted uncertainty sets. The new decision rules on the lifted uncertainty set are linear again and strictly improve on the classical linear decision rules and existing lifting methods. We show that the resulting model with our new decision rules on the lifted uncertainty set are not more difficult from a computational standpoint than the models that use classical linear decision rules. The power of our new decision rules is shown on existing applications taken from the literature on adjustable robust optimization. apply multiple approaches to validate our model output. 4 - Urban Analytics: A Case Study in Philadelphia TD81

380B Technology Management Contributed Session Chair: Aditya Kulkarni, Virginia Tech, Blacksburg, VA, United States, aditya88@vt.edu 1 - Generating Large Size Social Networks Xu Dong, Research Assistant, University of Miami, 1251 Memorial Drive, Coral Gables, FL, 33145, United States, x.dong3@umiami.edu, Nazrul I.Shaikh Researchers conduct various tests on simulated social networks, because of the infeasibility of testing certain conditions (e.g. node removal, link adding, etc.) and processes (e.g. spread of diseases, opinions, influence, etc.) that take place in real- world social networks. However, the existing social network models are only able to recover small social networks with few members in it. In this work, we focus on generating large-scale networks with desired social network topologies, such that they can serve as the test beds for modeling specified social processes and activities. 2 - Blockchain for Security of Cyber Physical Systems Leili Soltanisehat, Research Assistant, Old Dominion University, Norfolk, VA, United States, lsolt001@odu.edu In recent decades, along with various cyber attacks to some important settings, security of cyber-physical systems (CPS) has become a challenging issue. In this paper, we are analyzing the implications of Blockchain Technology in boosting the security of CPS. General idea is that CPS will be more secure and resilient if both information and physical security are controlled. Using the concept of cryptography and decentralization, Blockchain technology enables the cyber system with a very high degree of certainty and reliability. Certainty and reliability ensure the integrity and accuracy of information. 3 - Analyzing the Economic and Environmental Impacts of Cloud Computing Chialin Chen, Professor, National Taiwan University, College of Management, National Taiwan University, Taipei, Taiwan, cchen026@ntu.edu.tw, Xiaodan Pan Cloud computing has fundamentally changed the way today’s companies use information technology. We use Systems Dynamics to construct a simulation model to analyze the economic and environmental impacts of cloud computing. On the demand sides, we classify customers in different market segments for cloud computing, and then characterize their specific demand patterns and preferences. On the supply side, we identify different infrastructure and network costs as well as different sources of energy consumption associated with cloud computing. By observing the dynamic systems behaviors, we identify the key determinants of the economic and environmental risks and benefits of cloud computing. 4 - A Decision-theoretic Approach to Verification in Systems Engineering Verification is a key step in the development of a complex system. Multiple decision-making agents across multiple firms are involved in verification activities. A comprehensive and rigorous mathematical framework for computing optimal verification strategies in this multi-agent and multi-firm environment has not yet been developed. We propose a new approach that builds upon principal- agent modeling, partially observable Markov decision processes and multiscale decision theory. Using this new approach, we identify incentive and information exchange mechanisms that allow for better verification at lower cost. 381C Urban Operations: Towards Livable Cities Sponsored: Energy, Natural Res & the Environment, Energy Sponsored Session Chair: Tim Williams, University of Michigan, Ann Arbor, MI, 48104, United States, tgw@umich.edu Co-Chair: Tom M. Logan, University of Michigan, Ann Arbor, MI, 48109, United States, tomlogan@umich.edu 1 - The Role of Behaviour in Undermining the Effectiveness of Adaptive Measures in Reducing Long-term Vulnerability to Natural Hazards Tom M. Logan, Industrial and Operations Engineering, University Aditya Umesh Kulkarni, Virginia Tech, 250 Durham Hall, 1145 Perry Street, Blacksburg, VA, 24061, United States, aditya88@vt.edu, Christian Wernz, Alejandro Salado TD80

of Michigan, Ann Arbor, MI, 48104, United States, tomlogan@umich.edu, Seth Guikema, Jeremy Bricker

405