2015 Informs Annual Meeting

SC63

INFORMS Philadelphia – 2015

4 - A New Approach for Optimal Electricity Planning and Dispatching with Hourly Time-scale Air Quality

measure). Clarity of action in a complex decision requires careful selection of distinctions in order to define useful prospects. The author will discuss several real-world applications in oil and gas projects where clear distinctions led to clarity of action. 2 - A Probabilistic Analysis of Drilling Strategies in Unconventionals Robert Hammond, Decision Analyst, Chevron, 1400 Smith St, Houston, TX, United States of America, rhammond@chevron.com This talk will focus on a probabilistic decision analysis of drilling strategies in an unconventional oil and gas play that has sporadic areas with low chances of drilling success. The analysis helped determine the optimal drilling strategy, including whether to drill multiple wells from a single surface location, which reduces development costs and environmental footprint, or a spaced well approach, which in some cases can be used to avoid drilling issues and additional development costs. 3 - Meta-modeling in Decision Analysis: A Case Study Brad Powley, Senior Consultant, Strategic Decisions Group, 745 Emerson Street, Palo Alto, CA, 94301, United States of America, bpowley@sdg.com, Eric Bickel A sophisticated physical model, despite representing an organization’s best thinking, may be excluded from a decision analysis because it cannot complete a requisite number of runs in a reasonable amount of time. When facing such a situation, we created a statistical model of a hydrocarbon reservoir model based on a handful of previous runs, and used it to conduct a probabilistic simulation on project economics. This talk introduces that approach, and discusses its merits and challenges. 4 - A Cognitive Decision Room for High-stakes Decision Making Jeffrey Kephart, IBM, T. J. Watson Research Center, Yorktown Heights, NY, 10598, kephart@us.ibm.com, Debarun Bhattacharjya We have built a cognitive room in which decision makers use a combination of speech and gesture to interact with a multi-agent system of decision and information agents. We overview the hardware and software infrastructure of the cognitive room, describe a set of interacting decision agents, and illustrate via several examples how the room enables human decision makers to make better, more informed decisions in the context of high-stakes decisions in domains such as mergers and acquisitions.

Paul Kerl, Georgia Tech, 755 Ferst Drive, NW, Atlanta, GA, 30332, United States of America, paul.kerl@gmail.com, Valerie Thomas

Energy production from coal, natural gas, oil and biomass generates air pollutants and health impacts. Pollutant exposure depends on the relative location to power plants and atmospheric conditions which vary by hour, day and season. We have developed a method to evaluate pollutant formation from source emissions which we integrate with an electricity production model. In a case study of Georgia we show how to reduce health impacts by shifting production during select hourly periods.

SC63 63-Room 112B, CC

Doing Good with Good OR II Cluster: Doing Good with Good OR Invited Session Chair: Itai Ashlagi, MIT, 100 Main st, Cambridge, Ma, 02139, United States of America, iashlagi@mit.edu Co-Chair: Lisa Maillart, Swanson School of Engineering, Hall Pittsburgh, PA, lisa.maillart@engr.pitt.edu 1 - Finding Patterns with a Rotten Core: Data Mining for Crime Series Detection Tong Wang, Graduate Student, MIT, 70 Pacific Street, Apt. 242A, Cambridge, MA, 02139, United States of America, tongwang@mit.edu We worked with the Cambridge Police Department to build a model that can automatically detect crime series, which analysts spend hours per day doing it manually. NYPD is currently working with our code, aiming to incorporate it into a custom software package they are developing which can assist in their daily job. This project has received widespread media attention. 2 - Infusion Center Process Improvement and Patient Wait Time Reduction Mengnan Shen, Georgia Tech, Atlanta, GA, United States of America, motion0720@gatech.edu, Xiaoyang Li, Allen Liu, James Micali, Jisu Park, Yunjie Sun, Emilie Wurmser, Sung Keun Baek Winship experienced long wait times and low patient satisfaction. Combining data analytics, stakeholder interviews, queuing network principles, and detailed simulation analysis, we improved flow, communication, and visibility throughout the process. Winship implemented our suggestions, resulting in a 28% reduction in patient wait times from check-in to chair, a 8.5% increase in patient satisfaction, and a 6 patients/day increase in throughput. 3 - Using Operations Research to Improve the Health of Patients with Type 2 Diabetes Yuanhui Zhang, NC State, United States of America, yuanhui.zhang@gmail.com We developed OR models for policy evaluation and robust optimization of clinical regimens for glycemic control for patients with type 2 diabetes. We used the models to address controversial questions including: whether protocols based on new medications are more effective than standard regimens. A publication from this work received substantial press and may help inform treatment recommendations in the future.

SC65 65-Room 113B, CC Systems Engineering and Decision Analysis Sponsor: Decision Analysis Sponsored Session

Chair: Robert Bordley, Expert Systems Engr Professional, Booz-Allen- Hamilton, 525 Choice Court, Troy, Mi, 48085, United States of America, Bordley_Robert@bah.com 1 - Limits to Rationality in Systems Engineering George Hazelrigg, Deputy Division Director, National Science

Foundation, Civil, Mech. & Mfg Innovation, 4201 Wilson Boulevard, Arlington, Va, 22230, United States of America, ghazelri@nsf.gov

Rationality is a worthwhile goal in any engineering activity enabling optimization and averting poor choices. But attempts to create a rational framework for systems engineering fail at the time the second person is assigned to the project. We outline the limits to rationality in systems engineering and illustrate consequences. Systems design approaches can have destructive impacts on system design. This paper presents simple procedures to avoid such problems. 2 - Improving Systems Engineering Trade-Off Studies Greg Parnell, Professor, University of Arkansas, Department of Industrial Engineering, Fayetteville, AR, 72701, United States of America, gparnell@uark.edu Today’s complex systems involve significant uncertainties, multiple stakeholders with conflicting objectives, and growing affordability concerns. SE trade-offs arise throughout the system life cycle. Surprisingly many of the published trade-off studies do not have a strong mathematical foundation, do not provide an integrated assessment of value and risk, and many do not even consider uncertainty. We report on a book project to provide best practices using decision analysis. 3 - Making Product Development Decisions with Decision Analysis Dennis Buede, President And Executive Director, Innovative Decisions, 8230 Old Courthouse Road, Suite 460, Vienna, VA, Formal decision processes during system design are commonly called trade studies or analyses of alternatives (AoAs). This paper will give an overview of the process for systems engineering and product development, describe the many kinds of trade studies that should be undertaken, relate decision analysis to these trade studies and discuss complexities of system design about which decision analysts should be aware. Numerous real world examples will be given along the way. 22182, United States of America, dbuede@innovativedecisions.com

SC64 64-Room 113A, CC Joint Session DAS/ENRE: Decision Analysis

Applications in Oil and Gas Sponsor: Decision Analysis & ENRE Sponsored Session

Chair: Brad Powley, Senior Consultant, Strategic Decisions Group, 745 Emerson Street, Palo Alto, CA, 94301, United States of America, bpowley@sdg.com 1 - Defining Prospects for Decision Analysis Ahren Lacy, Decision Analysis Advisor, Chevron, 1400 Smith St, #31-128, Houston, TX, 77007, United States of America, Ahren@chevron.com The prospect is the building block of decision analysis. We express our belief about the likelihood of a prospect’s occurrence (by assigning a probability), and we express our preference should we obtain it (often using a monetary value

114