2016 INFORMS Annual Meeting Program

SC33

INFORMS Nashville – 2016

3 - Minimizing The Total Number Of Late Multi-task Jobs On Identical Machines Hairong Zhao, Purdue University Calumet, 2200 169 Street, Hammond, IN, 46323, United States, hairong@pnw.edu Lingxiang Li, Haibing Li We consider scheduling multi-task jobs on identical machines in parallel. Each job consists of one or more tasks that can be processed by any machine. The tasks of a job can be processed concurrently. Preemptions are not allowed. Each job has a release date and a due date. The completion time of a job is the time when all of its tasks have been completed. We focus on the problem of minimizing the number of late jobs. We show that while some special cases are solvable, the general problem is NP-hard and admits no constant approximation algorithm unless =NP. We then present a framework of a general algorithm for the problem and derive from it six heuristics whose performance is evaluated by experimental results. 4 - An Improvement In NSGA II For Resource Constrained Project Scheduling Problem Fikri Kucuksayacigil, Iowa State University, 610 Squaw Creek Drive, Unit 18, Ames, IA, 50010, United States, fksayaci@iastate.edu Resource constrained project scheduling problem has been extensively studied. For multi-objective form of this problem, since finding an optimum solution is nearly impossible, several metaheuristic methods have been proposed and implemented. Non-dominated sorting genetic algorithm (NSGA II) has been one of the most effective algorithms in this respect. In this study, we develop a hybrid simulated annealing / NSGA II algorithm to find more diverge and better quality results. The results show that our algorithm visits more solutions in the solution space. 5 - Production Scheduling Of Jobs With Fixed Processing Property On Parallel Machines Sangoh Shim, Hanbat National University, Dept of Business Administration, Deokmyung-Dong, Daejeon, 305-719, Korea, Republic of, mizar0110@gmail.com One of the important things for smart factory is an intelligent production scheduling, how to schedule jobs effectively and efficiently. This problem is for scheduling jobs on parallel machines with the fixed processing property in which a group of specific jobs can be processed on the predetermined machine. Usually, even though parallel machines can process various types of jobs, fixed processing are preferred not to deteriorate products’ quality. Also, in this problem, when changing process of different groups of jobs, operations for changing type of groups, called as setup, are necessary. To minimize makespan of jobs, several heuristic algorithms are devised. SC33 203B-MCC Simulation and Optimization III Contributed Session Chair: Prasanna Kumar Ragavan, Virginia Tech, Durham Hall, Blacksburg, VA, 24061, United States, rpkumar@vt.edu 1 - Managing Escalations: Equipment Failure And Response Capacity Allocation Marc Christiaan Jansen, PhD Candidate, Cambridge Judge Business School, Downing College, Regent Street, Cambridge, CB2 1DQ, United Kingdom, mcj32@cam.ac.uk Nektarios Oraiopoulos, Daniel Ralph Failure of medical equipment represents a cause of downtime for hospitals and may lead to life-threatening circumstances for patients. At the onset of such failure, the scale of the disruption is typically unknown. This paper examines how contracting decisions between a maintenance service provider and multiple clients can enable efficient allocation of response capacity under imperfect and asymmetric information on the true nature of the disruption. 2 - Illusion Of Control In Resource Allocation Decision Making Howard Charles Ralph, Visiting Assistant Professor, Western Carolina University, 201 Edgemont Avenue, Liberty, SC, 29657-1110, United States, r_11l1f@hotmail.com Resource allocation decisions drive the managerial function of control and are basic to business school curricula. Decision problems, deterministic or probabilistic, seek to equip future managers with mathematical tools for optimized solutions, and flexibility to operate under uncertainty. But, “illusion of control” or cognitive biases giving the decision-maker unwarranted confidence, interferes with learning. An exercise has inexperienced decision-makers prioritize a set of realistic allocation problems and explores recorded rationales for features of illusion of control biases.

3 - Flexible High Density Puzzle Storage System Ehsan Shirazi, West Virginia University, 1204, Van Voorhis Road, Unit B, Morgantown, WV, 26505, United States, ehshirazi@mix.wvu.edu A puzzle-based storage system has been introduced to replenish and retrieve items from the top and bottom of a highly dense storage system. Each cell of the puzzle storage is considered as a grid. Each grid is able to store an item and or to move items in the south direction. We describe a high density storage system that can retrieve and replenish items from all sides. A puzzle storage with this characteristic is a lot more flexible than what has been introduced before. We will illustrate how this puzzle storage scheme affects replenish and retrieve time based on different network policies, distributions of replenishing and retrieving items, and number of free spaces on the puzzle network. 4 - The Use Of Simulation For Evaluating Forecast Models Sanjeewa Naranpanawe, Sr Analytical Consultant, SAS Institute, 100 SAS Campus Drive, Cary, NC, 27513, United States, sanjeewa.naranpanawe@sas.com The normal process of evaluating forecast models are by fitting the model using historical data, evaluating using holdout samples to select the model. However, this single point evaluation of forecast accuracy may not be good at predicting how the model is going to perform in the future. This presentation examines how simulation can be use to evaluate different forecast models. 5 - Adaptive-spline For Integer-order Simulation Optimization

Prasanna Kumar Ragavan, Virginia Tech, Durham Hall, Blacksburg, VA, 24061, United States, rpkumar@vt.edu Raghu Pasupathy, Michael Taaffe

We present Adaptive-SPLINE to solve simulation optimization (SO) problems where the decision variables are integer-ordered, and the objective function can only be estimated through “noisy” observations from a simulation. Adaptive- SPLINE iterates between a line search and an enumeration procedure, and adaptively determines sampling effort by trading-off stochastic error with structural error. We will discuss consistency and finite-time performance.

SC34

204-MCC Joint Session HAS/MSOM-HC: Analytics in Drug Development General Session Chair: Elisa Frances Long, UCLA Anderson School of Management, Los Angeles, CA, United States, elisa.long@anderson.ucla.edu 1 - Continuity In Gatekeepers: Quantifying The Impact Of Care Fragmentation Vishal Ahuja, SMU Cox School of Business, vahuja@smu.edu Bradley R Staats, Care coordination is increasingly being recognized as an criticla aspect of overall patient care. We attempt to establish a quantitative measure of care coordination and study its impact on patient health outcomes. Further, we investigate the mechanism by which coordination affects these outcomes. We use data on patients with diabetes, a chronic condition. 2 - Flexible FDA Approval Thresholds: A Dynamic Programming Approach Current FDA approval standards require drug companies to demonstrate the efficacy of their product by presenting statistically significant results from clinical trials. Traditionally, this significance level is set to 0.05 or 0.01, but this choice ignores the complexity of the drug approval process. In particular, the current approval threshold does not incorporate the severity and prevalence of the disease being treated, the level of research and development taking place, and the quantity of existing drugs available for the disease. We develop a continuous time dynamic programming model to study how the optimal significance level should depend on characteristics of the drug pipeline. Taylor Corcoran, University of California-Los Angeles, taylor.corcoran.1@anderson.ucla.edu, Elisa Frances Long, Fernanda Bravo

80