2016 INFORMS Annual Meeting Program

132 / 561

INFORMS Nashville – 2016
132
3 - Policy Uncertainty And Real Options On Switching Of
Peak Power Plants
Stein-Erik Fleten, Norwegian University of Science and
Technology, , Marius Johansen,
Alois Pichler, Carl J Ullrich
We examine empirically how economic factors, government policy, and strategic
interactions a
ffect managers’ decisions to switch between operating and stand-by states for
peaking electric power generators. We model the switching decisions using a
structural model of a dynamic optimal decision game. We focus on the power
markets in the Northeastern United States, where annual observations of such
decisions are available. The results indicate that regulatory uncertainty
significantly increases firms’ perception of switching costs, and that large power
producers are noticeably more influenced by their economic environment in their
decision-making than small firms.
MA32
203A-MCC
Scheduling V
Contributed Session
Chair: Nermine Harraz, Associate Professor, Alexandria University -
Faculty of Engineering, Alexandria, Egypt,
1 - Two Stage, Single Lot, Lot Streaming Problem For A 1 + M Hybrid
Flow Shop
Sanchit Singh, Virginia Tech, 1206 University Terrace,
Apt F, Blacksburg, VA, 24060, United States,
This paper considers a single lot, lot-streaming problem for a hybrid flow shop
consisting of a single machine at Stage 1 and M identical machines at Stage 2. The
aim is to find optimal number, sizes and allocation of sublots to the machines at
Stage 2 for minimizing make span. An exact algorithm is developed with
polynomial time search to obtain an optimal schedule when the sublot sizes are
continuous. A branch-and-bound based method is also developed for the case of
integer sublot sizes that employs tight bounds. Its performance is tested against
the direct solution of the MIP formulation by CPLEX.
2 - Ideal Schedules
Kangbok Lee, City University of New York, York College, 94-20
Guy R Brewer Boulevard, Queens, NY, 11451, United States,
, Joseph Leung, Michael L Pinedo
An ideal schedule is referred to as a schedule minimizing both maximum and
total completion times simultaneously and an ideal problem is a problem where
every instance of the problem has an ideal schedule. We identify the boundary
between ideal and non-ideal problems and investigate the algorithms for ideal
schedules under different job characteristics such as processing times, release
dates, precedence constraints, and eligibility constraints.
3 - An Exact Model For Solving The Hybrid Flow Shop
Scheduling Problem
Nermine Harraz, Associate Professor, Alexandria University -
Faculty of Engineering, Alexandria, Egypt,
This work presents a model to solve the hybrid Flow Shop Scheduling problem.
The developed Mixed Integer Programming model considers optimizing a dual
criteria objective function to minimize a convex sum of makespan and the
number of tardy jobs under various constraints. The constraints consider
unrelated parallel machines, precedence relations, stage skipping, machine
eligibility, operations lag time, machine release times, and sequence dependent
anticipatory and non-anticipatory setup times. An exact solution is used to solve
the model and to implement it in a real world packaging case. The output shows
that the model is efficient and applicable easily under different conditions.
MA33
203B-MCC
Simulation I
Contributed Session
Chair: Simon Mak, Graduate Student, Georgia Institute of Technology,
North Ave NW, Atlanta, GA, 30332, United States,
1 - Dynamic Simulation Of Marine Port Resiliency
Henry Lester, University of South Alabama, PO Box 8172,
Mobile, AL, 36689, United States, ,
Raymond L Smith
Marine port system resiliency indicates the degree a port can resist and recover
from a potential disruption. Essential measures of this port resiliency include
operational factors subject to excessive burdens. This paper presents an approach
to capturing marine port behaviors with respect to system operational factors. The
procedure utilizes a system dynamic simulation model to analyze port operations
under unpredictable and fluctuating loads to isolate significant critical port
infrastructure components. The resultant critical components will provide a basis
for future public policy decision analysis pertaining to marine port resiliency
subject to extreme events.
2 - An Optimal Stopping Approach To Portfolio Risk Measurement
Kun Zhang, Mr, City University of Hong Kong, Tat Chee Avenue,
Kowloon, Hong Kong, Hong Kong, 000, Hong Kong,
Guangwu Liu, Bingfeng Zhang
Portfolio risk measurement under the nested setting involving a two-level
simulation procedure is a challenging computational problem. To reduce the
computational burden, we propose an optimal stopping approach in which only
one observation is required in the inner-level simulation for each outer-level
scenario. In particular, we show that conditional value-at-risk (CVaR) of a
portfolio can be represented as the solution to an appropriately constructed
optimal stopping problem. Then the least squares method is employed to solve it,
leading to a fast lower bound algorithm for portfolio CVaR. Numerical results
show that this approach works well and produces accurate estimates of portfolio
CVaR.
3 - A Simulation Model For Assessing Mitigation Strategies
Against Cyber Attacks
Soumyo D Moitra, Senior Member of Technical Staff, Software
Engineering Institute, 5000 Forbes Avenue, Pittsburgh, PA, 15213,
United States,
Defending information networks from cyber attacks is a major concern. We
propose a simulation model that helps analyze the risks involved in different
defense strategies. The model considers attack categories, vulnerabilities in target
hosts and the probabilities of a negative impact as a function of attack categories
and host states. The host states are modeled as functions of mitigation strategies.
Thus the effectiveness of alternative mitigation strategies can be estimated. An
illustrative example is presented and the advantages of this approach are
discussed.
4 - Support Points: A New Method For Compacting Distributions
Simon Mak, Graduate Student, Georgia Institute of Technology,
North Ave NW, Atlanta, GA, 30332, United States,
In this presentation, we introduce a new method for compacting a continuous
distribution into a finite set of representative points, which we call support points.
These points not only provide theoretical advantages over both Monte-Carlo
(MC) and quasi Monte-Carlo (qMC) methods in integration, but can also be
computed for any distribution in practice. To generate support points efficiently in
high-dimensions, a blockwise coordinate descent algorithm is proposed which
exploits closed-form majorization-minimization updates. Simulation studies and a
real-world Bayesian example show that support points provide a sizable reduction
in integration error compared to MC and qMC methods.
MA34
204-MCC
Networks in Healthcare
Sponsored: Manufacturing & Service Oper Mgmt, Healthcare
Operations
Sponsored Session
Chair: Itai Gurvich, Kellogg School of Management, Evanston, IL,
United States,
1 - Network Effects In Epidemic Propagation
Kimon Drakopoulos, Massachusetts Institute of Technology,
In this paper we analyze data on the evolution and propagation of influenza
across the United States and discover that compartmental epidemic models
enriched with environment dependent terms have fair prediction accuracy, and
that the effect of inter-state traveling is negligible compared to the effect of intra-
state contacts.
2 - Distance, Quality, Or Relationship? Interhospital Transfer Of Heart
Attack Patients
Lauren X Lu, Associate Professor, University of North Carolina,
Chapel Hill, NC, United States, , Susan F Lu
We empirically investigate the pattern of where heart attack patients are
transferred between hospitals. Our conditional logit analysis shows that the
relationship of being affiliated with the same multihospital system plays a
dominant role in the choice of transfer destinations, compared to distance and
quality. When using 30-day readmission rate to evaluate the health outcome of
transferred patients, we find that relationship-based transfers are associated with
a much higher readmission rate than distance-based and quality-based transfers.
MA32