2016 INFORMS Annual Meeting Program

230 / 561

INFORMS Nashville – 2016
230
MD67
Mockingbird 3- Omni
IEEE T-ASE Invited Session II
General Session
Chair: Jingshan Li, University of Wisconsin - Madison,
1 - Sparse Modeling And Recursive Prediction Of Space-time
Dynamics In Stochastic Sensor Networks
Hui Yang, Penn State University,
Wireless sensor network has emerged as a key technology for monitoring space-
time dynamics of complex systems, e.g., and body area sensor network. However,
sensor failures are not uncommon in traditional sensing systems. As such, we
propose the design of stochastic sensor networks to allow a subset of sensors at
varying locations within the network to transmit dynamic information
intermittently. Experimental results on real-world data and different scenarios of
stochastic sensor networks demonstrated the effectiveness of sparse particle
filtering to support the stochastic design and harness the uncertain information
for modeling space-time dynamics of complex systems.
2 - Integration Of Data Fusion Methodology And Degradation
Modeling Process To Improve Prognostics
Kaibo Liu, UW-Madison, 1513 University Avenue, Madison, WI,
53706, United States, , Shuai Huang
The rapid development of sensing and computing technologies has enabled
multiple sensors embedded in a system to simultaneously monitor the
degradation status of an operation unit. Unlike other existing data fusion
methodologies that treat the fusion procedure and the degradation modeling as
two separate tasks, this paper aims at solving these two challenging problems in a
unified manner. A case study that involves a degradation dataset of an aircraft gas
turbine engine is implemented to numerically evaluate and compare the
prognostic performance of the developed health index with existing literature.
3 - Bottleneck Analysis To Reduce Surgical Flow Disruptions:
Theory And Application
Xiang Zhong, UW-Madison, , Jingshan Li
The work flow of surgical operations in emergency department and operating
rooms can be interrupted due to various disruptions. Reducing such disruptions is
of significant importance to ensure successful operations. In this paper, we
introduce a continuous-time Markov chain model to analyze the disruptions and
their impacts. A continuous improvement method is developed to identify the
bottleneck disruption, so that reducing the interruption time and frequency of the
bottleneck can lead to the largest improvement. An application of the method at
an emergency department of a large academic medical center is presented to
illustrate the effectiveness of the model and the improvement approach.
4 - Integrating Optimal Simulation Budget Allocation And
Genetic Algorithm To Find The Approximate Pareto Patient
Flow Distribution
Zekun Liu, Washington University,
The imbalanced development among different levels of hospitals caused by the
irrational patient flow distribution has become a major social issue in China’s
urban healthcare system. In this work, we propose a methodology framework
integrating discrete-event simulation, multi-objective optimization and multi-
objective optimal computing budget allocation to find the optimal macro-level
patient flow distribution. A case study validating the optimization framework
shows the Pareto optimal patient flow distribution can improve overall system
performances.
MD68
Mockingbird 4- Omni
Quality Engineering Invited Session
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Murat Caner Testik, ., Hacettepe University, Hacettepe
Universitesi, Muhendislik Fakultesi Endustri Muhendisligi Bolumu,
Ankara, 06800, Turkey,
1 - Metamodel Based Method For Optimization Of Multilayer Thin
Film Architecture
Srikant Nekkanty, Intel Corporation, ,
Danel Draguljic, Thomas Santner, Angela Dean, Rajiv Shivpuri
The application of thin, hard coatings is one of the most effective ways to protect
engineering components operating under heavy contact. We describe a
metamodel based approach for improving the performance of a multilayer coating
architecture using finite element models. From finite element models, the stresses
in the coating material were evaluated and used to build the metamodel for
optimizing the multilayer system. The complexity of this engineering application
involved (1) a nonrectangular input region of coating designs and (2) opposing
objectives to be minimized simultaneously.
2 - Measurement Error Of Binary Quality Inspections In Industry
Thomas Akkerhuis, University of Amsterdam,
The evaluation of the reliability of binary measurements, such as quality
inspections in industry, is challenging. This especially holds in gold standard
unavailable situations, where the true conditions of the inspected items are
unobservable. Our studies show that methods in literature for such evaluation
perform poorly. We found that binary measurement error can be decomposed in a
random and a systematic part. Although this is a well-known decomposition for
numerical measurement, it is a new insight for binary measurement. It turns out
that, in a gold standard unavailable situation, generally only the random
component is identifiable. We have developed a robust and efficient method to do
so.
3 - Qfd Customer-requirement Prioritization Based On The Law Of
Comparative Judgments
Domenico Augusto Maisano, Associate Professor, Politecnico di
Torino, Corso Duca degli Abruzzi 24, Turin, 10129, Italy,
, Fiorenzo Franceschini
Quality Function Deployment (QFD) is a structured process to design and develop
products/services that better fulfill customers’ requirements (CRs). The initial
collection and analysis of the CRs is particularly critical, as any distortion can
propagate to the whole process results. The focus of this article is on the
prioritization of CRs, which can be improved by introducing a new prioritization
technique based on the Thurstone’s Law of Comparative Judgment. The greatest
strength of this technique is combining a refined theoretical model with a simple
and user-friendly data collection process. The description is supported by a
realistic application example.
MD69
Old Hickory- Omni
Routing and Allocation in Military Operations
Sponsored: Military Applications
Sponsored Session
Chair: Alexandra M Newman, Colorado School of Mines, 1104 Maple
Street, Golden, CO, 80401, United States,
1 - Minimum Risk Routing Through A Mapped Mine Field
Chris Richards, Colorado School of Mines,
A typical risk-additive routing model in a network or network representation of
continuous space may be unrealistic. We present an alternative ``threat-additive
model’’ formulated as an integer program. We then develop a specialized,
shortest-path approximation to this model. In a realistic model of a ship seeking
to traverse a mapped minefield with minimum risk, we show that the
approximation provides ``true risk’’ while constraining computation time via
implicit enumeration.
2 - Building An Optimization-based, Decision Support System For
Routing Vehicles For Randomized Inspections
Doug Altner, MITRE Corporation, , Justin Nave,
Abby Ng
Suppose we want to maximize the number of 500,000+ sites that 60+ vehicles
can visit subject to constraints on time, area covered, overnight travel, and the
mix of sites selected, while also ensuring selections are “sufficiently random” and
vehicle territories are “well defined”. This talk discusses how we heuristically
solved this problem, and how we turned our methods into an optimization-based,
decision support system for our customer.
3 - Daily Aircraft Routing For Amphibious Ready Groups
Robert Dell, Professor of Operations Research, Naval Postgraduate
School, Monterey, CA, United States, ,
Travis Hartman, Connor McLemore, Ertan Yakici
An Amphibious Ready Group (ARG) consists of ships capable of conducting flight
operations that routinely require the transport of personnel and cargo (PMC) to
remain operationally viable. Planning PMC transport for an ARG and nearby
airfields is a unique vehicle routing problem (VRP) characterized by a
heterogeneous capacitated fleet, two cargo types, multiple depots, disjoint time
windows, and synchronized routing of two aircraft required between some but
not all node pairs. We solve most realistic instances optimally but solution time
can become excessive so we also solve instances using a metaheuristic. This talk
describes this unique VRP, our metaheuristic and our computational experience.
MD67