Informs 2015

401 / 552

INFORMS Philadelphia – 2015
399
3 - Modeling Processing Times for Based on Expert’s Estimates:
Pert or Triangular? Uniform or Beta?
Martha Centeno, Full Professor, University of Turabo, P.O. Box
3030, Industrial and Management Engineering, Gurabo, 00778-
3030, Puerto Rico, , Kimberly Diaz,
Karla Acevedo
Simulating a system may require relying on the estimates of the experts to obtain
a range and a most likely value. The question is: what is to use: Triangular or
Pert? We present a study of the effect on typical measures of performance of using
Pert or Triangular, Uniform or Beta. Based on results, we present guidelines to
select one of these distributions. For example, Pert and concave Beta should be
used for conservative decision making and triangular and convex Beta for an
optimistic one.
4 - Agent Based Modeling of Uncertain Dynamic Markets
with Contracts
Pratik Misra, Air Products and Chemicals, Inc., 7201 Hamilton
Boulevard, Allentown, PA, 18195, United States of America,
, Sanjay Mehta, Cem Ozen, Yang Liu,
Erdem Arslan
Agent based modeling technique is employed to simulate uncertain markets that
have geographical limits due high distribution costs and have time-bound
contracts. Suppliers and customers are modelled as agents and macro-economic
conditions are modelled as environment in which the agents interact and follow
their programmed decision-rules. In this presentation, we will present general
features of these models and share example case studies to show their utility in
understanding such markets.
5 - An Agent Based Modeling Approach to Predicting Adoption of
Critical Health Practices
Noshir Contractor, Northwestern University, 2145 Sheridan Road,
Tech D241, Evanston, IL, 60208, United States of America,
, Aaron Schecter
The adoption of new health practices in rural areas is driven by a variety of
factors, including opinions, opportunity, and external influences. However, it is
analytically infeasible to determine exactly how opinions spread. Thus, we
propose a series of agent based models to uncover the processes that lead to
widespread adoption, as well as which individuals are most influential. Our
models are based on survey data collected from over 10,000 government agents
in India.
WA77
77-Room 300, CC
Supply Chain Closed Loop I
Contributed Session
Chair: Mohammadsadegh Mobin, PhD Fellow, Western New England
University, 1215 Wilbraham Road,, Springfield, MA, 01119, United
States of America,
1 - Quality Uncertainty and the Value of Coordination in a Closed
Loop Supply Chain
Juan Pedro Sepúlveda-Rojas, Associate Professor, University of
Santiago of Chile, 3769 Ecuador Ave, Santiago, Chile,
We analyze total cost optimization in a SC with returns through global
coordination. We develop a LP model that incorporates a quality factor about
returning items for remanufacturing processes;the model is validated with
different demand and returns scenarios, allowing comparison between instances
with different variability in customer demand.The results highlight the value of
the quantity and quality of returned products for supply chain members and how
it might influence the uncertainty
2 - An Accelerated Benders Decomposition for Closed-Loop Supply
Chain Network Design
Mohammad Jeihoonian, PhD Candidate, Concordia University,
1455 De Maisonneuve Blvd. W., Montreal, QC, H3G 1M8,
Canada, Michel Gendreau,
Masoumeh Kazemi Zanjani
We present a mixed-integer programming formulation to design a closed-loop
supply chain network for modular-structured products. The choice of the
recovery option depends on the quality level of the composing components in the
returned product. We develop an accelerated Benders decomposition-based
solution algorithm. Computational results illustrate the efficiency of the solution
method.
3 - Performance Evaluation of Closed Loop Systems with General
Failure and Repair Times
Mohammadsadegh Mobin, PhD Fellow, Western New England
University, 1215 Wilbraham Road,, Springfield, MA, 01119,
United States of America, Morteza Assadi,
S. Hossein Cheraghi, Zhaojun Li
This paper evaluates the performance of closed loop supply systems using the
proposed modified extended bottleneck algorithm. It is shown that the algorithm
exhibits better performance than the existing bottleneck algorithm for closed loop
systems with generally distributed failure and repair times. The effectiveness of
the developed algorithm is verified using a simulation model.
WA78
78-Room 301, CC
Energy Applications
Contributed Session
Chair: Zhenhong Lin, Oak Ridge National Laboratory, 2360 Cherehala
Blvd, Knoxville, TN, 37932, United States of America,
1 - A Risk-Based Approach to Modeling Industrial Loads in
Non-Residential Buildings
Seyed Vaghefi, Research Associate, Center for Advanced
Infrastructure and Transportation, 100 Brett Rd, Piscataway
Township, NJ, 08854, United States of America,
Mohsen Jafari, Farbod Farzan
This work aims to develop a data-driven framework to predict and optimally
control industrial loads in nonresidential buildings. In this framework, first, a set
of predictive analytics tools are employed to identify the patterns of industrial
loads over time. This includes a high-dimensional cluster analysis and a
classification model to predict the day-ahead load profiles. The results are fed into
a cost-based risk model to calculate and evaluate the total risk of energy decisions.
2 - An Iterative Two-stage Convex Relaxation Approach for Natural
Gas Pipeline Transmission: A CNPC Case
Mengying Xue, Tsinghua University, Department of Industrial
Engineering, Beijing, 100084, China,
Dingzhi Liu, Tianhu Deng
The optimal dynamic planning of natural gas consumption is important to a
nation’s economic sustainability and environmental protection. We study how
China National Petroleum Corporation, the largest oil and natural gas producer
and supplier in China, should dynamically plan its gas production, transportation
and sales amount under certain nonlinear physical requirements. The designed
system has been implement and used and is projected to save $34Billion from
2016-2020.
3 - Natural Gas Storage Valuation under Uncertainty
Ebisa Wollega, Assistant Professor, Colorado State University-
Pueblo, 2200 Bonforte Blvd, Pueblo, CO, 81001, United States of
America, , Hank Grant
This paper presents a heuristic algorithm that natural gas storage decision makers
can use to make storage decisions under uncertainty. The algorithm decreases the
computation time significantly from hundreds of days to fractions of a second at a
reasonable solution quality.
4 - Scenario Generation via Copula-arima Models: Risk Management
for a Gas-fired Power Plant
Xiaojia Guo, University College London, Dept. of Managment and
Innovation, Gower Street, London, WC1E 6BT, United Kingdom,
, Afzal Siddiqui, Giampiero Marra
Gas-fired power plants face uncertainty in both electricity and natural gas prices,
which tend to be positively correlated. We propose a copula-based approach to
link two independent models and to generate scenarios for solving stochastic
programming problems. We compare our approach with established methods,
e.g., independent ARMA models and transfer functions, in terms of forecasting
performance and providing solutions for the stochastic programming problem.
5 - Optimize Electric Driving Range under Range Uncertainty
Zhenhong Lin, Oak Ridge National Laboratory, 2360 Cherehala
Blvd, Knoxville, TN, 37932, United States of America,
This paper optimizes the driving ranges of battery electric vehicles (BEV) for U.S.
drivers based on driving pattern, household vehicle flexibility, vehicle price, range
anxiety, range uncertainty. Key results are the cumulative share of U.S. BEV
consumers for a given optimal range and the sensitivity of such range
distributions to battery cost, charging infrastructure, and range uncertainty.
WA78