|
Case study - The Advanced Refinery
Modeling System
David S. Hirshfeld, MathPro INC.
The Advanced Refinery Modeling System (ARMS)
is a high-performance, desktop linear programming (LP) modeling
system for analyzing the technical and economic performance
of :
- the refining
sector of the U.S. and other countries, disaggregated by
region, refinery type, or other sub-division, and
- individual refineries.
ARMS is used intensively in a consulting practice
specializing in refining economics, for policy analysis and
business planning studies dealing with the technical and economic
response of the refining industry (or individual refineries)
to real or prospective changes in public policy, regulation,
and/or market conditions. Through its advanced technology
and ease of use, ARMS conveys a competitive advantage to the
consulting practice.
The refinery LP model in ARMS is a partial equilibrium
model. It simulates the operation of the refining sector such
that:
- the market for each product clears at the
computed prices;
- each refinery is in competition with all
others in the given region; and
- all competitors have full information
about the market.
A typical instance of the model comprises approximately
1,100 constraints (equations and inequalities), 3,500 activities
(variables), and 45,000 non-zero coefficients.
Features
ARMS is an advanced system embodying established and
proven modeling technology for creating, modifying, and managing
model statements, databases, and cases. It is implemented
by means of an interactive development environment (MathPro
2000®) and an advanced model solver (Xpress-MP).
By virtue of its advanced design and implementation, ARMS
supports quick response analyses, even when the analysis calls
for modifying or extending the model structure and for running
many scenarios in a short time. The LP model in ARMS is expressed
as a symbolic, computer-readable model statement, in matrix
schematic form. Each distinct pairing of the model statement
and a set of input values leads to a distinct model instance,
or case, that ARMS processes and solves. Typical studies of
policy and planning issues involve hundreds of cases.
ARMS is "data-driven" and easily enhanced.
Data-driven operation enables one to change data values and
create scenarios without changing the mathematics or any computer
programs. It is the result of strict separation of the model's
mathematics and data within ARMS, an advanced design feature.
ARMS's design and implementation facilitate a continuing stream
of enhancements to the model's structure, without conventional
computer programming and its attendant costs in dollars and
time.
For more information contact Dave Hirshfeld of MathPro
Inc. e-mail: mathpro@mathproinc.com
|
|
