CSIM 19/OptQuest is
a version of the OptQuest Callable Library (OCL), a product of
OptTek Systems, which has
been adapted to work with the CSIM 19 Simulation Engine. OptQuest
is a simulation run supervisor. It makes repeated calls to a simulation
model, searching for the "best" configuration of the
system being modeled.
Systems designers and
systems managers need to find the best possible configuration
for their systems. The best configuration is typically the selection
of components, schedules, software, and operating strategies which
accomplish one of the following goals:
Finding
this best configuration can be a long, difficult and expensive process.
One way of conducting
the search for the best configuration is to create a simulation
model of the system and then to use this model to evaluate different
configurations. CSIM 19/OptQuest offers an efficient way of accomplishing
this search for the best configuration.
In some cases, performance
can be improved significantly, while incurring no or little additional
costs, by implementing some simple changes, such as modifying
manpower schedules or changing other operating procedures. In
many cases, there are alternative components which deliver performance
that is "good enough" to achieve a performance goal,
but have much lower costs. Often, it is imperative that the system
deliver the best possible performance, in order to avoid performance
related penalties or other costs associated with less than optimal
performance. OptQuest efficiently automates the search for the
best configuration. CSIM 19 lets you build comprehensive and efficient
simulation models. The combination lets you realize the benefits
of finding the best configuration.
The analyst creates
a CSIM 19 simulation model of a complex system. This model is
written so that important components of the model (in particular
those components which control important features of the system's
operation) are supplied as inputs to the model. The analyst also
"sets up" the optimization problem for OptQuest; in
particular the analyst specifies an objective of the problem as
well as limits on the configuration variables and additional requirements,
which must be satisfied in order to have a feasible configuration.
During a run of the
application, OptQuest supplies a "configuration" (a
set of values of the input configuration variables). The model
executes using this configuration and returns the value of the
objective as well as the values of the additional requirements.
OptQuest then evaluates these returned values, and, based on these
values, presents another configuration to the model. This process
continues automatically until the application either "gives
up" and quits or until OptQuest has generated every feasible
configuration. At any time, the application can query OptQuest
to retrieve the "best configuration so far."
The general technique
is called Tabu Search. OptQuestCallable (OCL) uses Tabu Search,
along with the metaheuristic, mathematical optimization, and neural
network components of OptQuest to guide the search for best solutions
to decision and planning problems of all types. The user-written
application passes to OptQuest OCL an indication of the quality
of each solution generated during the search, by calling an evaluator
(such as an objective function) which can take any form. In some
situations, the evaluator may simulate a complex system to determine
its behavior based on a proposed configuration.
OptQuest and OCL are
from Optimization Technologies, Inc. (OptTek) in Boulder, CO.
The primary developers of OptQuest are Professors Fred Glover,
Jim Kelly and Manuel Laguna. OptTek and Mesquite Software have
entered into an agreement which covers marketing, supporting and
developing OptQuest for use with CSIM 19 simulation models.
Most certainly! The
CSIM 19 model must be structured (or restructured) so that:
-
The model can be executed repetitively,
-
The important features of the configuration
(the ones that will be adjusted as part of the study) are inputs
to the model; in particular these need to be inputs and not
compiled-in constants, and
-
The model returns the values of the objective
and any requirements that must be met.
If a CSIM 19 model has to be modified
to operate in this way, the changes required are usually minimal.
Many simulation models
are used to develop the best possible configuration of a system.
Before OptQuest, doing this with a CSIM 19 model usually entailed
successive runs of the model, trying to find a configuration that
is "good enough", to meet the cost-performance criteria
for the system.
With OptQuest, this
search for improving configurations is done automatically. The
search techniques in OptQuest usually drive system configurations
toward the optimal configuration in a very efficient manner. This
approach, based on automatic searching, is usually more thorough
and more efficient than the manual approach mentioned above.
You would use OptQuest
to help find the best possible configuration for the system you
are designing and/or evaluating.
There are some sample
models available for examination on the www.mesquite.com web site.
There are several articles and application studies on the www.optQuest.com
web site which describe a variety of successful applications of
OptQuest in many different problem domains.
OptQuest is an automatic
search supervisor. It uses simulation models to find the best
possible configuration of the system being analyzed. It turns
out that with a faster simulation model, more evaluations can
be conducted and leading to better configurations. In addition,
the simulation model must accurately reflect the structure and
operation of the real system.
CSIM 19 models are very
efficient, executing with short elapsed times. In addition, CSIM
19 models are computer programs; as such, they can be constructed
to accurately model the important features of almost any system,
regardless of the nature and structure of that system.
By incorporating all
of these features (efficient automatic searching, fast execution
speeds for the model and accurate representation of the system),
analysts can use CSIM 19/OptQuest to build "industrial strength"
applications to find the best possible configuration of the system
being analyzed. Finding the best possible configuration of a system
can result in both significant reductions in system costs and
significant improvements in system performance.
You can order CSIM 19/OptQuest
from Mesquite Software, Inc. There are two kinds of licenses:
For pricing and availability, contact
Mesquite Software:
Phone: (512) 338-9153
In the US: (800) 538-9153
Fax: (512) 338-4966
e-mail: