Rockwell Automation Arena OptQuest Users Guide - Manual (Page 50)

A RENA O PT Q UEST U SER ’ S G UIDE
46
• •
•
•
•
The objective on the left in the example above has significant amounts of noise caused by
the probability distributions used to model the problem's uncertainty. For these types of
objectives, OptQuest might have trouble discerning the minimum or maximum value. You
can detect noisy functions by watching the Best Solutions window for best solutions that
seem to “bounce around” from one set of values to completely different sets of values. To
help solve this problem, you can increase the number of replications per simulation.
On the right, the objective appears smooth due to the relative certainty in the model
assumptions. In these cases, OptQuest should quickly converge to the best solution.
Simulation speed
By increasing the speed of each simulation, you can increase the number of simulations
that OptQuest runs in a given time period. Some suggestions to increase speed are:
„
Quit other applications
„
Reduce the amount of uncertainty in your model
„
Reduce the size of your model
„
Increase your system’s RAM
Designing your model for optimization
When you are setting up your Arena model, there are certain things you can do to prepare
it for optimization. Keep in mind that OptQuest automates Arena. You want to be sure that
any automation or control logic that you set up does not interfere with the functionality or
reliability of OptQuest or its results.
The Arena model should be able to run all the way through from beginning to end without
any pauses or interruptions for user input. For example, UserForms and Message boxes
that wait for input from the keyboard will cause the model to pause and wait for some user
interaction. With OptQuest running, you may not realize that Arena has paused and is
waiting for some input. OptQuest is waiting to be signaled that Arena is finished running
a replication and will not know that Arena has paused. In this situation, it may look like
OptQuest is “stuck” and is not calculating any results when, in fact, Arena is simply wait-
ing for user interaction, as it was told. This situation can be avoided if your model is able
to run from beginning to end with no user interaction required.
Be sure that any changes to the values of resource capacities or variables in the model do
not affect the controls, constraints, or objectives unless this is integral to the validity of the
model. For example, control logic could be used to increase gradually the value of a
variable, and this may be an accurate representation of the system being modeled. In this
situation, OptQuest may be set to vary the beginning value of the variable, and then the
control logic would simply increase the value from that starting point. As long as this is
taken into consideration when viewing the results, this is perfectly correct. However, if the
variable is set by OptQuest then immediately changed by the control logic, the results
reported in OptQuest will reflect the changed value and not the value OptQuest had set for
the control.