Rockwell Automation Arena OptQuest Users Guide - Manual (Page 48)

A RENA O PT Q UEST U SER ’ S G UIDE
44
• •
•
•
•
Although the optimization model is correct, the control bounds are not meaningful,
because the upper limits cannot be reached given the constraint above. A better set of
bounds for these controls would be:
0 <= X <= 20
0 <= Y <= 16.667
0 <= Z <= 10
These bounds take into consideration the values of the coefficients and the constraint limit
to determine the maximum value for each control. The new “tighter” bounds result in a
more efficient search for the optimal values of the controls. However, this efficiency
comes at the expense of missing the optimal solution if it lies outside the specified
bounds.
Constraints with varying bounds
When plotting an efficient frontier graph, each point on the graph represents an entire opti-
mization consisting of numerous simulations. For example, if the constraint defined five
different bounds, then the overall optimization will consist of five separate optimizations.
Naturally, getting good results for each sample increases the overall time required
compared to an optimization with no varying bounds. Fortunately, OptQuest does not start
each sample from scratch, but rather uses what it has already learned to save time.
The number of simulations specified on the Options node is divided among the number of
bounds on the constraint. You should increase the number of simulations when your opti-
mization contains a constraint with varying bounds.
Complexity of the objective
A complex objective has a highly nonlinear surface with many local minimum and maxi-
mum points.
OptQuest is designed to find global solutions for all types of objectives, especially
complex objectives. However, for more complex objectives, you generally need to run
more simulations to find high-quality global solutions.
Feasibility
A feasible solution is one that satisfies all constraints. OptQuest can check a solution
against linear constraints before running a simulation. Only solutions that satisfy linear
constraints are set to Arena for evaluation. After the simulation has completed, OptQuest
checks non-linear constraints for constraint feasibility.