@RISK/Evolver: Debugging RISKOptimizer and Evolver Models

To debug RISKOptimizer/Evolver: remove redundant constraints, ensure initial feasible solution, adjust penalty functions, and use Constraint Solver if available.

Dec 18, 2025
Knowledge

Title

Answer

Applies to:

RISKOptimizer in @RISK 5.x and up Industrial Edition
Evolver 4.x and up
Evolver Developer Kit (EDK)
RISKOptimizer Developer Kit (RODK)

How to Debug RISKOptimizer or Evolver Models

Question:
What are some strategies to debug my RISKOptimizer or Evolver model?

Answer:
Here are several suggestions to help ensure your model runs efficiently and produces meaningful results:

1. Check for Redundant Constraints

Only the main constraint should directly apply to the adjustable cell range.
Avoid constraints on constraints (“subconstraints”), as they can cause RISKOptimizer to waste time finding solutions that satisfy the main constraint but violate a subconstraint.

2. Consider Penalty Function Magnitude

When using penalty functions, ensure the formula scales appropriately with your model’s numbers.
For large deviations (thousands or millions), the default penalty formula:
```
100*(EXP(deviation/100)-1) 
```
may exaggerate deviations. Instead, you can use a simpler formula like:
```
deviation 
```
Note: This applies only to the Genetic Algorithm. RISKOptimizer 6.0+ allows choosing between the Genetic Algorithm and the OptQuest engine.

3. Satisfy Constraints in the Initial State

For the Genetic Algorithm, ensure all constraints are satisfied before starting optimization.
A good starting point prevents the software from exploring solutions far from feasible regions.
See: Starting Values of Adjustable Cells for guidance.

4. Use the Constraint Solver (Evolver & RISKOptimizer 5.5+)

The Constraint Solver (under the Utilities menu) can find an initial feasible solution automatically.

Alternative for Older Versions (≤5.0):

Create individual Excel statements for each constraint, evaluating to TRUE or FALSE.
Use an IF statement:
- If the constraint is FALSE (not satisfied), return an arbitrary number (e.g., 1).
- If the constraint is TRUE, return 0.
Sum the column of IF statements into a target cell.
Set the optimization objective to minimize this sum (goal: reach 0), keeping adjustable cells unchanged.

This creates an initial feasible solution before running the main optimization.

URL Name

debugging-riskoptimizer-and-evolver-models