OPTIMIZATION FEATURES

Optimization searches for new parameter values that will drive multiple response values to desired targets. This is accomplished by defining a range of feasible values for each parameter (continuous, integer, or discrete) and by defining constraints and goals for the statistics (mean, standard deviation, and/or probability of non-compliance) of each response. Optimization thus enables you to identify the parameter values that achieve the right balance between performance, quality, reliability and cost across multiple responses.

Features:

• Discover new designs (combinations of parameter values) that have higher performance, higher quality, lower cost, or all of the above
• Discover designs that are robust to parameter variability
• Easy-to-use and learn interface
• Unlimited numbers of factors, responses, goals and constraints*
• Extremely robust custom genetic algorithm for truly multi-objective, nonlinear, statistical, global optimization
• Mixed continuous/ integer/discrete search
• Full support of index variables and table lookup searches

*Constrained only by worksheet size/memory limits in MS Excel™

Using Optimization in Product Development

At its heart, optimization is nothing more than a tool to help people make decisions. For any decision we want to identify the best solution from a set of many different possible alternatives. For determining what is “best” we almost always have several criteria. These criteria often conflict, and some criteria are more important than others.

We perform an optimization by defining the range of different solution alternatives, defining the criteria for evaluating them, and then having an optimization algorithm perform an automated, intelligent search.

Alternatives are represented mathematically as specific values for one or many parameters.

Criteria are divided into two groups: those that must be satisfied and those that ideally should be satisfied. Criteria that must be satisfied are called Constraints, and criteria that ideally are satisfied are called Goals. Each constraint and goal is represented by its own response equation.