COSMOS/M Design Optimization

To optimize a design, you start with a base model and then determine the parameter (called the design objective) that you wish to either maximize or minimize. Then you choose the restrictions you want placed on the model's response (called behavior constraints), and you're on your way to an improved design. COSMOS OPSTAR helps you optimize both size and shape, giving you a wide variety of design variables.

Your objective function can be volume, weight, stress, strain, displacement, velocity, acceleration, reaction force, natural frequency, buckling load factor, temperature, heat flux or temperature gradient. Behavior constraints can be stresses, strains, displacements, frequencies, buckling loads, temperatures, heat fluxes, temperature gradients or any combination!

One of the newest features of OPSTAR lets you define your own objective function and behavior constraints for optimization and response quantities for sensitivity. And OPSTAR will support multiple load cases to give you a product optimized for many conditions. It will also support structural and thermal transient analyses and applications involving material and geometric nonlinearities, and contact problems.

Optimization requires many analysis iterations and solutions. With OPSTAR and FFE, you can reduce analysis time and disk storage dramatically. FFE analyzes complex designs up to 100 times faster than conventional FEA technology, and reduces disk storage requirements by over 20 times. That means you can optimize large models on desktop platforms in just hours.

Sensitivity Studies

Sensitivity analysis enables you to define design variables and examine their effects on your model's behaviour. OPSTAR offers three types of sensitivity studies prior to optimization.

Global Sensitivity - You specify a number of steps over which your design variables will be changed between their lower and upper bounds. The number of steps will remain the same for all variables, and you can choose to change all the design variables simultaneously or one at a time.

Local Sensitivity - You can perturb one variable at a time with a value you define, while all others remain constant. You can then use the results to find the derivatives of the response. You also have the option to retrieve the sensitivity information used in optimization.

Offset Sensitivity - You specify the values of a series of design variables and COSMOS updates the model to correspond to those values. You can then choose the type of analysis to perform.

With OPSTAR, you have all the tools for making sure - very fast - that your design meets all its performance criteria and still gives you the greatest economy of resources.