About Design Objectives
Use
Design Objective (TOBJ or TINDEX) to define physical features to optimize by, for example, reducing the mass, minimizing the frequency, or minimizing temperature. You select response types (RTYPE) that are treated as the objective function during topology optimization. Some types of responses require you to select load cases.
The computational cost and disk space are dictated by the number of responses retained in the optimization. For example, the sensitivity calculation file is a temporary file that is generated during the optimization. Its file size could be as big as the number of direct responses multiplied by the number of design variables, multiplied by eight bytes.
In topology optimization, the number of design variables is often large, because each element can have its own design variable. So if the number of responses is also large, the multiplication can result in a huge value. There are also cases that require additional multiplication in the calculation, such as von Mises stresses, that link to six direct responses, one for each component stress.
For constraints, the responses created are screened, and only the 20 (default) active constraints are retained for each load case. However, for objectives, all the responses created for an objective are retained during the optimization process. Therefore, for design objectives, it is important not to create a large number of responses.
| Tip: The general guideline is that for an objective, the number of responses created should not exceed 20 responses per load case. Tip: If you are creating more than 20 responses for an objective, consider using other formulations that are more efficient. For example, select only the critical region, and not the entire geometry. Or consider using the responses as constraints instead of objectives. |