Design Variable Form

_images/Design.png — Design Variable Form

The Design Variable form is created by clicking the Design Variable button on the main form. This form allows the user to manage design variables. Like in the State Variable form, design variables can be made active or inactive. An “active” design variable is one whose value is allowed to change in the optimization process. An “inactive” design variable does not change during the optimization process but is set to the current value. An active design variable is shown with black text in the table; an inactive design variable is shown in gray. To change the active status, click the checkbox in the Active column. To change multiple at once, select the desired design variables from the table, right click and select Toggle.

The values (initial or current) shown in the table are normalized to a 1.0 nominal value. The initial value corresponds to the value specified on the Nastran DESVAR card. The current value of selected design variables can be changed by entering the new design value in the corresponding row of the Current column of the table. Values can also be entered by using the copy and paste functionality in the context menu of the table. User-specified changes to the current design are not propagated beyond the Design Variable form until the user clicks Update. The background color of the current value column changes to reflect the magnitude of the change in design variable value depending on the preference set on the global preferences design tab.

Bounds on design variables may be specified in two ways. The “upper bound” and “lower bound” are permanent bounds on the design variable. They do not change during the optimization and model-updating process unless the user changes them. The “limit” defines a fractional change above and below the initial value of the design variable that defines the valid range for the design variable. As the initial value changes (it changes whenever the DESVAR card is updated and Nastran is rerun), the range specified by the “limit” will move with it. The bounds used during optimization will be the effective bounds created by the combination of the upper and lower bound and the limit. For example, if the initial value of a design variable is 1.0, the limit is set to 0.5, the upper bound is set to 2.0, and the lower bound is set to 0.8, the effective bounds will be 0.8 and 1.5.

The limits and bounds for a design variable can be set in two ways. For setting a single design variable, the value in the table can be edited directly. For setting multiple design variables at once, values can be copied and pasted. To copy a value or range, select Copy from the context menu or press Control+C on the keyboard. Values can then be pasted on any editable column in the table by using the corresponding Paste context menu or Control-V. A value can be unset by deleting the value in the cell. The blank cell can be copied and pasted to clear multiple values as once.

Just as individual weights can be set on the state variables, they can also be set for design variables. Again, the nominal value is 1.0. Increasing the weight on a design variable increases the cost of changing that design variable, resulting in a smaller change to that design variable. Decreasing the weight on a design variable is likely to increase the change in the design variable.

The same selection buttons that were available on the State Variable form are available on the Design Variable form. The “All” button selects all design variables. “None” deselects all design variables. “Active” selects all active design variables. “Inactive” selects all inactive design variables. This is helpful for toggling the active status of the variables when they have been previously set.

The Design Variable form also has a mechanism that allows the user to filter design variables based on sensitivities. The sensitivity value displayed in the table refers to a “composite sensitivity,” which is a single value that characterizes the sensitivity of all state variables to this design variable. The derivation of composite sensitivity is given in the Theory section of this manual. The minimum and maximum composite sensitivities are displayed in the lower-right-hand portion of the form. The user can specify the minimum composite sensitivity for an active design variable by entering a value in the field labeled “Cutoff.” Design variables with sensitivities less than the cutoff value (in absolute terms) will be marked inactive. This will override any user-specified active status for the design variables; however, the user may alter the status after the cutoff has been applied.

To further explore the sensitivity of each design variable, select a desired design variable and press Plot. This will produce a bar chart that shows the sensitivity of each selected design variable to all active state variables.