Correlate your simulation model to experimental data?

The first step in utilizing a virtual prototype for design is to have confidence in the predictions of the virtual prototype (simulation model). Within Siemens, when we have the virtual prototype matching the physical prototype, we call this a Digital Twin. Most people think of HEEDS as helping to improve the performance of the design via virtual prototype design space exploration. While this is a huge use-case, it should be noted that HEEDS can also help with the development of the Digital Twin itself; or in other words it can help the virtual simulation model results match the physical experimental model results. Typically, this is done upon the baseline (starting) design concept, where within the simulation model there are unknown parameters which must be tuned to give the intended performance whereas the simulation model matches the experiment. As with design optimization, correlation of these parameters can be challenging without a tool like HEEDS.

To help with this correlation task, HEEDS uses a special response type called a ‘curve fit’ to make setup and result interpretation easy, relying upon the same optimization technology used to later then optimize designs with the correlated virtual prototype. The curve fit response in HEEDS is used to measure the closeness-of-fit between an experimental data curve and a curve extracted from the simulation model. The quality of fit between the two curves is quantified through a root-mean square method (RMS). Better matching curves have lower RMS values.

Consider an example below where we are trying to match the Force-Displacement curve from simulation data to test data. Figure 1 shows the setup inside HEEDS, while Figure 2 shows an example of a correlated solution.

In HEEDS, the X values of the simulation response and the X values of the reference curve (experimental/test data) do not need to match. If they are not equal, HEEDS uses the reference curve (under manage curves section ) to automatically interpolate the Y values between simulation and target. The RMS value is calculated from all of the data points to compute a ‘closeness’ between the curves.

Once the response is setup, it should be set as an objective to minimized for an optimization. This will minimize the RMS, hence finding optimal designs with the closest fit between the simulation curve and the target curve. The curve fit plot in HEEDS POST will show the fit along with the calculated RMS for each design, as shown in Figure 2.

For more information on this technique, please contact our Global HEEDS Technical Support Team for help on implementing on your problem: [email protected]