Image courtesy of Mentor Graphics.
The latest version of FloTHERM has a functionality that automatically calibrates simulations to measurements collected by T3Ster, Mentor Graphics’ temperature test and measuring system. This calibration will help improve simulation accuracy and validation for future tests.
“By using an uncalibrated model for simulation, manufacturers run the risk of simply guessing performance. Calibrating the package model constants ensures the product will respond accurately in all steady state and transient applications,” stated Roland Feldhinkel, general manager of Mentor Graphics’ mechanical analysis division.
T3Ster thermal transient tester. Image courtesy of Mentor Graphics.
FloTHERM and T3Ster convert their respective simulated and experimental transient thermal responses into structure function curves using identical mathematical algorithms. These curves correlate to the tested device and allow for optimal comparisons between the simulated and experimental data.
As with traditional validation methods, inconsistencies between the simulation model and experimental model typically point to errors in the simulation. Often these errors are estimated values that are hard to measure, such as the thermal interface material thickness or the interfacial thermal contact resistances.
Using optimization algorithms, FloTHERM will change the parameters of the simulation until the results match with the experimental model. The now validated and calibrated simulation will produce accurate estimations in future tests performed by the design team.
“Working with a fully calibrated model of a package is often critically important in steady state and transient analyses when accurate prediction of the junction temperature is the objective,” said Ir. Clemens Lasance, retired principal scientist at Philips Research. “The calibration is usually performed by creating a structure function to be compared with the one measured with a T3ster in the lab. The latest version of FloTHERM embeds an automated ability to fine-tune a FloTHERM model to achieve this match, making this best practice modeling methodology far more accessible to thermal engineers."
Using optimization techniques to calibrate a simulation to an experimental model is rather clever. In many industries, not all the values are known, or can be measured, before the simulations start. As a result, these simulations will not always match the experimentations when the validation process starts. The optimization algorithms can help to further estimate these unknown values and ensure models run properly.
However, a tool like this could, theoretically, run the risk of giving a user a false sense of security. It is vital that engineers understand the physics behind their simulations or they risk a garbage in, garbage out situation. The FlowTherm/T3Ster calibration method may appear to calibrate the model. However, in a situation where the engineer is uncertain of the underpinnings of the system, an engineer may not recognize that this apparent calibration might be a localized event.
As a result, the calibration tool creates a stronger need for engineers to understand the physics of their systems.
Other additions to FloTHERM include:
- DC electrical calculations for joule heating
- 64-bit support for bigger MCAD imports
- MCAD voxelation processes further optimized for speed and accuracy
- Localized grid spaces can now overlap for larger, more cluttered geometry
- DCIM software development kit to display FloTHERM simulations in DCIM software like FieldView 2015