23 March, 2017
Aerospace and defense giant Thales recently released a paper entitled ‘State of the Art of Thermal Characterization of Electronic Components using Computational Fluid Dynamic tools’ . The paper focuses on the creation of a realistic 3D model of copper traces and board structure (as opposed to modeling effective thermal conductivity) and, amongst other things, validates a variety CFD software against experimental data and compares computational performance.
The component investigated by Thales was a Ball Grid Array containing a thermal test die. Steady-state and dynamic thermal measurements were taken in accordance with JEDEC-JESD51. The following simulation tools were compared: for ICEPAK®, SC-STREAM®, FLOTHERM – XT® and 6SigmaET. 6SigmaET was shown to provide excellent agreement with all experimental data presented; the maximum error observed was -1.5%, with most errors being ±0.6% or less.
In terms of computation performance, the first area of comparison was meshing. 6SigmaET – with its automatic, intelligent gridding that is parallelized to improve performance - fared well against its competitors taking 4 minutes to generate a 9.5 million cell versus 2 hours 45 minutes (1.9 million cells) and 8 hours 4 minutes (28.9 million cells) for FLOTHERM-XT® and ICEPAK ® respectively.
The next area of comparison was time taken to solve. Thales performed each simulation on the same system: 2 Intel Xeon E5-2680 running at 2.7GHz with 32 cores and 256GB RAM. The number of cores used was dependent on available number of licenses for each software tool. The results, alongside meshing time, are summarized below .
The time savings displayed by 6SigmaET, especially in terms of meshing, are considerable. Taking 4 minutes compared to 8 hours 4 minutes effectively saves an entire working day of time. This combined with the speed of solving would accelerate the thermal design cycle without sacrificing accuracy.
A full copy of the paper can be found here.
 E. Monier-Vinard, R. Brice, V. Bissuel, N. Laraqi, O. Daniel and M. Kotelon, “State of the Art of Thermal Characterization of Electronic Components using Computational Fluid Dynamic tools,” IEEE, Piscataway, 2016.