22 April, 2020
When designing a new electronics product, engineers must balance the various demands of safety, product reliability, time-to-market and costs. When striking this balance, it’s essential that electronics engineers select the very best CFD (Computational Fluid Dynamics) software for the thermal simulation of their projects.
But with multiple players in the market, and a huge variety of features to choose from, how can electronics engineers ensure they’ve selected the most reliable tool for their project?
Which CFD software is best really depends on the unique challenges of the project at hand. Different solutions offer different strengths, but a combination of speed, accuracy and ideally low costs is always vital.
So, when deciding which CFD platform is best for thermal simulation of electronics, what are the features that you should look out for?
When looking for the best CFD software for the thermal simulation of your electronics equipment, there are a number of things that need to be considered:
Accuracy is vital for a CFD platform. To guarantee accuracy the equipment needs to be modelled with sufficient grid to describe both the physical geometry and features of the flow field. To do this efficiently usually requires an unstructured grid system which can concentrate necessary grid cells in important locations. Ideally some form of automated gridding should be offered to both guide the user as to where grid is needed and reduce the amount of manual effort required.
For maximum accuracy, CFD platforms should also incorporate simulation of a wide variety of objects and environments. This includes the ability to simulate infra-red heat radiation and solar radiation, as well as accounting for material property variation due to things like environment temperature or altitude. These external considerations all add to the accuracy of a simulation, ensuring that there are no nasty surprises when the end product goes to market.
Fast solving speeds
After accuracy, the next biggest consideration should be time — the time required to obtain a solution. The total time taken to get a solution will consist of two parts — the time needed to build the model and the time taken to solve it. Fast model building requires efficient use of pre-existing mechanical CAD and electronic designs with options to model parts with an appropriate level of detail.
For a detailed electronics equipment, solving the Navier-Stokes flow equations involved in generating a thermal simulation can be a huge undertaking.
Structured grid solvers can be most efficient in terms of time taken per grid cell but the reduction in grid cell count offered by effective unstructured gridding can outweigh this advantage meaning that unstructured grid models will often be the fastest and keep solving times to a minimum. An effective solver should be able to ensure that areas which do not call for highly detailed grids are treated at a suitable (lower) resolution, while smaller, more complex objects are provided with a far finer simulation grid.
In this way, a design that would have demanded nearly one billion ‘structured’ grid cells can be simplified to just six million ‘unstructured’ cells, providing an accurate solution in a matter of minutes rather than hours or days.
Of course, solving the equations in a parallel rather than serial manner can also reduce solution time significantly and the more CPUs available for solving the faster the solution. The second big consideration is where exactly this solving takes place. If your project requires a lot of compute power — but needs to be turned around in a short space of time — then you’ll need a CFD platform that offers cloud solving. Through cloud solving, electronics engineers can offload complex CFD simulations to cloud-based servers on demand, drastically reducing simulation times without the need for expensive on-site hardware.
While accuracy and speed are by far the most important factors for choosing a CFD platform, cost is inevitably also going to be a consideration. The costs of simulation suites vary, and direct cost comparisons are available from various independent review sites.
For students and those looking to simply experiment with thermal simulations, there are reduced price academic license and even free open source CFD platforms available. If, however, engineers are looking to generate simulation for a device that is actually going to market, then a professional-grade thermal simulation platform is always the better option.
The final consideration, when deciding which CFD software is best, is flexibility. The problem that a lot of engineers have with thermal simulation packages is that they get ‘locked in’ to a particular system, making it very hard to compare or switch in future. In fact, despite 71% of thermal engineers not being completely satisfied with their simulation packages, 60% haven’t switched their provider in over 3 years.
To address this issue, 6SigmaET recently worked with Intel to develop a ‘neutral file format’ for CFD. As a result, models and simulations can now be shared and switched between different CFD software platforms. This means that engineers can easily move away from old or outdated platforms, and can test their products in multiple simulation suites without losing access or functionality.
Want to find out more about how the world’s biggest electronics brands generate fast, accurate thermal simulations? Click here to read our case studies.
Blog written by: Tom Gregory, Product Manager