Future Facilities calls itself a digital twin company that offers an array of different options for companies to model and monitor data centers. The 6SigmaET division of the company, however, has another focus: thermal simulation of electronics components. 6SigmaET was born in 2009 and built around a computational fluid dynamics (CFD) solver. The new Release 16 of 6SigmaET has several improvements related to faster solutions and better visual representation of those solutions.

Joule heating calculations run faster so system solve times run faster. (Image courtesy of 6SigmaET.)

The company says that the solver is the most important part of the simulation. Model size, physical conditions, mesh size and the number of objects in the system are all contributing factors that decide how long a model’s solve time will take. Using multi-core processors is the main method that allows 6SigmaET to get shorter solve times. Options for customers range from single-user single-machine licenses to HPC scheduling from IBM or Microsoft and Rescale cloud solutions on a pay-per-solve model. Using 2.7 GHz processors and 352 GB RAM cores while testing 6Sigma showed that a hard-to-visualize 352 core system could run simulations with over 75 million grid cells more than 55 times faster than a single-core machine.

Using those fast speeds as a base, 6SigmaET continues to make incremental improvements to its system solution speeds. Release 16 speeds up the base calculations for Joule heating, reduces the time needed for importing PCB data and simplifies the process of one-dimensional flow analysis. Enhanced view tools give the user more options to create promotion-ready graphics and more thorough results reports. 6SigmaCommander gives more options for team collaboration with improvements to how data is stored and shared.

Joule Heating Uses GPU Instead of CPU for Speed Boosts

Joule heating exists at an interesting place between mechanical and electrical engineering concerns. The easiest way to think of joule heating is as the electrical energy flowing through conductive material as it turns into thermal energy. As a mechanical engineer, if the subject of joule heating came up in a lecture, it was often described as negligible, and the instructor moved on to the next topic. 6SigmaET says that high current applications, such as inverters, motor controls and power supplies, can generate enough joule heating that it needs to be studied.

6Sigma partnered with NVIDIA to use some of the graphics capabilities of the CUDA Toolkit and AmgX library to help speed up some of the base calculations. 6SigmaET’s Thermal Expert Slava Semin said that manipulating the electrical potential equation led to finding ways to optimize the base calculations.

“Our process allows coupled solutions using multiple processors for thermal simulations, via MPI and GPU processes,” Semin said. “During the iterations, the Computational Fluid Dynamics (CFD) solver assembles thermal properties to update electrical conductivity. After solving electrical potential, the GPU redistributes the Joule Heating and uses the MPI processes to get the final thermal solution.”

Taking the smallest calculations and making incremental changes allowed 6SigmaET to boost the speed of some simulation studies to be up to 325 times faster using an NVIDIA A100 Tensor Core GPU instead of a traditional CPU-based solve. The Release 16 joule heating information does note that each specific application is different. Single station customers and small-team customers can check with their service representatives to see if their specific NVIDIA processor and application will work with the faster solving methods.

PCB Analysis Import Changes and 1D Flow Incremental Bumps

Simulating printed circuit boards (PCBs) can be difficult because of the different manufacturer variations and industry specifications. 6SigmaET gives the user the option to import according to Institute for Interconnecting and Packaging Electronic Circuits (IPC) or Open DataBase (ODB) standards. After the standard system is selected, the level of detail can be chosen for the import. Detailed PCB level means that the conductors and traces of the PCB will exist in the model as distinct blocks of copper. Simplified Mode instead calculates the amount of copper in each grid block and uses that calculation for the simulation studies.

One-dimensional flow in 16 gives faster results and improved graphics. (Image courtesy of 6SigmaET.)

A Mixed mode also exists in which users can control some of the import parameters for a board and find some compromise between accuracy and efficiency. The company recommends that a detailed PCB mode be used for applications when a specific section of the board is being targeted for study to get the best possible results for a possible problem area.

One-dimensional flow studies also see some incremental changes. Previous releases broke flow into Air Side and Liquid Side Networks, but Release 16 shows a single Flow Network. The ability to solve multi-component systems as a whole or as co-simulations still exists. Valve objects are new in this release, giving the option to turn fluid flow on or off and control the flow direction. Sensor objects and component objects can be placed in the model to calculate temperature and pressure drops across model points.

Data Handling and Results Visualization Benefit from Changes

6SigmaCommander has a long list of new features and commands added to it for the ease of moving, sharing, and translating data. Users can automate CAD and PCB imports, update, and results generation.

Enhanced View feature brings the software closer to photo realism. (Image courtesy of 6SigmaET.)

Enhanced View is the new visualization tool built to bring the software toward more photo-realistic graphics. Shadow details have been added, and materials now look more reflective and distinct using Enhanced View. Model manipulation also benefits from these changes. An example on the graphics page shows an Enhanced View spinning at 60 fps next to a Graphics View moving at 1 fps. New camera placement and camera movement functions should also allow the software to give better results animations.

What Does It All Mean?

Releasing a new update to software that’s been in existence since 2009 is tough. A company needs to have some big changes along with lots of smaller changes. 6SigmaET’s big changes seem to be the transfer of Joule heating simulation over to a graphics model instead of the CPU model and the push to enhance the way that the software is doing visualization. There are numerous small changes made to how data is handled as it moves both in and out of the software and lots of areas where it looks like one or two enhancements were made.

The graphics shift follows the full introduction of Enhanced View after a soft debut in Release 15, and it feels like the software was ready to update its graphics game in response to society’s shift toward more screens and more visuals. Many of the other changes look like continuous improvement items added into the release either to fix issues in the code or to make refinements addressing customer complaints or requests.

6SigmaET wants to be a tool for the specific application of thermal simulation for electronics. Because of this, a few differences stand out over the last several revisions. The software continues to allow the building of models right in the software. If the macro goal of simulation is to help engineers and designers to work more efficiently and create better products, then the changes made here in this release push the software in the right direction.