Thermal Design Stages-Phase 2

About a week ago I wrote the first part of a four part series on how I believe that Computational Fluid Dynamics (CFD) accelerates thermal design from the “Need” to “Acceptance”, which can be found here.  This second part focuses on the Computer Analysis portion of the process.

Computer Analysis

I’d like to think that I am not remotely experienced enough to know first hand when desktop computers really changed the way engineers approach their designs.  When I went to college, for the second time, my dad had just purchased a 386.  I remember, like it was yesterday, telling him that he would never be able to fill his 350Mb hard drive.  Times have changed, to say the least.

Spreadsheets and CFD are now essential tools in a lot of design processes and are certainly the two computer based design tools I use the most.   The beauty of a spreadsheet calculation is the speed.  Though the number of assumptions is high and the accuracy low, when properly used it can quickly guide you in the proper direction.

With CFD, the number of assumptions is greatly reduced which increases the value of the analysis, but not all CFD will necessarily reduce the computer analysis time.  General purpose CFD based tools can definitely provide insight into complicated physics that would be very difficult, if not impossible, to test.  If these same tools were applied to less complicated engineering applications the time spent during analysis may not be worth the effort.  If an engineer told his team that it would take a week to get one set of CFD results I think that rules-of-thumb and “doing it the way we did it last time” would become very attractive.  In a nut shell, not all CFD tools are the same and some are better at working within and improving an existing design process.

In addition to general ease-of-use concerns, a CFD tool can reduce the computer analysis portion of the design process the following ways:

Leverage existing data

3D CAD solid model-Many designs are started in a 3D  CAD tool which contains a mechanical definition of the design.  This data can either be exported as a neutral file format (i.e.  sat,  step, etc.) or simply in the native format which can be imported into CFD  tools.  The downside is that once this translation has occurred all parametric relationships between the parts are lost. Many times it makes sense to perform the CFD design within the 3D CAD tool bypassing the translation phase and allowing the designer to directly access to all of the geometric relationships and dimensions.   (Mentor’s FloEFD product is this type of CFD tool)

IDF Board Layout-In the Electronics cooling industry the Intermediate Data Format (IDF) is widely used to communicate the PCB mechanical layout from the board layout tool to the CFD tool. The standard may (not sure) support thermal information in terms of heat dissipation and thermal resistances but in my experience it is used primarily for the layout.  There are better links between some of the board layout, CFD, and 3D CAD tools that are gaining traction, but that is another topic.

Application specific software

When a thermal/airflow model can be created out of parts that contain more than the geometric definition the design process can be greatly accelerated.  When this is taken one step further and the designer has the ability to create application specific parts (i.e. IC components, PCBs, fans, heat sinks, etc.) by inputting data  as presented in a data sheet the designer becomes even more productive.

Thermal Design SmartPart

Heat Sink SmartPart

The advantages are that a thermal model can be more quickly defined and iterated because the designer has controls over industry specific parts like heat sinks or TECs, for instance.  When this approach is taken it is quite easy for the designer to extend the analysis to include Design of Experiments (DOE) and optimization.

(Mentor’s FloTHERM product is this type of CFD tool)

Standardized thermal/airflow models

Since the late 80’s  Electronics Cooling Designers have  been able to include the effects of fans, thermal interface materials, heat sinks, etc., in their analysis because the various data sheets included the specific thermal/airflow information they needed.  It hasn’t been so easy on the IC component side and the prediction of junction temperature though.  Not so long ago when a thermal engineer looked at an IC data sheet the only information for them was a mechanical drawing.  This is a far cry from adequate.  If, by some miracle, the IC vendor provided a complete detailed definition, the computational overhead would preclude it’s effective use in a system design model.  It has taken quite a lot of years but the industry and most IC vendors are accepting, and providing, quality thermal information in their data sheet.  This information can be junction-to-case and junction-to-board resistances or, even better, a JEDEC DELPHI standard compact thermal model.  (Mentor’s FloTHERM.PACK product is this type of  tool)

When airflow/thermal data can be efficiently and accurately captured in an analysis model a more complete predictive model can be created.  In recent years vendors are not only providing the necessary information in data sheets but also provide CFD tool specific parts that allow their customers to more quickly achieve a quality design.

Though CFD, in general, is a valuable design tool all are not created equal.  Choosing the right one can reduce the design time spent during computer analysis.

Post Author

Posted May 30th, 2009, by

Post Tags

, , , , , ,

Post Comments

No Comments

About John R Wilson's Blog

Insights into the practical application of CFD to the thermal and airflow design process. John R Wilson's Blog


Add Your Comment