Recently I was thinking about how Computational Fluid Dynamics (CFD) accelerates all phases of Thermal Design. I’m sure it is debatable but here is what I came up with for the phases of Thermal Design . The arrows could be looped over and back and round and round but I think the flow is reasonable.
The basic 4 step design process that leads to the desired Design Approval depicted is:
- Design Need and Ideas: Basically the process would be started by the need for a design or an idea related to a design.
- Computer Analysis: Once the light bulb is lit or the fire is stoked the designer would set out to study that idea. If all the design tools known to man are available a fair place to start would be in front of a computer.
- Build and Test: Realizing that nothing is true until it is tested, this is a necessary step. If Step 2 is done well enough then this step should be quite a bit easier.
- Share Design Solution: Everything needs to be sold at some point, and this is where a compelling argument for the design needs to be made.
I plan on dedicating this post to discussing Phase 1 a bit further and follow up with the remaining phases during subsequent posts.
Phase I:Design Need and Ideas
This phase would be the first of the four phases leading up to approval and would begin because either:
- There is a Need-A designer needed to develop a thermal design to address a specific issue
- There is an Idea-A designer conceptualized a thermal design that might address a future need (the kind that might make you a bit more valuable to your employer or customer)
Okay…CFD doesn’t accelerate the “Need” but if you think about what it takes to get out of Phase I into Phase II, CFD can play a big role. Performing CFD analysis allows such great insight and intuition, just as years of lab work would, into what would or might work, and what won’t work. I believe the data learned from the analysis, presumably based on “good” models, allows for more insight into the physics than lab tests. I appreciate a good experiment almost as much as the next person but there is no denying that a CFD analysis gives a broader scope of the phenomenon than a test set up to measure discrete data points. A couple of examples to illustrate knowledge that could be gained easily in an analysis software are shown in the images below.
In the first example a designer analyzes 4 heat sink designs and simulates them with various emissivity values. Based on the results this designer can look at this and realize that if the future design incorporated either of the lower two heat sinks (green and purple) in the plot there is no reason to bother with radiation as a mode of heat transfer to enhance; The emissivity for a practical purposes is a non-factor. In the upper two heat sinks (blue and red) huge performance increases can be realized if radiant heat transfer is considered when designing the cooling solution.
In the second example a designer looks at one heat sink design and varies the thermal conductivity in the analysis. The results show that there is really no point in manufacturing this heat sink out of a material greater than 5 W/mK. A lot of engineers would interpret this as, ” this is one lousy heatsink, are there any fins at all ??”. It is safe to say that the thermal conductivity of this heat sink is not the biggest thermal bottle neck.
The point I am trying to make is that CFD allows you to quickly set up studies and provides a complete set of results that allow the designer to gain huge insight into how heat and fluid move around a system. This information will lead to the ability of a designer to reject or discourage a proposed design idea based on intuition and experience. Intuition and experience that was gained more simply by being exposed to more designs that worked and failed.