It has been way too long since I last wrote a blog article…It is amazing to me how fast time flies between blogs. A lot of ideas pop into my head that I think would be an interesting read but all of them still reside in my head. One idea that has been haunting me is the topic of this blog. Part of the problem was that I had a hard time deciding what would be the point of the blog. Rather than wait any longer deciding what a single good point was I have decided to run with a few less significant points, and a few pretty pictures.
When I first started working for Flomerics my primary role was to support customers in their use of FloTHERM. It was an exciting time learning important (and true) concepts like-all software has bugs, feature instead of bug, and cad translation is not seamless, to name a few. I remember one support case, which now being a part of Mentor we call service request, that is the main thrust of this blog.
It was a very simple model provided to me by a customer…imagine a heated block in thermal contact with a heat spreader in natural convection. The customer felt that the model was over-predicting the temperature of said block. As a good support technician I ran through the model setup including grid independence and radiation. It was all good but still the customer thought it was too hot. I was still a relatively inexperienced engineer so I couldn’t totally rule out that there wasn’t something else I could do with the model. This is when I entered what I consider to be the “Jumping the Shark” phase. I tried advanced radiation, different turbulence models and none at all, and as a total desperation move I decided that the culprit must be the Boussinesq approximation. None of which made any significant difference. As you may have guessed by the title of this blog, the issue may have been heat spreading. I never did reach this conclusion while trying to help the customer though. Just so we are all aware of the effect of heat spreading I have taken the liberty of running a very simple model that illustrates the effect of heat spreading. Below is a series of images that show a heat generating block positioned at different locations on a heat spreader. Notice that the maximum temperature decreases as the block works towards the center of the spreader.
Looking back on this I really feel like the customer actually wanted this support case to not be resolved. I suspect that he appreciated that he could tell his manager that he was working with Flomerics support and he suspects there may be a bug in the software. I could be wrong but this thought does give me some level of comfort so this is what I choose to believe.
Because the points of this blog are a bit murky, allow me to summarize:
- Don’t expect an analysis model to predict wildly different results when tweaking knobs. (If it does I would get a different analysis tool)
- Senior engineers keep a look out for when junior engineers “Jump the Shark”.
- If the analysis model has appropriate detail and assumptions for the result resolution desired then be open to the possibility that the prediction may be something you didn’t expect. (To be honest this is the exception rather than the norm…product marketing would love if every model showed something we didn’t expect)
- The best way to help a customer is not always answering their question; Answering the specific question often times under values the level of assistance we can provide. In this specific case the customer would probably have been better served if I could have re-educated him and his expectations rather than jumping the shark like Fonzie.