topic Re: What accuracy can be expected in rotating region analyses? in CFD Forum

What accuracy can be expected in rotating region analyses?

PaulNielsen — Tue, 22 Dec 2015 09:38:05 GMT

Following the steps provided in

https://knowledge.autodesk.com/search-result/caas/sfdcarticles/sfdcarticles/What-are-the-best-practices-for-running-a-rotating-region-analysis-in-Simulation-CFD.html

What kind of result accuracy could be expected compared to a live test run on a physical prototype?

Could we expect accurate results within a 5% error or should we expect larger error and that the simulations could only be used for trending analyses?

Re: What accuracy can be expected in rotating region analyses?

Jon.Wilde — Wed, 23 Dec 2015 07:55:20 GMT

I would expect us to be in that region, yes. It depends how well the model is set up.

I tend to say less than 10% but I know we can do better if we have the correct approach.

Re: What accuracy can be expected in rotating region analyses?

Royce_adsk — Mon, 04 Jan 2016 16:27:27 GMT

I don't like to contradict Jon, but this is my opinion in this regards this topic.

0 - 5% ... Possible (I would be really ecstatic to be in this range): Best practices followed, appropriate mesh, type of device being modeled, and simulation assumptions fit testing assumptions perfectly.

5 - 15% ... Typical range: Where I would hope we would end up most of the time and when one or more of the above are slightly off.

15 - 20% ... Still acceptable range and can be used very well for trending for multiple operating points.

When you are doing simulations of rotating regions you always want to compare the full useful operating curve and not just a single point. You will many times find single points where the physics just don't work out very well, but the rest of the curve looks great. No reason to fight a single hard to solve point when everything else looks excellent. Case in point, you can fairly easily solve for the lift curve of an airfoil before it stalls, once it stalls it becomes more difficult.

Re: What accuracy can be expected in rotating region analyses?

PaulNielsen — Tue, 19 Jan 2016 13:20:55 GMT

Interesting. We are currently investigating how a customer of ours can make use of this kind of simulations to automate the process of calculating head-capacity curves for their line of pumps.

Our problem is that despite more or less following the guidelines and recommended settings we still get an error of about 30% on pressure and flow. The efficiency curve is ok though... This could be due to faulty input from the client, but do you have any kind of verification or benchmark model for these kind of simulations?