High BHE Error with low flow rates

Hello everyone, I have the following problem: I created an FEFLOW Problem to model a BHE Array consisting of 64 BHEs connected in parallel, so the Input goes to all BHEs which then goes to the output. For bigger flow rates of lets say 150 m3/d the simulation runs totally fine but for flow rates below 20 m3/d the BHE Error Norm raises astronomically high causing completely wrong calculations. We coupled the BHE Array to a TRNSYS Model, with an IFM Plugin so the low flow rates somehow need to be calculated correctly. Has someone an Idea what could be the issue? We first thought laminar flow inside BHEs could be the problem, but reading the White Papers FEFLOW should handle this right? Is there any numerical parameter in the settings we are missing? I can provide the information for our chosen settings if needed! Thanks, Max

Hi Max, Not surely whether I understood entirely. Is the problem occuring in a FEFLOW model being coupled to TRNSYS? Or are you using also in the "uncoupled" problem? Please note that the plug-in is not anymore supported by DHI. Cheers Carlos

Hi Carlos, sorry for explaining it a bit confusing. We don’t use the plugin at the moment so its uncoupled! The problem is we have a time series for the flow rates of the BHE Array and when the flow rates drop below 20 m3/h, the Bhe error gets really high and then all temperatures are wrong. hope this helps, cheers Max

Here are some screenshots for further explanation.

Hello Max, have you found a solution to your problem in the meantime? I am also experiencing high errors and extreme temperatures at low flow rates with BHEs. Kind regards Mario

Hi Max, The problem is that SAMG did not converge properly. I see you are working with a RMS norm. This is practical for standard cases, but in situations, where you have specific "hot spots" (e.g. not convergence directly at the BHE locations), it will be much useful to work with a MAX norm. The Max norm will focus on the spots with maximum error, presumably the BHE locations. Alternatively, you can give a try with the direct solver PARDISO. Best regards Carlos Rivera