Learning Forum

[cfdjannis]

Dear forum,

I would like to simulate the melting of ice. For this I have created a 2-dimensional transient model. The model represents ¼ of a circle. Unfortunately it comes very often to the error "Error at Node 0: floating point exception" I have already googled but could not find a solutin. Attached are some pictures from Fluent and I've coppied some text from the Console. Can anyone tell me what settings I have to make so that the error no longer occurs and the model calculates cleanly?

Best regards and thanks for your efforts

Jannis

Check and Quality.PNG

[Rob]

As you may have seen in the rules, staff are not permitted to open/download attachments so please re-post the images. nThe error suggests the solver diverged, so check mesh quality, boundary/cell conditions and models. How do the residuals look prior to the failure? n

[cfdjannis]

https://www.directupload.net/file/d/6124/fzbws2yn_png.htmnhttps://www.directupload.net/file/d/6124/tu6vlp3o_png.htmHello Rob, nthank you for your reply. nI've uploaded the pictures, can you see them now?.n

[cfdjannis]

Ahh now i've marked your answer as a solution... can i change it? n

[Rob]

Not sure, but you need to embed the images. Drag the image into the text box. n

[cfdjannis]

Hello Rob,ok, I will try it again.nnThe calculation stops before I can see the Residuals...nnBest wishes and thank you for your supportJannisn

[Rob]

Cell quality looks OK, have you got good refinement of the mesh? What time step did you use, and what is the material density?n

[cfdjannis]

Thank You for your answer Bob. n nWhat do you mean with refinement of the mesh?nI will attach a picture of the meshing-tool.nI've also tried to activate and disable adaptive size but it doesn't changed something. I've tried a lot with the time steps. I've used one second up to one minute….n nFor the density of water I‘ve used the "piecewise-linear" function. nAnd for the solids and the air I’ve used a fixed value.

[Rob]

If the density of water varies, where does the extra volume go? n

[cfdjannis]

This is a good question. nIn a 2-dimensional model isn't any space for the growing volume... right?.I've changed the water density to constant and it looks like Fluent is calculating. nBut I think I have to use different values for the density, otherwise i do not represent the reality. Right?.

[cfdjannis]

Its still calculating...Here I've upload a picture of the residuals. Do they look good?n

[Rob]

If density is constant I suspect there's no motion to calculate. Have a look at the Bousinesq approximation. n

[cfdjannis]

Okay, I think I've found the problem, but I'm not sure.I used measured values including the density anomaly of water. nIs it possible that Ansys has problems with this?nIf I take values that do not reflect the density anomaly, Ansys seems to calculate correctly. nIs there no way to represent the density anomaly of water?But I get results in any case. nThank you very much for your efforts and your help!n

[Rob]

What density anomaly? Water is weird but still only has one density for each temperature? n

[cfdjannis]

I mean that the value of the density isn't straight or linear:Usually, substances expand when they are heated. But if you heat water starting at 0°C, you will notice something weird:At first, the volume of the liquid does not get bigger, but smaller! Up to a temperature of 4°C, the water contracts when heated. Once it has reached that temperature, it starts to expand with increasing temperature just like other fluids. That means water has a higher density at 4°C than at any other temperature. [http:/www.vias.org/kas/en/water_anomaly.html]nBut yes, it still has just one density for each temperature...

[Rob]

Your problem may well be that you're changing the density of the water in a sealed volume. The extra volume has to come from, or go to somewhere. Bousinesq approximation fudges the body force to mimic the temperature effect on density without changing the material density. nI knew about the water density peaking at 4C, didn't know it had a term. n

[cfdjannis]

Ah okay, nI will check it out. nSo I think we can close this question. n I'm very glad that you helped me!nThank you very much for your support.nBest wishes and have a nice weeknnjannisnn