## Fluids

Topics relate to Fluent, CFX, Turbogrid and more

#### 2D negative cell volume when change fluid from air to water

• pgalves
Subscriber

Hello

I have a simple 2D problem of a pipe in which a sphere is placed in the middle (see figure). I want to simulate the motion of the sphere due to the fluid flowing through the pipe. The pipe has 1 inlet in which the velocity is prescribed as 0.30 m/s and 1 outlet where the pressure gauge is set to 0. The flow is laminar. The sphere has a 0.5 mm diameter, which gives a 6.54e-11 m3 in volume. Since the sphere material has a density of 5200 kg/m3, that gives a mass of 3.40e-7 kg.

• YasserSelima
Subscriber
First, as this is 2D problem, Fluent recognizes the sphere as a cylinder. Unfortunately you need to do 3D simulation for this case.nSecond, the drag force is proportional to the fluid density. This means the force in the case of water is almost 800 times that in case of air assuming same velocity.nThird, 20 iterations per time step is not enough. If the flow does not converge, the values of the drag and lift force could be extremely high. This would result in higher particle velocity leading to negative volume cells.nFourth, it seems something is wrong with your remseshing criteria. Can you show the pannel of remeshing as well as your mesh statistics.n
• YasserSelima
Subscriber
One more thing to add, you might have separation around the sphere as well. Use transition turbulent modeln
• pgalves
Subscriber
Thank you for your quick answers, Array!nI just have some questions about two points you make:n1): I understand that fluent will interpret the circle as a cylinder and not a sphere, but would that impact its motion in 2D? What I mean is, if the depth is very small, the slice from the middle of a sphere and the slice from a cylinder would be similar/same, no?n4) Which criteria would you think could be wrong? I don't understand what you mean by panel of remeshing. I even posted the smoothing and remeshing settings in duplicate, without noticing it! Regarding the mesh statistics, please check the attached pic:nnThanks for the other suggestions, I will look into it!n
• YasserSelima
Subscriber
1) The drag coefficient on a sphere is different than that on a cylinder.n2) Sorry my bad, I looked at the mesh but not the panels. Make the minimum length scale zero. Also press on mesh scale info button and post a screenshot Now I believe these cells are coming from the sizing function. Your sizing function makes the new mesh sizes similar to the smallest one. And the change is linear .. so, all remeshed cells becomes very small. Put a positive number in variation and rate .. say 0.1. Or disable the resizing function
• pgalves
Subscriber
Thank you nHere's a screenshot of the mesh scale info. nRegarding the sizing function, I intended the change to be linear and the remeshed cells to remain small, and did not want to have large cells away from moving circle. You think this is bad? Why?Thank you again for the discussion.nBest regardsn
• YasserSelima
Subscriber
You ask Fluent to create mesh that is finer than the original ones.nMake the minimum length scale 0, the max 0.07 or so ... and the skewedness 0.49, similar to the original mesh .nsizing function, put small positive numbers and tryn