different results in 3d and 2d axis symmetric with same geometry

billycrc · November 14, 2023, 04:32

Hi,

I am running two simulations related to drag coefficients of a sphere. One simulation runs in 3d Cartesian with a sphere in a cylinder, the other simulation runs in 2d axis symmetric with a half circle in a rectangle. Both of them have the same geometric relation such as size of sphere, position of sphere and size of cylinder. I apply the same physics (default laminar flow in Comsol) and the same boundary conditions to both of the simulations. Vinlet=0.4m/s, p=0 at outlet, and non-slip wall for sphere.

For drag coefficient
Cd=Cf+Cp where Cf is the friction coefficient and Cp is pressure coefficient

Vinlet is applied in Z direction and I want to derive the coefficients in Z direction.

please visit the link below to see the different results

https://1drv.ms/i/s!AnYEOGuLofRYg88q..._sWBA?e=bFJccj

Theoretically, The two simulations should have the same result.
But the results are quite confusing.
Cpz(3d)=Cpz(2d axis symmetric) => same
Cfz(3d)≠Cfz(2d axis symmetric) => unequal
Cdz(3d)≠Cdz(2d axis symmetric) => unequal

Is there anyone who can explain this to me? How can I derive the same result from axis symmetric geometry?

I can share my two Comsol files if you want.

Thanks a lot

tomchiah · November 21, 2023, 06:06

It seems the results are quite close for me. Even though Cf and Cd with certain discrepancy but it seems acceptable if your error acceptable range is within 20%.

I think you need to refer to some literature that is close to your simulation and judge what is the acceptable range of the error for your case. However, 3D simulation should be more accurate.

November 14, 2023, 04:32	different results in 3d and 2d axis symmetric with same geometry	#1
billycrc New Member Billy Cao Join Date: Nov 2023 Posts: 1 Rep Power: 0	Hi, I am running two simulations related to drag coefficients of a sphere. One simulation runs in 3d Cartesian with a sphere in a cylinder, the other simulation runs in 2d axis symmetric with a half circle in a rectangle. Both of them have the same geometric relation such as size of sphere, position of sphere and size of cylinder. I apply the same physics (default laminar flow in Comsol) and the same boundary conditions to both of the simulations. Vinlet=0.4m/s, p=0 at outlet, and non-slip wall for sphere. For drag coefficient Cd=Cf+Cp where Cf is the friction coefficient and Cp is pressure coefficient Vinlet is applied in Z direction and I want to derive the coefficients in Z direction. please visit the link below to see the different results https://1drv.ms/i/s!AnYEOGuLofRYg88q..._sWBA?e=bFJccj Theoretically, The two simulations should have the same result. But the results are quite confusing. Cpz(3d)=Cpz(2d axis symmetric) => same Cfz(3d)≠Cfz(2d axis symmetric) => unequal Cdz(3d)≠Cdz(2d axis symmetric) => unequal Is there anyone who can explain this to me? How can I derive the same result from axis symmetric geometry? I can share my two Comsol files if you want. Thanks a lot tomchiah likes this.

November 21, 2023, 06:06		#2
tomchiah Member Chiah Zi Jing Join Date: Oct 2023 Location: Xiamen University Malaysia (XMUM) Sepang, Malaysia Posts: 38 Rep Power: 2	It seems the results are quite close for me. Even though Cf and Cd with certain discrepancy but it seems acceptable if your error acceptable range is within 20%. I think you need to refer to some literature that is close to your simulation and judge what is the acceptable range of the error for your case. However, 3D simulation should be more accurate.