Bi conjugate gradient solver did not converge !

johnsherjy · July 17, 2018, 07:29

Hi

I am trying to simulate flow over blades of a helicopter. I get the following error

Bi conjugate gradient solver did not converge !
* CFL 50 -> 5...
Correction limited to 13 cells in subtract-overset region

I am using LES technique to simulate the flow with overset mesh technique for meshing with Cell count 15 Million cells.. Could anyone please help me out why does this occur and solution to this problem...thanks in advance

me3840 · July 18, 2018, 02:08

Are you sure you're going to be getting a well resolved LES solution on merely 15 million cells?

Do you have to use overset? I'm sure you're going to get diffusion and reflection issues. Why not just use sliding mesh for something easy like helicopter blades?

This error primarily suggests you have mesh or boundary condition issues. You're going to have to describe your case in greater detail for more help.

johnsherjy · July 18, 2018, 02:26

technique :LES method
airfoil : NACA 23012
Implicit unsteady with time size of 7.861e-4 ie 2deg rotation .
The helicopter is rotating at 424rpm at hover configuration.
Boundary condition
The helicopter blades are given no slipwall BC
The outer cubical fluid domain-All the 6 sides are given as slip wall condition

me3840 · July 18, 2018, 02:34

For something like this it's probably better to base your timestep on the blade passing frequency instead of an angle.

Your domain will have to be excessively large to allow the acoustic waves to damp if you're going to use slip walls, I'd guess.

How much of the inertial range are you intending on resolving?

Review your mesh quality, LES is heavily, heavily dependent on a good mesh.

johnsherjy · July 19, 2018, 03:11

My chord length =0.0656m and rotor diameter=2m... The domain size 9*9*9m. I think it is sufficiently large... Could you explain why blade passing frequency should be used?... i all the simulation i have seen they have used 1 or 2deg rotation as timestep..

me3840 · July 19, 2018, 09:46

Because choosing an angle ignores that there's important physics happening at a certain rate. Especially for an LES calculation where you expect frequency content to be very important, that should factor in to your timestep choice.

If you believe your setup is valid it's probably a good idea to show your mesh.

johnsherjy · July 19, 2018, 11:14

This is the mesh i have used for the simulation

me3840 · July 21, 2018, 12:18

If you wish to continue using overset (which I don't recommend, but anyway), you need to make sure that the overset mesh and background mesh are nearly the same size in the area where they interface.

It would be a good idea to re-read the overset best practices in the user guide.

Can you post images of the mesh closer to the blades?

johnsherjy · July 21, 2018, 12:44

This is the mesh near to the blade..

me3840 · July 21, 2018, 13:02

Yes, your mesh is very clearly the problem. You need to lower the prism height by a lot and get them to stop collapsing, in addition to fixing the transition from the overset to the background as stated earlier.

Have you reviewed the length scales of your physics to give you an idea of how the mesh should be sized?

johnsherjy · July 21, 2018, 13:17

Thank you for your response. I will try to change the prism layer height and get back to you. The transition bw overset and background layer is done by using a overlap region

johnsherjy · July 26, 2018, 07:14

Hi

could you explain in detail what does blade passage frequency actually mean? and its significance in rotating blades?

me3840 · July 31, 2018, 15:55

The blade passage frequency is the frequency at which blades move past a fixed point in space.

So a single blade rotating at 10 RPM has a blade passing frequency of 0.167 Hz.
But two blades on a rotor at the same rate has double the frequency.

A fourier transform of a point in space near the blades will reveal this easily. Therefore the flow has a tonal content which is generated primary from the passing frequency and its harmonics. If you are interested in a scale resolving simulation like LES, this content could be of great importance.

johnsherjy · July 31, 2018, 22:43

Thank you for your response!!

Kinng · November 4, 2019, 19:30

Hi,

I am trying to simulate a similar case where my rotor blades are in Hover condition inside a sufficiently large wind tunnel.

I am getting periodic results which is as suggested by the Cd-adapco literature.
But the results that I am looking are about 50-100% higher than the experimental values.

I believe the bi-conjugate gradient solver did not converge is the result of improper meshing. I too faced that error and have overcome it by improving my mesh.

Any comments.

Thank you.

steve lee · November 5, 2019, 00:35

try to use MRF
don't need to make the same size back ground and overset mesh.
interface tech is pretty good.
but the most important thing is boundary mesh if you stick overset, you should make more than 100m meshs for LES overset

July 17, 2018, 07:29	Bi conjugate gradient solver did not converge !	#1
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	Hi I am trying to simulate flow over blades of a helicopter. I get the following error Bi conjugate gradient solver did not converge ! * CFL 50 -> 5... Correction limited to 13 cells in subtract-overset region I am using LES technique to simulate the flow with overset mesh technique for meshing with Cell count 15 Million cells.. Could anyone please help me out why does this occur and solution to this problem...thanks in advance

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Periodic flow using Cyclic - comparison with Fluent	nusivares	OpenFOAM Running, Solving & CFD	30	December 12, 2017 05:35
OpenFOAM: Adding a transient pressure gradient to a solver	dsh5400	OpenFOAM Programming & Development	7	April 7, 2016 15:13
fluent divergence for no reason	sufjanst	FLUENT	2	March 23, 2016 16:08
compressible two phase flow in CFX4.4	youngan	CFX	0	July 1, 2003 23:32
Poisson eq and conjugate gradient method	Gary Dantinne	Main CFD Forum	2	March 19, 1999 00:24

July 18, 2018, 02:08		#2
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Are you sure you're going to be getting a well resolved LES solution on merely 15 million cells? Do you have to use overset? I'm sure you're going to get diffusion and reflection issues. Why not just use sliding mesh for something easy like helicopter blades? This error primarily suggests you have mesh or boundary condition issues. You're going to have to describe your case in greater detail for more help.

July 18, 2018, 02:26		#3
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	technique :LES method airfoil : NACA 23012 Implicit unsteady with time size of 7.861e-4 ie 2deg rotation . The helicopter is rotating at 424rpm at hover configuration. Boundary condition The helicopter blades are given no slipwall BC The outer cubical fluid domain-All the 6 sides are given as slip wall condition

July 18, 2018, 02:34		#4
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	For something like this it's probably better to base your timestep on the blade passing frequency instead of an angle. Your domain will have to be excessively large to allow the acoustic waves to damp if you're going to use slip walls, I'd guess. How much of the inertial range are you intending on resolving? Review your mesh quality, LES is heavily, heavily dependent on a good mesh.

July 19, 2018, 03:11		#5
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	My chord length =0.0656m and rotor diameter=2m... The domain size 999m. I think it is sufficiently large... Could you explain why blade passing frequency should be used?... i all the simulation i have seen they have used 1 or 2deg rotation as timestep..

July 19, 2018, 09:46		#6
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Because choosing an angle ignores that there's important physics happening at a certain rate. Especially for an LES calculation where you expect frequency content to be very important, that should factor in to your timestep choice. If you believe your setup is valid it's probably a good idea to show your mesh.

July 21, 2018, 12:18		#8
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	If you wish to continue using overset (which I don't recommend, but anyway), you need to make sure that the overset mesh and background mesh are nearly the same size in the area where they interface. It would be a good idea to re-read the overset best practices in the user guide. Can you post images of the mesh closer to the blades?

July 21, 2018, 13:02		#10
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Yes, your mesh is very clearly the problem. You need to lower the prism height by a lot and get them to stop collapsing, in addition to fixing the transition from the overset to the background as stated earlier. Have you reviewed the length scales of your physics to give you an idea of how the mesh should be sized?

July 21, 2018, 13:17		#11
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	Thank you for your response. I will try to change the prism layer height and get back to you. The transition bw overset and background layer is done by using a overlap region

July 26, 2018, 07:14		#12
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	Hi could you explain in detail what does blade passage frequency actually mean? and its significance in rotating blades?

July 31, 2018, 15:55		#13
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	The blade passage frequency is the frequency at which blades move past a fixed point in space. So a single blade rotating at 10 RPM has a blade passing frequency of 0.167 Hz. But two blades on a rotor at the same rate has double the frequency. A fourier transform of a point in space near the blades will reveal this easily. Therefore the flow has a tonal content which is generated primary from the passing frequency and its harmonics. If you are interested in a scale resolving simulation like LES, this content could be of great importance.

July 31, 2018, 22:43		#14
johnsherjy Member JOHN SHERJY SYRIAC Join Date: May 2018 Posts: 35 Rep Power: 7	Thank you for your response!!

November 4, 2019, 19:30		#15
Kinng New Member Karthik Join Date: Mar 2019 Posts: 3 Rep Power: 7	Hi, I am trying to simulate a similar case where my rotor blades are in Hover condition inside a sufficiently large wind tunnel. I am getting periodic results which is as suggested by the Cd-adapco literature. But the results that I am looking are about 50-100% higher than the experimental values. I believe the bi-conjugate gradient solver did not converge is the result of improper meshing. I too faced that error and have overcome it by improving my mesh. Any comments. Thank you.

November 5, 2019, 00:35		#16
steve lee Member Steve Join Date: Mar 2018 Location: South Korea Posts: 69 Rep Power: 8	try to use MRF don't need to make the same size back ground and overset mesh. interface tech is pretty good. but the most important thing is boundary mesh if you stick overset, you should make more than 100m meshs for LES overset