Error: Divergence detected in AMG solver: x-momentum

moabdi · December 5, 2015, 12:30

Hi every body

I faced "Error: Divergence detected in AMG solver: x-momentum" when i used UDF. that UDF define velocity for geometry walls.
it's my pleasure if any one could suggest/help for this ERROR.

BEST REGARDS

RTN3000 · December 7, 2015, 16:25

I have two questions for you:

What does your mesh look like?
How many iterations does it take to diverge?

Your error is very open ended, but here are my thoughts:

I have some inlet velocity UDFs that I run that I've seen immediately diverge (at iteration 1) due to a coarse mesh. After making the diverging direction mesh finer, the divergence went away.
If your UDF is assigning a velocity through some coupled calculations in your domain, consider increasing the "Profile Update Interval" to help stabilize it.
From what I've seen, DEFINE_PROFILE is very sensitive to garbage data, so make sure that your UDF is outputting the correct values by using Message0 or printf to help debug.

moabdi · December 15, 2015, 08:35

Hi RNT3000
thanks for your attention

it's accrued at iteration 1.
The UDF has described rotation of a curved pipe (section of TORUS) around it's axial axis, and it's no problem during use it for a simple curved pipe, but when i use this UDF for a series of several curved pipes with different diameter that attached continuously, this ERROR has accrued at iteration 1.
note that i applied UDF for each curved pipe separately.

RTN3000 · December 15, 2015, 17:53

Yeah the main times where I get divergence at iteration 1 are either when:

I don't set the udf correctly as a parameter (ie. when I change the name of the profile or the actual compiled UDF and forget to reassign it)
I have the output of the define_profile macro outputting garbage, causing the solver to immediately error out from infinite residuals
Coarse mesh, in which some cases some of my parameters I'm getting from the faces/cells act weirdly and cause infinite/garbage behavior

For your problem, if you hard code a value into the profile assignment, does the divergence still occur? If it's still diverging, it could be because you're assigning a velocity with the direction being parallel to the face instead of normal to it (negative velocities work fine).

Also, what do you mean by "attached continuously" when you refer to the curved sections?

December 5, 2015, 12:30	Error: Divergence detected in AMG solver: x-momentum	#1
moabdi New Member mohsen Join Date: Jun 2015 Posts: 25 Rep Power: 10	Hi every body I faced "Error: Divergence detected in AMG solver: x-momentum" when i used UDF. that UDF define velocity for geometry walls. it's my pleasure if any one could suggest/help for this ERROR. BEST REGARDS

December 7, 2015, 16:25		#2
RTN3000 New Member Join Date: Sep 2013 Posts: 13 Rep Power: 12	I have two questions for you: What does your mesh look like? How many iterations does it take to diverge? Your error is very open ended, but here are my thoughts: I have some inlet velocity UDFs that I run that I've seen immediately diverge (at iteration 1) due to a coarse mesh. After making the diverging direction mesh finer, the divergence went away. If your UDF is assigning a velocity through some coupled calculations in your domain, consider increasing the "Profile Update Interval" to help stabilize it. From what I've seen, DEFINE_PROFILE is very sensitive to garbage data, so make sure that your UDF is outputting the correct values by using Message0 or printf to help debug.

December 15, 2015, 17:53		#4
RTN3000 New Member Join Date: Sep 2013 Posts: 13 Rep Power: 12	Yeah the main times where I get divergence at iteration 1 are either when: I don't set the udf correctly as a parameter (ie. when I change the name of the profile or the actual compiled UDF and forget to reassign it) I have the output of the define_profile macro outputting garbage, causing the solver to immediately error out from infinite residuals Coarse mesh, in which some cases some of my parameters I'm getting from the faces/cells act weirdly and cause infinite/garbage behavior For your problem, if you hard code a value into the profile assignment, does the divergence still occur? If it's still diverging, it could be because you're assigning a velocity with the direction being parallel to the face instead of normal to it (negative velocities work fine). Also, what do you mean by "attached continuously" when you refer to the curved sections?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
divergence detected in AMG solver: momentum	sudhakar	FLUENT	41	January 23, 2024 03:56
Error: Divergence detected in AMG solver: pressure correction	wanna88	FLUENT	19	April 6, 2016 02:57
AMG solver divergence;	Mike	FLUENT	17	December 14, 2015 13:01
Divergence detected in AMG solver: pressure correction	xinquanzhoucn	FLUENT	5	July 21, 2014 04:49
Error: Divergence detected in AMG solver: x-momentum/ epsilon/ temperature	bubuchacha	FLUENT	6	February 26, 2013 03:30

December 15, 2015, 08:35		#3
moabdi New Member mohsen Join Date: Jun 2015 Posts: 25 Rep Power: 10	Hi RNT3000 thanks for your attention it's accrued at iteration 1. The UDF has described rotation of a curved pipe (section of TORUS) around it's axial axis, and it's no problem during use it for a simple curved pipe, but when i use this UDF for a series of several curved pipes with different diameter that attached continuously, this ERROR has accrued at iteration 1. note that i applied UDF for each curved pipe separately.