Fatal overflow in linear solver error. Why?

May 30, 2006, 06:31

| ERROR #004100018 has occurred in subroutine FINMES. | | Message: | | Fatal overflow in linear solver.

hi i run my analysis using k-epsilon turbulence model and after 9 iterations, above error ocuured. i mesh my model in Gambit and using boundary layer as follows;

-first layer thickness = 0.0001 m -expnasion factor = 1.2 -layer = 20 layer

can anybody give me an idea..

May 30, 2006, 19:20

Hi,

Your simulation has diverged. Look in the documentation under obtaining convergence.

Glenn Horrocks

June 1, 2006, 01:31

this problem happened again to me!Is there any BC not apprcieate for me???i set pressure for inlet,and pressure for outlet ,is it allowed???

June 9, 2006, 10:12

As a quick fix: Did you try adjusting your timestep from automatic to manual?

Divide the recommended auto time step by 10^3 and try it again.

Bdew8556 · August 10, 2016, 09:25

Great response Glenn,

Could you perhaps elaborate on what the physical timescale does?

I am aiming to run a steady flow simulation, what effect does it have in this case?

Cheers (Fellow Australian I see)

Brett

ghorrocks · August 10, 2016, 20:25

Hi Brett,

You certainly have dug up a thread from ancient history here. You are lucky that I am still around

I have not seen the other people on this thread for 10 years....

Physical time scale is the rate at which CFX advances the numerical values in a steady state simulation. There are other ways of doing it used in other software such as under-relaxation, but CFX does it by a physical time scale. So a large time scale advances the values a large amount, a small time scale advances it a small amount. Large time scales can approach convergence quicker, but you can only go so fast before the numerics becomes unstable and the simulation diverges.

Note that the physical time scale in a steady state simulation is not the same as time step size in a transient simulation. A steady state simulation has many transient terms removed as they are zero for steady state, whereas a transient simulation includes all the transient terms.

Bdew8556 · August 11, 2016, 05:26

Thanks Glenn, Lucky indeed.

How does that related to the mesh density, ie the dx,dy,dz values? Is it the increment it adds on from the initial values at each cell?

Brett

ghorrocks · August 11, 2016, 06:59

Have a look at the CFX documentation and a CFD textbook for an answer to that question. The spatial discretisation is handled very differently to the time discretisation (for a transient simulation) or the convergence advancement procedure (for a steady state run). I will not describe the difference on the forum as it is far too complex. Hence see a textbook.

May 30, 2006, 06:31	Fatal overflow in linear solver error. Why?	#1
zaidun Guest Posts: n/a	\| ERROR #004100018 has occurred in subroutine FINMES. \| \| Message: \| \| Fatal overflow in linear solver. hi i run my analysis using k-epsilon turbulence model and after 9 iterations, above error ocuured. i mesh my model in Gambit and using boundary layer as follows; -first layer thickness = 0.0001 m -expnasion factor = 1.2 -layer = 20 layer can anybody give me an idea..

May 30, 2006, 19:20	Re: Fatal overflow in linear solver error. Why?	#2
Glenn Horrocks Guest Posts: n/a	Hi, Your simulation has diverged. Look in the documentation under obtaining convergence. Glenn Horrocks

June 1, 2006, 01:31	Re: Fatal overflow in linear solver error. Why?	#3
prayskyer Guest Posts: n/a	this problem happened again to me!Is there any BC not apprcieate for me???i set pressure for inlet,and pressure for outlet ,is it allowed???

June 9, 2006, 10:12	Re: Fatal overflow in linear solver error. Why?	#4
Brendan Guest Posts: n/a	As a quick fix: Did you try adjusting your timestep from automatic to manual? Divide the recommended auto time step by 10^3 and try it again.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Fatal Overflow when using RNG k-e	Pascal	CFX	2	February 5, 2008 17:41
free C code for large sparse matrix linear solver	ztdep	Main CFD Forum	7	May 24, 2007 15:14
Solver: Fatal Bounds error detected	hagupta	CFX	5	March 24, 2006 11:17
linear solver overflow	peggy	CFX	1	February 8, 2001 02:39
solver for linear system with large sparse matrix	Yangang Bao	Main CFD Forum	1	October 25, 1999 05:22

August 10, 2016, 09:25		#5
Bdew8556 Senior Member Brett Join Date: May 2013 Posts: 217 Rep Power: 14	Great response Glenn, Could you perhaps elaborate on what the physical timescale does? I am aiming to run a steady flow simulation, what effect does it have in this case? Cheers (Fellow Australian I see) Brett

August 10, 2016, 20:25		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Hi Brett, You certainly have dug up a thread from ancient history here. You are lucky that I am still around I have not seen the other people on this thread for 10 years.... Physical time scale is the rate at which CFX advances the numerical values in a steady state simulation. There are other ways of doing it used in other software such as under-relaxation, but CFX does it by a physical time scale. So a large time scale advances the values a large amount, a small time scale advances it a small amount. Large time scales can approach convergence quicker, but you can only go so fast before the numerics becomes unstable and the simulation diverges. Note that the physical time scale in a steady state simulation is not the same as time step size in a transient simulation. A steady state simulation has many transient terms removed as they are zero for steady state, whereas a transient simulation includes all the transient terms.

August 11, 2016, 05:26		#7
Bdew8556 Senior Member Brett Join Date: May 2013 Posts: 217 Rep Power: 14	Thanks Glenn, Lucky indeed. How does that related to the mesh density, ie the dx,dy,dz values? Is it the increment it adds on from the initial values at each cell? Brett

August 11, 2016, 06:59		#8
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Have a look at the CFX documentation and a CFD textbook for an answer to that question. The spatial discretisation is handled very differently to the time discretisation (for a transient simulation) or the convergence advancement procedure (for a steady state run). I will not describe the difference on the forum as it is far too complex. Hence see a textbook.