Problem with Convergence (Over flow)

armlic · July 12, 2014, 11:43

Hello everyone.
I´m trying to simulation the flow in a turbine but I get only over flow problems in the solver.

-Inlet stationary domain (A) then to a rotor (rotary domain B) and finally to the outlet which form part of the same domain (A).
-Multiphase transient. 2D analysis (its 3d but only 1 layer depth)
-65,000 nodes

Inlet 1 - velocity (air)
Opening 1 - Opening Pres and Dirn. (0 Pas.) only for air
Outlet 1 - average static pressure

-I have tried countless times to finish a simulation but I always get this Overflow error. I´m simulating with very small time steps equivalent to every 5 degrees of rotation of the rotor. (timestep: .0029 s) and with a total time of 0.21 s which is the time that takes for only 1 revolution.
max 10 iterations, transient scheme: second order backward euler, Advection scheme: High resolution, convergence criteria RMS 1E-4.

- Since the problem requires to have a water level inside the turbine (below the runner) I incorporated an expression for the volume of fraction for the stationay doman:
Waterlevel=-235mm, WaterVolumefraction= if(y<Waterlevel,1,0)

-I have tried with Double Precision/Normal, Parallel(2,4,8,12)/serial, but always around the same time step (65-75) i get the overflow problem.

Any idea what could be the problem??
(I tried the same mesh before with steady state and worked, and before introducing the volume fraction equations; but i don’t see any problem with them)

Thank you very much in advance!

ghorrocks · July 13, 2014, 08:36

Have you read the FAQ? http://www.cfd-online.com/Wiki/Ansys...do_about_it.3F

What makes you think a 5 degree rotation is a small timestep? It sounds like a huge one to me. Your time step will need to be much smaller than that.

armlic · July 14, 2014, 04:26

Hello ghorrocks,

Thank you! I reduced the time step duration from 5degree/interval (70 intervals) to 4 degree/interval (90 intervals) and it converged in every simulation since.
Regards!

ghorrocks · July 14, 2014, 04:30

You have reduced your time step size from huge to big.

Have you done a sensitivity study of your time step size? You need to make the time step smaller until your results do not change. I would expect this would give you a time step of more like 0.5 degrees or maybe less if you have a fine mesh.

armlic · July 14, 2014, 05:48

Hello ghorrocks
My only further plan was to make a Mesh quality study but since you mentioned it, now it makes total sense to include the timestep duration into it. Thank you for your advice.

July 12, 2014, 11:43	Problem with Convergence (Over flow)	#1
armlic New Member Join Date: Jul 2014 Location: BW, Germany Posts: 9 Rep Power: 12	Hello everyone. I´m trying to simulation the flow in a turbine but I get only over flow problems in the solver. -Inlet stationary domain (A) then to a rotor (rotary domain B) and finally to the outlet which form part of the same domain (A). -Multiphase transient. 2D analysis (its 3d but only 1 layer depth) -65,000 nodes Inlet 1 - velocity (air) Opening 1 - Opening Pres and Dirn. (0 Pas.) only for air Outlet 1 - average static pressure -I have tried countless times to finish a simulation but I always get this Overflow error. I´m simulating with very small time steps equivalent to every 5 degrees of rotation of the rotor. (timestep: .0029 s) and with a total time of 0.21 s which is the time that takes for only 1 revolution. max 10 iterations, transient scheme: second order backward euler, Advection scheme: High resolution, convergence criteria RMS 1E-4. - Since the problem requires to have a water level inside the turbine (below the runner) I incorporated an expression for the volume of fraction for the stationay doman: Waterlevel=-235mm, WaterVolumefraction= if(y<Waterlevel,1,0) -I have tried with Double Precision/Normal, Parallel(2,4,8,12)/serial, but always around the same time step (65-75) i get the overflow problem. Any idea what could be the problem?? (I tried the same mesh before with steady state and worked, and before introducing the volume fraction equations; but i don’t see any problem with them) Thank you very much in advance!

July 13, 2014, 08:36		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Have you read the FAQ? http://www.cfd-online.com/Wiki/Ansys...do_about_it.3F What makes you think a 5 degree rotation is a small timestep? It sounds like a huge one to me. Your time step will need to be much smaller than that. armlic likes this.

July 14, 2014, 04:30		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	You have reduced your time step size from huge to big. Have you done a sensitivity study of your time step size? You need to make the time step smaller until your results do not change. I would expect this would give you a time step of more like 0.5 degrees or maybe less if you have a fine mesh. armlic likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Convergence Problem icoFoam steady flow over an airfoil	Lucas	OpenFOAM Running, Solving & CFD	6	July 12, 2018 16:06
Swirl flow convergence problem with simpleFoam	iqbalsk8	OpenFOAM Running, Solving & CFD	7	November 28, 2012 01:54
convergence problem with compressible flow	S.B.J.	Siemens	2	February 2, 2009 05:05
Convergence problem with compressible flow	S.B.J.	Siemens	2	January 7, 2009 04:29
Problem with Convergence at high flow rates	Syed	Siemens	1	April 10, 2007 16:18

July 14, 2014, 04:26		#3
armlic New Member Join Date: Jul 2014 Location: BW, Germany Posts: 9 Rep Power: 12	Hello ghorrocks, Thank you! I reduced the time step duration from 5degree/interval (70 intervals) to 4 degree/interval (90 intervals) and it converged in every simulation since. Regards!

July 14, 2014, 05:48		#5
armlic New Member Join Date: Jul 2014 Location: BW, Germany Posts: 9 Rep Power: 12	Hello ghorrocks My only further plan was to make a Mesh quality study but since you mentioned it, now it makes total sense to include the timestep duration into it. Thank you for your advice.