Transient analysis-help

PYJG · February 11, 2014, 13:33

Hi,

I am working on a high speed flow with shock waves. The flow is turbulent and reaches mach 1 as it reaches the mouth if the expansion region.
I have a steady state solution for the fluid region in the model. Next step is introducing the solid boundary to this fluid region. The interaction is transient. I want to run this model for first 30 seconds. I used time steps of the order of 10^-6 to obtain the steady state solution. I want to efficiently but accurately model the phenomena occurring in the solid region.
a) are there any tips for setting up time steps in transient analysis.If I were to use the same time step as steady state it would take a lot of time.
b)will reducing the mesh size and increasing the time steps help?
So far I have used-
- total time 30 sec, adaptive meshing and tried different minimum time 10e-4, 10e-5, 10e-6 and #of loops are 7
-time steps 0.01 I get Finmes error

I am using high speed model and also introduced laminar to turbulent smoothness factor of 0.1. Please help

ghorrocks · February 11, 2014, 17:22

Time step size for most transient simulations is best set using adaptive time steps, homing in on 3-5 coeff loops per iteration.

Reducing mesh size usually results in smaller time steps being required. But you should set the mesh size from a sensitivity analysis, and then you get the time step size from 3-5 coeff loops per iteration and then you know the simulation run time. A common mistake is to decide how long a simulation should take to run, and then set mesh and time step size accordingly. This is obviously wrong.

Your time step size of 0.01s is obviously too big and causing divergence. (At least for the start of the simulation anyway)

PYJG · February 12, 2014, 17:07

Thanks Glen. That helps.

Another question- I have a polyline that traces the outline of a contour how can I extract the coordinates of the polyline in results?

ghorrocks · February 12, 2014, 17:35

In CFD-Post do File/Export and use the polyline object as the object to export.

PYJG · February 12, 2014, 17:45

Wow! Thanks Glenn. You have been very helpful. Thank you for the wonderful help you provide.

PYJG · February 16, 2014, 21:33

I extracted data from polyline but some points are missing. How do I make sure it does not skip points?

ghorrocks · February 17, 2014, 00:12

The polyline should have a point each time the line cuts a control volume boundary. So if it is missing points then you probably don't have a mesh in that area, or have a weird distorted mesh.

February 11, 2014, 13:33	Transient analysis-help	#1
PYJG New Member Join Date: Oct 2013 Posts: 22 Rep Power: 12	Hi, I am working on a high speed flow with shock waves. The flow is turbulent and reaches mach 1 as it reaches the mouth if the expansion region. I have a steady state solution for the fluid region in the model. Next step is introducing the solid boundary to this fluid region. The interaction is transient. I want to run this model for first 30 seconds. I used time steps of the order of 10^-6 to obtain the steady state solution. I want to efficiently but accurately model the phenomena occurring in the solid region. a) are there any tips for setting up time steps in transient analysis.If I were to use the same time step as steady state it would take a lot of time. b)will reducing the mesh size and increasing the time steps help? So far I have used- - total time 30 sec, adaptive meshing and tried different minimum time 10e-4, 10e-5, 10e-6 and #of loops are 7 -time steps 0.01 I get Finmes error I am using high speed model and also introduced laminar to turbulent smoothness factor of 0.1. Please help

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
FSI Transient Thermal Analysis	Abduri	CFX	1	January 1, 2017 22:54
Transient analysis of a synchronous PM motor	daddo88	Structural Mechanics	0	January 25, 2014 05:52
vortex shedding, transient or steady state analysis type?	alfonsojurado	CFX	0	October 25, 2012 05:33
transient analysis _porous medium	shaswat	CFX	4	January 25, 2012 00:18
transient analysis of silencer using cfx.	mehul	CFX	0	October 16, 2005 04:48

February 11, 2014, 17:22		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,708 Rep Power: 143	Time step size for most transient simulations is best set using adaptive time steps, homing in on 3-5 coeff loops per iteration. Reducing mesh size usually results in smaller time steps being required. But you should set the mesh size from a sensitivity analysis, and then you get the time step size from 3-5 coeff loops per iteration and then you know the simulation run time. A common mistake is to decide how long a simulation should take to run, and then set mesh and time step size accordingly. This is obviously wrong. Your time step size of 0.01s is obviously too big and causing divergence. (At least for the start of the simulation anyway)

February 12, 2014, 17:07		#3
PYJG New Member Join Date: Oct 2013 Posts: 22 Rep Power: 12	Thanks Glen. That helps. Another question- I have a polyline that traces the outline of a contour how can I extract the coordinates of the polyline in results?

February 12, 2014, 17:35		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,708 Rep Power: 143	In CFD-Post do File/Export and use the polyline object as the object to export.

February 12, 2014, 17:45		#5
PYJG New Member Join Date: Oct 2013 Posts: 22 Rep Power: 12	Wow! Thanks Glenn. You have been very helpful. Thank you for the wonderful help you provide.

February 16, 2014, 21:33		#6
PYJG New Member Join Date: Oct 2013 Posts: 22 Rep Power: 12	I extracted data from polyline but some points are missing. How do I make sure it does not skip points?

February 17, 2014, 00:12		#7
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,708 Rep Power: 143	The polyline should have a point each time the line cuts a control volume boundary. So if it is missing points then you probably don't have a mesh in that area, or have a weird distorted mesh.