CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > CFX

Time integration of quantities

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   January 13, 2014, 12:31
Default Time integration of quantities
  #1
HLo
New Member
 
HLo
Join Date: Dec 2013
Location: Germany
Posts: 25
Rep Power: 3
HLo is on a distinguished road
Hello experts,
I was looking for some means for time integration of a quantity to establish some kind of control (PI-control). There are proposed approaches:
1) User fortran
2) some calc with a prevoius value, which is not availabe by default (keyword: Update Loop = TRANS_LOOP)
3) sth. I worked upon using internal equations: Initialization Additional Variables for time integration
x) more ?
As I want to control to a certain steady state, I may tend to do a steady state analysis but in this case, the approaches have disadvantages
1) fortran: I didn't find any hint how to get values of the "false timestep" (or related values) used in steady state analyses to be able to calculate the integral (in fact a accumulated sum of quantity*timestep)
2) trans_loop: the setup for previous value needs a two step setup, and for the aquisition of the timestep value it is the same problem
3) internal equations: solution is more complex, but should yield (more) correct(?) values, as the internal equations' integration underlies the same timestep as the steady state solution itself
What is your opinion?
What are your proposals?
Thanks in advance for upcoming feedback
HLo
HLo is offline   Reply With Quote

Old   January 13, 2014, 17:45
Default
  #2
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,954
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
It will be very difficult to implement a PI type controller inside a steady state simulation. You are going to have convergence difficulties. You are almost certainly going to have to run this transient.

What are you trying to control anyway? Implementing a controller will be tricky and difficult to converge so there might be another way.
ghorrocks is offline   Reply With Quote

Old   January 14, 2014, 03:27
Default PI-Control
  #3
HLo
New Member
 
HLo
Join Date: Dec 2013
Location: Germany
Posts: 25
Rep Power: 3
HLo is on a distinguished road
Hello Glenn,
I investigate some switch setup with different housing and thermal insulating material. What I wanted to achieve is a constant steady state temperature (or at end of trans. run) in the electric current carrying part (some kind of heating) and derive and compare the heating powers needed from PI input. Of course works with a P-only control if one accepts a certain deviation from target temperature (but deviation may evolve different in different setups).

Thanks
HLo
HLo is offline   Reply With Quote

Old   January 14, 2014, 05:04
Default
  #4
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,954
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Why not use a boundary condition to fix the temperature at that region then? Or a source term to fix the temperature in that region? These options are MUCH simpler.
ghorrocks is offline   Reply With Quote

Old   January 14, 2014, 05:20
Default Power
  #5
HLo
New Member
 
HLo
Join Date: Dec 2013
Location: Germany
Posts: 25
Rep Power: 3
HLo is on a distinguished road
Hello Glenn,
thought about that, but then -I assumed- I get no information about the electrical power needed to achieve the same target temperature for different setups.
BTW: the electrical power is input by profile data from an ANSYS electrical calc with a weight factor, the latter PI-controlled
But your remark now takes me to an idea.
I'll gonna check for equivalence: surface integral (areaInt@...) of all fluxes around the heating should be equal to input power in final steady state.
You agree?
... and: what -apart from the actual example- is your opinion to the integration proposals?
... and: is there access to the stead state analysis' "false timesteps" in user fortran?
HLo
HLo is offline   Reply With Quote

Old   January 14, 2014, 05:46
Default
  #6
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,954
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
If the electrical side of things generates a known temperature and that temperature generates (or interacts with) some fluid flow and/or heat transfer then the electrical stuff and the fluid/thermal stuff is decoupled. You can do the electrical stuff in one simulation in the knowledge that it gives a known temperature, and the fluid/thermal stuff can have the known temperature as a boundary condition. This is far simpler than the everything model you propose.

Or is there some reason why this is inappropriate? Maybe the temperature distribution is important?

To answer your specific questions:
1) All 3 methods should work. All are a bit yukky, but should work given perservence. I would give each a go and use whatever looks the most promising after you have tried them.
2) I do not know this, but hopefully it is documented. If not contact ANSYS support. Or you might be lucky and somebody else on the forum knows it.
ghorrocks is offline   Reply With Quote

Old   January 14, 2014, 05:58
Default Temperature distribution
  #7
HLo
New Member
 
HLo
Join Date: Dec 2013
Location: Germany
Posts: 25
Rep Power: 3
HLo is on a distinguished road
Hello Glenn,
thanks for your answers.
Temperature distribution is a point of interest, but maybe in a pre-analysis one may find that temp is spatially almost the same (within 5 .. 10 K tolerance at 500 K). Then I can applay a constant temp and the total heat flow balance at surface "heating" should give the power input.
I'll keep you and the forum on track when I got further.
HLo
HLo is offline   Reply With Quote

Old   January 14, 2014, 06:15
Default
  #8
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,954
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
That sounds much more sensible.

You could consider doing a preliminary analysis with constant temperature, and that gives you the heat flow. Then you can repeat the analysis with the heat flow you just got and that will give you an approximation of the temperature distribution. This would give you a better representation of the temperature distribution, but keep the electrical and fluid/thermal stuff uncoupled.
ghorrocks is offline   Reply With Quote

Old   January 23, 2014, 04:52
Default Further time integration pros and cons
  #9
HLo
New Member
 
HLo
Join Date: Dec 2013
Location: Germany
Posts: 25
Rep Power: 3
HLo is on a distinguished road
some more on time integration pros and cons see Post #21 in: Initialization Additional Variables for time integration

HLo
HLo is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Simulation seems to converge but crashes suddenly xxxx OpenFOAM 16 September 12, 2014 08:07
dynamic Mesh is faster than MRF???? sharonyue OpenFOAM Running, Solving & CFD 14 August 26, 2013 07:47
Orifice Plate with a fully developed flow - Problems with convergence jonmec OpenFOAM Running, Solving & CFD 3 July 28, 2011 05:24
calculation diverge after continue to run zhajingjing OpenFOAM 0 April 28, 2010 04:35
Transient simulation not converging skabilan OpenFOAM Running, Solving & CFD 12 September 17, 2007 17:48


All times are GMT -4. The time now is 15:50.