XANSYS Forum Index
    Register    

FAQ    Search    Memberlist    Usergroups    SmartFeedSmartFeed    Profile    Log in
[Xansys] {Transient Strutural} Time step dependent solutions
 
This forum is locked: you cannot post, reply to, or edit topics.   This topic is locked: you cannot edit posts or make replies.    XANSYS Forum Index -> XANSYS
Author Message
ayo.brimmo
User


Joined: 01 Jul 2013
Posts: 38
Location: Masdar City, Abu Dhabi

PostPosted: Thu Feb 12, 2015 4:23 pm  Reply with quote

Dear All,

I developed a model to estimate the Q factor (measure of the energy loss) of a damped cantilever. I did this in two ways: Using the Fluid structural Interaction (FSI) and the FLUID136 elements on Mechanical. Everything seems to be okay except that the solution (Q factor) increases significantly -- by about 100%-- as I continuously half the time step. I expected to see some convergence in time step but this is not the case. Both models -- FSI and FLUID136-- have similar results at a given time step and show this strong dependence on time step trend.

As I have a fixed simulation time (1e-2s), reducing the time step increases the simulation time and I have now hit a brick wall where my simulation takes close to 10days to complete -- am now at a time step of 1e-8s and can't possibly afford to have a run time of more than 10days.

My question is what is the "best practice" approach for a model like this that doesn't attain time step convergence and what could be possible causes of this strong dependence in time step.

Any comment, jest or suggestion would be highly appreciated.

Best,
Ayo Brimmo
Masdar
UAE

+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
_________________
Ayo Brimmo
Research Engineer
Masdar Institute of Science and Technology
UAE.
Back to top
View user's profile Send private message
peter.attar
User


Joined: 21 Oct 2008
Posts: 67

PostPosted: Thu Feb 12, 2015 5:21 pm  Reply with quote

The transient portion of the simulation will change as you decrease the timestep as you are able
to resolve higher frequencies with smaller timesteps..however I believe the Q factor calculations should be done
with the "particular solution" only, e.g. after most of the transient ("homogeneous solution")
has decayed.

Peter

Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Thursday, February 12, 2015 5:11 PM
To: ANSYS User Discussion List
Subject: [Xansys] {Transient Strutural} Time step dependent solutions

Dear All,

I developed a model to estimate the Q factor (measure of the energy loss) of a damped cantilever. I did this in two ways: Using the Fluid structural Interaction (FSI) and the FLUID136 elements on Mechanical. Everything seems to be okay except that the solution (Q factor) increases significantly -- by about 100%-- as I continuously half the time step. I expected to see some convergence in time step but this is not the case. Both models -- FSI and FLUID136-- have similar results at a given time step and show this strong dependence on time step trend.

As I have a fixed simulation time (1e-2s), reducing the time step increases the simulation time and I have now hit a brick wall where my simulation takes close to 10days to complete -- am now at a time step of 1e-8s and can't possibly afford to have a run time of more than 10days.

My question is what is the "best practice" approach for a model like this that doesn't attain time step convergence and what could be possible causes of this strong dependence in time step.

Any comment, jest or suggestion would be highly appreciated.

Best,
Ayo Brimmo
Masdar
UAE

+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
Back to top
View user's profile Send private message
ayo.brimmo
User


Joined: 01 Jul 2013
Posts: 38
Location: Masdar City, Abu Dhabi

PostPosted: Fri Feb 13, 2015 3:48 am  Reply with quote

Dr. Peter,

Thanks a lot for the response. This explanation makes a lot of sense.

Another question: After what point during the transient response is it safe to say the solution is homogenous?

Best,
Ayo Brimmo
Masdar
UAE


-----Original Message-----
From: Xansys [mailto:xansys-bounces@xansys.org] On Behalf Of Attar, Peter J.
Sent: Friday, February 13, 2015 4:21 AM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

The transient portion of the simulation will change as you decrease the timestep as you are able to resolve higher frequencies with smaller timesteps..however I believe the Q factor calculations should be done with the "particular solution" only, e.g. after most of the transient ("homogeneous solution") has decayed.

Peter

Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Thursday, February 12, 2015 5:11 PM
To: ANSYS User Discussion List
Subject: [Xansys] {Transient Strutural} Time step dependent solutions

Dear All,

I developed a model to estimate the Q factor (measure of the energy loss) of a damped cantilever. I did this in two ways: Using the Fluid structural Interaction (FSI) and the FLUID136 elements on Mechanical. Everything seems to be okay except that the solution (Q factor) increases significantly -- by about 100%-- as I continuously half the time step. I expected to see some convergence in time step but this is not the case. Both models -- FSI and FLUID136-- have similar results at a given time step and show this strong dependence on time step trend.

As I have a fixed simulation time (1e-2s), reducing the time step increases the simulation time and I have now hit a brick wall where my simulation takes close to 10days to complete -- am now at a time step of 1e-8s and can't possibly afford to have a run time of more than 10days.

My question is what is the "best practice" approach for a model like this that doesn't attain time step convergence and what could be possible causes of this strong dependence in time step.

Any comment, jest or suggestion would be highly appreciated.

Best,
Ayo Brimmo
Masdar
UAE

+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
_________________
Ayo Brimmo
Research Engineer
Masdar Institute of Science and Technology
UAE.
Back to top
View user's profile Send private message
peter.attar
User


Joined: 21 Oct 2008
Posts: 67

PostPosted: Fri Feb 13, 2015 5:05 am  Reply with quote

The extent of the transient will be dependent on a number of things. You will likely have to do a bit of a convergence study to determine this. As a first stab at it I would wait for at least 5 to 10 periods of the fundamental mode (or whatever mode you expect it to oscillate it given your IC) before trying to start computing the Q factor.

Peter


Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Friday, February 13, 2015 4:48 AM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

Dr. Peter,

Thanks a lot for the response. This explanation makes a lot of sense.

Another question: After what point during the transient response is it safe to say the solution is homogenous?

Best,
Ayo Brimmo
Masdar
UAE


-----Original Message-----
From: Xansys [mailto:xansys-bounces@xansys.org] On Behalf Of Attar, Peter J.
Sent: Friday, February 13, 2015 4:21 AM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

The transient portion of the simulation will change as you decrease the timestep as you are able to resolve higher frequencies with smaller timesteps..however I believe the Q factor calculations should be done with the "particular solution" only, e.g. after most of the transient ("homogeneous solution") has decayed.

Peter

Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Thursday, February 12, 2015 5:11 PM
To: ANSYS User Discussion List
Subject: [Xansys] {Transient Strutural} Time step dependent solutions

Dear All,

I developed a model to estimate the Q factor (measure of the energy loss) of a damped cantilever. I did this in two ways: Using the Fluid structural Interaction (FSI) and the FLUID136 elements on Mechanical. Everything seems to be okay except that the solution (Q factor) increases significantly -- by about 100%-- as I continuously half the time step. I expected to see some convergence in time step but this is not the case. Both models -- FSI and FLUID136-- have similar results at a given time step and show this strong dependence on time step trend.

As I have a fixed simulation time (1e-2s), reducing the time step increases the simulation time and I have now hit a brick wall where my simulation takes close to 10days to complete -- am now at a time step of 1e-8s and can't possibly afford to have a run time of more than 10days.

My question is what is the "best practice" approach for a model like this that doesn't attain time step convergence and what could be possible causes of this strong dependence in time step.

Any comment, jest or suggestion would be highly appreciated.

Best,
Ayo Brimmo
Masdar
UAE

+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
Back to top
View user's profile Send private message
christopher.wright
User


Joined: 17 Jun 2009
Posts: 927

PostPosted: Fri Feb 13, 2015 9:36 pm  Reply with quote

On Feb 12, 2015, at 5:11 PM, Ayoola Brimmo wrote:

Quote:
My question is what is the "best practice" approach for a model
like this that doesn't attain time step convergence and what could
be possible causes of this strong dependence in time step.

I expect that 'best practice' varies a great deal. The Q-factor usage
implies you're dealing with steady-state response of a linear
eigensolution--if that isn't the case I suppose you can go through
the motions, but it gets very complicated when frequencies and
responses and damping all start varying with input loading and
practically everything else, as happens with non-linear solutions.

I've never tried it with fluid elements, but I've used harmonic
analysis to extract the modal damping from the frequency response
diagram. Doing it with time-domain analysis sounds like a grotesque
pain in the ass. If you're not sure of the basics get hold of a good
advanced dynamics book (Thompson or Timoshenko) and read through the
section on damping.

As mentioned, you're probably picking up higher and higher response
contributions from the displacement discontinuity at the start of the
transient.

Christopher Wright P.E. |"They couldn't hit an elephant at
chrisw@skypoint.com | this distance" (last words of Gen.
.......................................| John Sedgwick, Spotsylvania
1864)
http://www.skypoint.com/members/chrisw/



+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
Back to top
View user's profile Send private message
ayo.brimmo
User


Joined: 01 Jul 2013
Posts: 38
Location: Masdar City, Abu Dhabi

PostPosted: Tue Feb 24, 2015 4:05 am  Reply with quote

Peter,

Thanks a lot for your insights. You were right about the reason for time step dependent Q factor -- smaller time steps have a greater capability to resolve higher frequencies. Tried to wait for 30 periods before estimating the Q factor as suggested but as I was using the Log Mean Decrement of the transient response to estimate the damping ratio and then Q factor, the effect of higher frequencies were still evident in my estimated Q factor, as the time step reduced.

However, transforming the transient response to the frequency domain (Using the FFT) helped to resolve the response of each modal frequencies and the Q factor for the fundamental frequency easily converged with the time step. From the response curve, it was easy to see the response of other modal frequencies become more telling as the time step reduces. This proved that the dependence of Q factor with time step I was experiencing was as a result of the smaller time steps being able to resolve higher modes.

Thanks again for your comments.

Best,
Ayo Brimmo
Masdar
UAE



-----Original Message-----
From: Xansys [mailto:xansys-bounces@xansys.org] On Behalf Of Attar, Peter J.
Sent: Friday, February 13, 2015 4:05 PM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

The extent of the transient will be dependent on a number of things. You will likely have to do a bit of a convergence study to determine this. As a first stab at it I would wait for at least 5 to 10 periods of the fundamental mode (or whatever mode you expect it to oscillate it given your IC) before trying to start computing the Q factor.

Peter


Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Friday, February 13, 2015 4:48 AM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

Dr. Peter,

Thanks a lot for the response. This explanation makes a lot of sense.

Another question: After what point during the transient response is it safe to say the solution is homogenous?

Best,
Ayo Brimmo
Masdar
UAE


-----Original Message-----
From: Xansys [mailto:xansys-bounces@xansys.org] On Behalf Of Attar, Peter J.
Sent: Friday, February 13, 2015 4:21 AM
To: ANSYS User Discussion List
Subject: Re: [Xansys] {Transient Strutural} Time step dependent solutions

The transient portion of the simulation will change as you decrease the timestep as you are able to resolve higher frequencies with smaller timesteps..however I believe the Q factor calculations should be done with the "particular solution" only, e.g. after most of the transient ("homogeneous solution") has decayed.

Peter

Peter Attar
Associate Professor
The University of Oklahoma
School of Aerospace and Mechanical Engineering
865 Asp Ave.
Felgar Hall Room 212
Norman, OK 73019-1052
phone:405-325-1749
fax:405-325-1088

________________________________________
From: Xansys [xansys-bounces@xansys.org] on behalf of Ayoola Brimmo [abrimmo@masdar.ac.ae]
Sent: Thursday, February 12, 2015 5:11 PM
To: ANSYS User Discussion List
Subject: [Xansys] {Transient Strutural} Time step dependent solutions

Dear All,

I developed a model to estimate the Q factor (measure of the energy loss) of a damped cantilever. I did this in two ways: Using the Fluid structural Interaction (FSI) and the FLUID136 elements on Mechanical. Everything seems to be okay except that the solution (Q factor) increases significantly -- by about 100%-- as I continuously half the time step. I expected to see some convergence in time step but this is not the case. Both models -- FSI and FLUID136-- have similar results at a given time step and show this strong dependence on time step trend.

As I have a fixed simulation time (1e-2s), reducing the time step increases the simulation time and I have now hit a brick wall where my simulation takes close to 10days to complete -- am now at a time step of 1e-8s and can't possibly afford to have a run time of more than 10days.

My question is what is the "best practice" approach for a model like this that doesn't attain time step convergence and what could be possible causes of this strong dependence in time step.

Any comment, jest or suggestion would be highly appreciated.

Best,
Ayo Brimmo
Masdar
UAE

+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+
+-------------------------------------------------------------+
| XANSYS web - www.xansys.org/forum |
| The Online Community for users of ANSYS, Inc. Software |
| Hosted by PADT - www.padtinc.com |
| Send administrative requests to xansys-mod@tynecomp.co.uk |
+-------------------------------------------------------------+

Post generated using Mail2Forum (http://www.mail2forum.com)
_________________
Ayo Brimmo
Research Engineer
Masdar Institute of Science and Technology
UAE.
Back to top
View user's profile Send private message
Display posts from previous:   
This forum is locked: you cannot post, reply to, or edit topics.   This topic is locked: you cannot edit posts or make replies.    XANSYS Forum Index -> XANSYS
All times are GMT - 7 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You cannot download files in this forum


Powered by phpBB © 2001, 2005 phpBB Group

sleek template created by Andrew Charron