Determining Stabilization Damping factor + Rikz analysis problems

[ihab_elaghoury]

Hi everybody,

I've modeled a simple steel beam that I've tested experimentally with 4-point loading. I've inputted the material exactly as the tested coupons. I've included Geometrical non-linearity. I'm measuring the response (Load-deflection and Load-Strains). Own weight is NOT included. I ran the analysis using two different methods :


1- GENERAL STATIC:
------------------
I've run the analysis using automatic stabilization using various damping factors until I reached a value of 0.0011 (without adaptive Stabilization) that gave nearly the same response as experimental (In the plastic part). Changing the damping factor always SIGNIFICANTLY changed the Plastic (non-linear) part of the response.

* So , I was wondering how would I accurately determine the damping factors of any other model if I don't have any experimental data to verify it and calibrate it.

===============================================================

2- STATIC RIKZ :
----------------
I've run the analysis with very small arc length and minimum increment, but always the analysis stops (for many trials) at a load of about half the failure load. I don't know how to fix it. When I removed Geometrical non-linearity, the analysis continues and gives a higher response than experimental. There are no other calibration options in the RIKZ analysis.

* So , I was wondering why is the big difference between including Geometrical non-linearity and not including it in Rikz analysis




In general, I need to know if my analysis is going in the right direction or not.


Thank you so much...

[Nicholas Fantuzzi]

Hi all,
I'm workin' on a dynamic analysis of a rock fall protection net. I  
made the model with autocad 3D and import as a IGS file. I assembled  
the model with all the net's parts. I added some connections, and  
finally put a predefined field as rock impact.
The analysis starts but it never ends because the PC becomes out of  
memory. How can I solve the problem?

Please answer soon I need to solve this problem!

Thanks

Nicholas

[Non-text portions of this message have been removed]

[Milton Deherrera]

It sounds like you are running under Windows.  Is your memory limit set to
the max (2 or 3 Gb, I forget the number for Windows)?  If it is, the only
solution is to run on a machine with more memory, or install Linux on your
PC and provide it with extra RAM, as Linux is not memory limited AFAIK. I
think some PC's allow a dual boot option, so you wouldn't have to initialize
the hard drive if you installed LINUX

On Wed, Oct 14, 2009 at 5:30 AM, Nicholas Fantuzzi <sniis84@...>wrote:

>
>
> Hi all,
> I'm workin' on a dynamic analysis of a rock fall protection net. I
> made the model with autocad 3D and import as a IGS file. I assembled
> the model with all the net's parts. I added some connections, and
> finally put a predefined field as rock impact.
> The analysis starts but it never ends because the PC becomes out of
> memory. How can I solve the problem?
>
> Please answer soon I need to solve this problem!
>
> Thanks
>
> Nicholas
>
> [Non-text portions of this message have been removed]
>
>  
>

[Non-text portions of this message have been removed]

[Nicholas Fantuzzi]

Il giorno 15/ott/2009, alle ore 00.31, Milton Deherrera ha scritto:

> It sounds like you are running under Windows. Is your memory limit  
> set to
> the max (2 or 3 Gb, I forget the number for Windows)? If it is, the  
> only
> solution is to run on a machine with more memory, or install Linux  
> on your
> PC and provide it with extra RAM, as Linux is not memory limited  
> AFAIK. I
> think some PC's allow a dual boot option, so you wouldn't have to  
> initialize
> the hard drive if you installed LINUX
>

Thanks so much Milton for your quick answer!
The CAE file runs on two different machines: the first is a Quad Core  
with 3,2GB of RAM, the second one is a Pentium 4 HT 3,20GHz with 2 GB  
of RAM.
The last chance is to try in a linux PC.Which Linux version should I  
install? Even though I think that memory limit is given by the  
hardware of the machine. For example if you have a 32bit system your  
memory can go over 2GB and if you a have a 64bit system you can  
install till 16GB RAM (if you have the money!!).
I have a PC with Ubuntu Linux but I asked 3ds and they told me that  
Abaqus isn't certified with Ubuntu so probably it doesn't work, so I  
have to use other versions, like RedHad, etc...
Could you please give me an advice on the Linux version that I have to  
use?

Thanks so much again!

Nicholas

[Non-text portions of this message have been removed]

[Milton Deherrera]

ABAQUS supports RedHat and SuSE, and other subsets like Fedora (since Fedora
is a version of RedHat). Quad cores should support 64-bit Linux, don't know
if the Pentium is limited to 32-bit. As far as the version of Linux, look in
http://www.simulia.com/support/v69/v69_platforms.html for version 6.9, and
there are other links for earlier versions.

If you're not using user subroutines, I would try installing it under Ubuntu
and see if it works for what you need.

On Wed, Oct 14, 2009 at 11:17 PM, Nicholas Fantuzzi <sniis84@...>wrote:

>
>
> Il giorno 15/ott/2009, alle ore 00.31, Milton Deherrera ha scritto:
>
>
> > It sounds like you are running under Windows. Is your memory limit
> > set to
> > the max (2 or 3 Gb, I forget the number for Windows)? If it is, the
> > only
> > solution is to run on a machine with more memory, or install Linux
> > on your
> > PC and provide it with extra RAM, as Linux is not memory limited
> > AFAIK. I
> > think some PC's allow a dual boot option, so you wouldn't have to
> > initialize
> > the hard drive if you installed LINUX
> >
>
> Thanks so much Milton for your quick answer!
> The CAE file runs on two different machines: the first is a Quad Core
> with 3,2GB of RAM, the second one is a Pentium 4 HT 3,20GHz with 2 GB
> of RAM.
> The last chance is to try in a linux PC.Which Linux version should I
> install? Even though I think that memory limit is given by the
> hardware of the machine. For example if you have a 32bit system your
> memory can go over 2GB and if you a have a 64bit system you can
> install till 16GB RAM (if you have the money!!).
> I have a PC with Ubuntu Linux but I asked 3ds and they told me that
> Abaqus isn't certified with Ubuntu so probably it doesn't work, so I
> have to use other versions, like RedHad, etc...
> Could you please give me an advice on the Linux version that I have to
> use?
>
> Thanks so much again!
>
> Nicholas
>
> [Non-text portions of this message have been removed]
>
>  
>

[Non-text portions of this message have been removed]

[Dave Lindeman]

Please explain the test in a little more detail.  Does the beam exceed a
buckling load, or does it just experience plastic deformation?  If the
beam only experiences plastic deformation, then you objective should be
to get the simulation to run WITHOUT stabilization.  If it does buckle,
then stabilization (or the Riks method) may be necessary.  An
alternative might be to run the simulation as a slow transient.

Regards,

Dave

-------------------------
Dave Lindeman
Lead Research Specialist
3M Company
3M Center 235-3F-08
St. Paul, MN 55144
651-733-6383


ihab_elaghoury wrote:

>  
>
> Hi everybody,
>
> I've modeled a simple steel beam that I've tested experimentally with
> 4-point loading. I've inputted the material exactly as the tested
> coupons. I've included Geometrical non-linearity. I'm measuring the
> response (Load-deflection and Load-Strains). Own weight is NOT included.
> I ran the analysis using two different methods :
>
> 1- GENERAL STATIC:
> ------------------
> I've run the analysis using automatic stabilization using various
> damping factors until I reached a value of 0.0011 (without adaptive
> Stabilization) that gave nearly the same response as experimental (In
> the plastic part). Changing the damping factor always SIGNIFICANTLY
> changed the Plastic (non-linear) part of the response.
>
> * So , I was wondering how would I accurately determine the damping
> factors of any other model if I don't have any experimental data to
> verify it and calibrate it.
>
> ===============================================================
>
> 2- STATIC RIKZ :
> ----------------
> I've run the analysis with very small arc length and minimum increment,
> but always the analysis stops (for many trials) at a load of about half
> the failure load. I don't know how to fix it. When I removed Geometrical
> non-linearity, the analysis continues and gives a higher response than
> experimental. There are no other calibration options in the RIKZ analysis.
>
> * So , I was wondering why is the big difference between including
> Geometrical non-linearity and not including it in Rikz analysis
>
> In general, I need to know if my analysis is going in the right
> direction or not.
>
> Thank you so much...
>
>

[ihab_elaghoury]

Hi Dave,

Thank you for your reply. It is a static loading for a simple beam at two points till failure (a very simple test). The actual test at failure has a little bit of local buckling in the compression flange (at the late stages of the test).

1- My problem is how to calibrate to this damping if I don't have experimental results (I'm 95% sure of the experimental results).

2- When I did the Static analysis WITHOUT stabilization + NL-GEOM ON -> gave nearly the same behaviour as : RIKS analysis WITHOUT NL-GEOM  which both of them give a higher load than the Experimental.

3- If I want to capture the post failure Load reduction part, I need RIKS analysis, which doesn't complete with NL-GEOM ON. How could I increase the number of trials per increment. Note that RIKS analysis doesn't have any stabilization options.

4- How to do a slow transient analysis, what do you mean by it ?

Please help me to understand the above,

Thank you so much

 Yours,
 
Ihab El-Aghoury

Research and Teaching Assistant
Department of Building, Civil & Environmental Engineering
Concordia University
1515 Rue Ste-Catherine West, EV 11.111
Montreal, Quebec, Canada H3G 2W1
Tel: [phone] Ext. 7198
Cellular: 514-996-2919




________________________________
From: Dave Lindeman <ddlindeman@...>
To: Abaqus@...
Sent: Fri, October 16, 2009 11:48:05 AM
Subject: Re: [Abaqus] Determining Stabilization Damping factor + Rikz analysis problems

 
Please explain the test in a little more detail.  Does the beam exceed a
buckling load, or does it just experience plastic deformation?  If the
beam only experiences plastic deformation, then you objective should be
to get the simulation to run WITHOUT stabilization.  If it does buckle,
then stabilization (or the Riks method) may be necessary.  An
alternative might be to run the simulation as a slow transient.

Regards,

Dave

------------ --------- ----
Dave Lindeman
Lead Research Specialist
3M Company
3M Center 235-3F-08
St. Paul, MN 55144
651-733-6383

ihab_elaghoury wrote:

>
>
> Hi everybody,
>
> I've modeled a simple steel beam that I've tested experimentally with
> 4-point loading. I've inputted the material exactly as the tested
> coupons. I've included Geometrical non-linearity. I'm measuring the
> response (Load-deflection and Load-Strains) . Own weight is NOT included.
> I ran the analysis using two different methods :
>
> 1- GENERAL STATIC:
> ------------ ------
> I've run the analysis using automatic stabilization using various
> damping factors until I reached a value of 0.0011 (without adaptive
> Stabilization) that gave nearly the same response as experimental (In
> the plastic part). Changing the damping factor always SIGNIFICANTLY
> changed the Plastic (non-linear) part of the response.
>
> * So , I was wondering how would I accurately determine the damping
> factors of any other model if I don't have any experimental data to
> verify it and calibrate it.
>
> ============ ========= ========= ========= ========= ========= ======
>
> 2- STATIC RIKZ :
> ------------ ----
> I've run the analysis with very small arc length and minimum increment,
> but always the analysis stops (for many trials) at a load of about half
> the failure load. I don't know how to fix it. When I removed Geometrical
> non-linearity, the analysis continues and gives a higher response than
> experimental. There are no other calibration options in the RIKZ analysis.
>
> * So , I was wondering why is the big difference between including
> Geometrical non-linearity and not including it in Rikz analysis
>
> In general, I need to know if my analysis is going in the right
> direction or not.
>
> Thank you so much...
>
>

   


      __________________________________________________________________
Be smarter than spam. See how smart SpamGuard is at giving junk email the boot with the All-new Yahoo! Mail.  Click on Options in Mail and switch to New Mail today or register for free at http://mail.yahoo.ca

[Non-text portions of this message have been removed]

[ihab_elaghoury]

Hi Dave,


Thank you for your reply. It is a static loading for a simple beam at two points till failure (a very simple test). The actual test at failure has a little bit of local buckling in the compression flange (at the late stages of the test).

1- My problem is how to calibrate to this damping if I don't have experimental results (I'm 95% sure of the experimental results).

2- When I did the Static analysis WITHOUT stabilization + NL-GEOM ON -> gave nearly the same behaviour as : RIKS analysis WITHOUT NL-GEOM  which both of them give a higher load than the Experimental.

3- If I want to capture the post failure Load reduction part, I need RIKS analysis, which doesn't complete with NL-GEOM ON. How could I increase the number of trials per increment. Note that RIKS analysis doesn't have any stabilization options.

4- How to do a slow transient analysis, what do you mean by it ?

Please help me to understand the above,

Thank you so much

Yours,

Ihab El-Aghoury

Research and Teaching Assistant
Department of Building, Civil & Environmental Engineering
Concordia University
1515 Rue Ste-Catherine West, EV 11.111
Montreal, Quebec, Canada H3G 2W1
Tel: [phone] Ext. 7198
Cellular: 514-996-2919

____________ _________ _________ __
From: Dave Lindeman <ddlindeman@mmm. com>
To: Abaqus@yahoogroups. com
Sent: Fri, October 16, 2009 11:48:05 AM
Subject: Re: [Abaqus] Determining Stabilization Damping factor + Rikz analysis problems

Please explain the test in a little more detail.  Does the beam exceed a
buckling load, or does it just experience plastic deformation?  If the
beam only experiences plastic deformation, then you objective should be
to get the simulation to run WITHOUT stabilization.  If it does buckle,
then stabilization (or the Riks method) may be necessary.  An
alternative might be to run the simulation as a slow transient.

Regards,

Dave

------------ --------- ----
Dave Lindeman
Lead Research Specialist
3M Company
3M Center 235-3F-08
St. Paul, MN 55144
651-733-6383

ihab_elaghoury wrote:

>
>
> Hi everybody,
>
> I've modeled a simple steel beam that I've tested experimentally with
> 4-point loading. I've inputted the material exactly as the tested
> coupons. I've included Geometrical non-linearity. I'm measuring the
> response (Load-deflection and Load-Strains) . Own weight is NOT included.
> I ran the analysis using two different methods :
>
> 1- GENERAL STATIC:
> ------------ ------
> I've run the analysis using automatic stabilization using various
> damping factors until I reached a value of 0.0011 (without adaptive
> Stabilization) that gave nearly the same response as experimental (In
> the plastic part). Changing the damping factor always SIGNIFICANTLY
> changed the Plastic (non-linear) part of the response.
>
> * So , I was wondering how would I accurately determine the damping
> factors of any other model if I don't have any experimental data to
> verify it and calibrate it.
>
> ============ ========= ========= ========= ========= ========= ======
>
> 2- STATIC RIKZ :
> ------------ ----
> I've run the analysis with very small arc length and minimum increment,
> but always the analysis stops (for many trials) at a load of about half
> the failure load. I don't know how to fix it. When I removed Geometrical
> non-linearity, the analysis continues and gives a higher response than
> experimental. There are no other calibration options in the RIKZ analysis.
>
> * So , I was wondering why is the big difference between including
> Geometrical non-linearity and not including it in Rikz analysis
>
> In general, I need to know if my analysis is going in the right
> direction or not.
>
> Thank you so much...
>
>

____________ _________ _________ _________ _________ _________ _
Be smarter than spam. See how smart SpamGuard is at giving junk email the boot with the All-new Yahoo! Mail.  Click on Options in Mail and switch to New Mail today or register for free at http://mail. yahoo.ca

[Non-text portions of this message have been removed]


   


      __________________________________________________________________
Yahoo! Canada Toolbar: Search from anywhere on the web, and bookmark your favourite sites. Download it now
http://ca.toolbar.yahoo.com.

[Non-text portions of this message have been removed]

[Dave Lindeman]

1) I'm note sure that you can calibrate the stabilization damping factor
-- it may be problem dependent.  However, if establish correlation to
existing experimental data, and the energy associated with the
stabilization is small compared to the other energies in the system,
then, for SIMILAR problems, you should have increased confidence that
the stabilization damping value will give reasonable results (but you
must always check the energy).

4) Use *DYNAMIC rather than *STATIC (or *VISCO).

Regards,

Dave

-------------------------
Dave Lindeman
Lead Research Specialist
3M Company
3M Center 235-3F-08
St. Paul, MN 55144
651-733-6383


Ihab El Aghoury wrote:

>  
>
> Hi Dave,
>
> Thank you for your reply. It is a static loading for a simple beam at
> two points till failure (a very simple test). The actual test at failure
> has a little bit of local buckling in the compression flange (at the
> late stages of the test).
>
> 1- My problem is how to calibrate to this damping if I don't have
> experimental results (I'm 95% sure of the experimental results).
>
> 2- When I did the Static analysis WITHOUT stabilization + NL-GEOM ON ->
> gave nearly the same behaviour as : RIKS analysis WITHOUT NL-GEOM which
> both of them give a higher load than the Experimental.
>
> 3- If I want to capture the post failure Load reduction part, I need
> RIKS analysis, which doesn't complete with NL-GEOM ON. How could I
> increase the number of trials per increment. Note that RIKS analysis
> doesn't have any stabilization options.
>
> 4- How to do a slow transient analysis, what do you mean by it ?
>
> Please help me to understand the above,
>
> Thank you so much
>
> Yours,
>
> Ihab El-Aghoury
>
> Research and Teaching Assistant
> Department of Building, Civil & Environmental Engineering
> Concordia University
> 1515 Rue Ste-Catherine West, EV 11.111
> Montreal, Quebec, Canada H3G 2W1
> Tel: [phone] Ext. 7198
> Cellular: 514-996-2919
>
> ____________ _________ _________ __
> From: Dave Lindeman <ddlindeman@mmm. com>
> To: Abaqus@yahoogroups. com
> Sent: Fri, October 16, 2009 11:48:05 AM
> Subject: Re: [Abaqus] Determining Stabilization Damping factor + Rikz
> analysis problems
>
> Please explain the test in a little more detail. Does the beam exceed a
> buckling load, or does it just experience plastic deformation? If the
> beam only experiences plastic deformation, then you objective should be
> to get the simulation to run WITHOUT stabilization. If it does buckle,
> then stabilization (or the Riks method) may be necessary. An
> alternative might be to run the simulation as a slow transient.
>
> Regards,
>
> Dave
>
> ------------ --------- ----
> Dave Lindeman
> Lead Research Specialist
> 3M Company
> 3M Center 235-3F-08
> St. Paul, MN 55144
> 651-733-6383
>
> ihab_elaghoury wrote:
>  >
>  >
>  > Hi everybody,
>  >
>  > I've modeled a simple steel beam that I've tested experimentally with
>  > 4-point loading. I've inputted the material exactly as the tested
>  > coupons. I've included Geometrical non-linearity. I'm measuring the
>  > response (Load-deflection and Load-Strains) . Own weight is NOT
> included.
>  > I ran the analysis using two different methods :
>  >
>  > 1- GENERAL STATIC:
>  > ------------ ------
>  > I've run the analysis using automatic stabilization using various
>  > damping factors until I reached a value of 0.0011 (without adaptive
>  > Stabilization) that gave nearly the same response as experimental (In
>  > the plastic part). Changing the damping factor always SIGNIFICANTLY
>  > changed the Plastic (non-linear) part of the response.
>  >
>  > * So , I was wondering how would I accurately determine the damping
>  > factors of any other model if I don't have any experimental data to
>  > verify it and calibrate it.
>  >
>  > ============ ========= ========= ========= ========= ========= ======
>  >
>  > 2- STATIC RIKZ :
>  > ------------ ----
>  > I've run the analysis with very small arc length and minimum increment,
>  > but always the analysis stops (for many trials) at a load of about half
>  > the failure load. I don't know how to fix it. When I removed Geometrical
>  > non-linearity, the analysis continues and gives a higher response than
>  > experimental. There are no other calibration options in the RIKZ
> analysis.
>  >
>  > * So , I was wondering why is the big difference between including
>  > Geometrical non-linearity and not including it in Rikz analysis
>  >
>  > In general, I need to know if my analysis is going in the right
>  > direction or not.
>  >
>  > Thank you so much...
>  >
>  >
>
> ____________ _________ _________ _________ _________ _________ _
> Be smarter than spam. See how smart SpamGuard is at giving junk email
> the boot with the All-new Yahoo! Mail. Click on Options in Mail and
> switch to New Mail today or register for free at http://mail. yahoo.ca
>
> [Non-text portions of this message have been removed]
>
> __________________________________________________________
> Yahoo! Canada Toolbar: Search from anywhere on the web, and bookmark
> your favourite sites. Download it now
> http://ca.toolbar.yahoo.com. <http://ca.toolbar.yahoo.com.>
>
> [Non-text portions of this message have been removed]
>
>