Atom topic feed | site map | contact | login | Protection des données personnelles | Powered by FluxBB | réalisation artaban

You are not logged in.

- Topics: Active | Unanswered

**Johannes_ACKVA****Member**- From: Ingenieurbüro für Mechanik, DE
- Registered: 2009-11-04
- Posts: 738
- Website

Bonjour,

I don't find how to apply an acceleration to the base of a model as a load in a harmonic analysis.

I could use the "Large-Mass-Method": Add a large mass to the base point, apply a force (instead of an acceleration) with the amount

Force=LargeMass*Acceleration.

Questions:

(1) is there a "acceleration load" in Code-Aster which I did not find?

(2) when using the Large-Mass-Method I use an unconstraint model. If I use modal reduction: with which boundary conditions must I calculate the modal basis?:

Must I fix the degree of freedom which afterwards will be accelerated in the harmonic analysis (2a)? In case (2a) there will be no superposition of the modes which allow a movement of the base point. Or must I leave free this degree of freedom (2b)? In case (2b) the modes contain a rigid-body-mode and the flexible frequencies are different from (2a). At which frequencies do resonances occur? At those of (2a) or (2b)?

Thank You for Your response

Best wishes for the new year to You all of this forum and kind regards,

Johannes_ACKVA

*Last edited by Johannes_ACKVA (2014-01-06 12:38:16)*

Ingenieurbüro für Mechanik

D 91717 Wassertrüdingen / Germany

www.code-aster.de Training & Support for NASTRAN and CODE-ASTER

Offline

Hi Johannes

I have not made this kind of analysis in CA, but with other FE codes.

So, for what it is worth.

The Force = Mass*Acceleration is a good approximation, the better, the larger, the mass of the large mass. Use a mass that is 10000x your object or more and it is fine.

a) You can apply displacement as a load if you wish, simply scale it such that Displacement = -Acceleration/(2*pi*frequency)^2. But, honestly, the force approach works just as good and is the more convenient to implement.

b) About BCs for the mass - as your structure CoG tends to be away from the large mass centre and the large mass is very very inert, it usually does not matter much what you do. Therefore, one usually connects the structure to the large mass using MPC equations and leave the large mass free. A limitation with this approach is that it does not account for deformation of the base, it assumes the base to be rigid. In a sense, what you try to mimic is a test on a electrodynamic- or hydraulic- shaker table.

If you need to account for deformation in the base structure, a whole bunch of approaches open. What applies for seismic analysis, often applies also for shock analysis. Some gentlemen working for NASA have turned this topic inside out. Take a look at the NASA handbook for dynamic envirnomental criteria. http://snebulos.mit.edu/projects/refere … K-7005.pdf

Sincerely

Claes

*Last edited by CLF (2014-01-03 19:55:22)*

Offline

**Dav****Member**- Registered: 2011-06-22
- Posts: 53

Hi,

Do you use DYNA_LINE_HARM function?

What about defining a acceleration in AFFE_CHAR_MECA.

Afterwards, you realise the calculation of an elementary vector based on the mechanical characteristic and more precisely base on you boundary conditions.

(CALC_VECT_ELEM).

Then you assemble with ASSE_VECT

Finally, in DYNA_LINE_HARM, you call this assembled vector in the keyword EXCIT.

Regards,

Dav

Offline

**Johannes_ACKVA****Member**- From: Ingenieurbüro für Mechanik, DE
- Registered: 2009-11-04
- Posts: 738
- Website

Dav wrote:

What about defining a acceleration in AFFE_CHAR_MECA

Dav, can You please try to do that?,! and then explain me

Johannes_ACKVA

Ingenieurbüro für Mechanik

D 91717 Wassertrüdingen / Germany

www.code-aster.de Training & Support for NASTRAN and CODE-ASTER

Offline

**lagrange****Member**- Registered: 2011-11-14
- Posts: 41

Dear Mr. Ackva,

Is it possible to apply a coordinate transformation?

In the simplest case you could clamp those points where you want to apply the prescribed acceleration and apply a gravity force with the negative acceleration to the rest of the structure. In that case you will get the relative motion of the structure with respect to the clamped nodes as result.

Best regards,

Lagrange

Offline

**Johannes_ACKVA****Member**- From: Ingenieurbüro für Mechanik, DE
- Registered: 2009-11-04
- Posts: 738
- Website

Dav,

sorry for my last ironic answer. To be clear now: There is no acceleration load in AFFE_CHAR_MECA

Lagrange,

this is a very surprising and interesting idea, thank You. But I ve now done already all the work using the LargeMassMethod.

Claes,

thank You for this paper which is very extensive and deep, so I hope to have later time to study it. Or do You have a direct answer from this paper for my question (below)?

question (2)

I return to my question (2). I use the LargeMassMethod, and I have only 1 node beeing accelerated (thus, no need to account for deformation in the base). For now, I do not use any modal reduction in DYNA_LINE_HARM. Now, If I want to use modal reduction, which base should be used? Modal basis with fixed LargeMass and modal basis with free LargMass? I understand now that the only difference is that the latter base comprises the additional rigid body mode (1 rigid-body-mode, because the acc acts in 1 direction), and the elastic modes are identic. So is it correct to use the base with the free LargMass?

Johannes_ACKVA

Ingenieurbüro für Mechanik

D 91717 Wassertrüdingen / Germany

www.code-aster.de Training & Support for NASTRAN and CODE-ASTER

Offline

Hi Johannes

As I wrote, a Large Mass is very very inert and as long as the object of analysis has its CoG offset from that of the Large Mass, it does not matter much which way you choose to analyze, i.e. free or with blocked BCs, the end result will nearly be the same.

The only way to cause a difference between Large Mass and BCs is if the rotational dofs are not inert. If it makes you feel better, you can always throw in a large rotational inertia number to cover the rotation dofs.

Both approaches are very simple. I would probably go for the Large Mass approach simply to reduce the risk of involuntarily constraining any dofs when you connect multiple positions to the large mass. Then again, CA, uses Lagrange Multipliers and these may be more forgiving than Nastran RBE2s.

Sincerely

Claes

*Last edited by CLF (2014-01-05 11:57:33)*

Offline

**jeanpierreaubry****Guru**- From: nantes (france)
- Registered: 2009-03-12
- Posts: 3,648

hello Johannes and happy new year

sorry for my last ironic answer. To be clear now: There is no acceleration load in AFFE_CHAR_MECA

of course yes!

with the keyword PESANTEUR

look here

http://www.code-aster.org/forum2/viewtopic.php?id=18760

but i wonder whether this claim holds true for a dynamic anlysis?

lagrange wrote:

Dear Mr. Ackva,

Is it possible to apply a coordinate transformation?

In the simplest case you could clamp those points where you want to apply the prescribed acceleration and apply a gravity force with the negative acceleration to the rest of the structure. In that case you will get the relative motion of the structure with respect to the clamped nodes as result.

Best regards,

Lagrange

jeanpierre aubry

consider reading my book

freely available here https://framabook.org/beginning-with-code_aster/

Offline

**Dav****Member**- Registered: 2011-06-22
- Posts: 53

No problem, i wanted to answer same thing than Jean-Pierre.

You can do with PESANTEUR.

No more thing to add after Jean-Pierre post

Regards,

Dav

Offline

Dear All

When I have used PESANTEUR, it has been to introduce, static, gravity load in fluid-structure analysis.

I doubt it will work for dynamic analysis.

The Large Mass approach works fine. If you do not like it, use displacements and a scaling function to enforce the Acc = -w^2*Disp function.

Sincerely

Claes

Offline

**lagrange****Member**- Registered: 2011-11-14
- Posts: 41

jeanpierreaubry wrote:

hello Johannes and happy new year

sorry for my last ironic answer. To be clear now: There is no acceleration load in AFFE_CHAR_MECA

of course yes!

with the keyword PESANTEUR

look here

http://www.code-aster.org/forum2/viewtopic.php?id=18760but i wonder whether this claim holds true for a dynamic anlysis?

lagrange wrote:Dear Mr. Ackva,

Is it possible to apply a coordinate transformation?

In the simplest case you could clamp those points where you want to apply the prescribed acceleration and apply a gravity force with the negative acceleration to the rest of the structure. In that case you will get the relative motion of the structure with respect to the clamped nodes as result.

Best regards,

Lagrangejeanpierre aubry

I cannot see any reason why it sould not work.

For example let us consider a two degree of freedom system:

m_1 \ddot{u}_1 + d_{12} (\dot{u}_1 - \dot{u}_2) + s_{12} (u_1 - u_2) = p_1

u_1 ... the unknown displacement

\ddot{u}_2 ... the known acceleration

One coud substitude:

\Delta u = u_1 - u_2

\Delta\dot{u} = \dot{u}_1 - \dot{u}_2

\Delta\ddot{u} = \ddot{u}_1 - \ddot{u}_2

Finally one gets an equation for the relative displacement $\Delta{u}$:

m_1 \Delta\ddot{u} + d_{12} \Delta\dot{u} + s_{12} \Delta{u} = p_1 - m_1 \ddot{u}_2

Best regards,

Lagrange

Offline

**sb1966****Member**- Registered: 2015-03-16
- Posts: 209

Hi,

Greetings to all.

From this, and other related posts, (http://www.code-aster.org/forum2/viewtopic.php?id=20013 and http://www.code-aster.org/forum2/viewto … ?id=20024) it is clear that:

1. Large mass method is suitable for harmonic loading of known frequencies.

2. The method proposed by Lagrange, using PESANTEUR might be easier to implement and better suited for applying recorded earthquake motion to particular node(s) of a structure.

3. This will be helpful for transient analysis directly in time domain, a must for nonlinear analysis.

Now I have the following questions:

1. How to apply PESANTEUR along with the operator DYNA_NONLINE ? ( I am assuming that this is the right operator for this purpose).

2. How can I apply the discrete acceleration record stored in a data file on the model ?

3. Do I need to use some other operator for this purpose in lieu of DYNA_NONLINE ?

Please excuse me if my questions are too elementary as I am relatively new to Code_Aster.

Regards and Thanks.

Sincerely

Sukumar

*Last edited by sb1966 (2016-05-24 10:42:57)*

Offline