Learning Forum

[glouvincs01]

Hi,

I'm a simulating a half sine wave shock on a mechanical structure. I have two two rigids parts linked by a frictional contact. One is fixed. The other one is allowed to move in one direction. 

I really don't understand why i don't observe any damping and why the displacement continues to increase.

Thank you for your help

[peteroznewman]

You will have to spend a bit more time to explain your model, show the geometry, the boundary conditions and loads, the details of the half sine load, where it is applied, is it a force, displacement, velocity or acceleration load? What damping did you use? Details of the Frictional Contact.  How is the normal force developed so there is some frictional force? Also, plot Directional Deformation and not Total Deformation.

 

[glouvincs01]

Hi Peter, thank a lot you for your help.

 

The simplified geometry is :

 

- A rigid black body. 

-A flexible yellow body 

- A  flexible blue body which links the yellow body and the green body. 

 

Boundary conditions:

- Black body is fixed 

-frictional contact between black and yellow (2)

-frictional contact between black and green (1)

- green body: only y direction is allowed. 

 

For the the contacts, iIused augmented lagrange contact

The system has Elastic materials 

 

Shock: 

It is defined as an acceleration load on the whole system in a transient analysis along the y direction. 

first step: preload of the blue part (screw), second step : half sine acceleration applied .

Sources of damping?  Numerical damping : 0.1 ; friction damping ? nothing else 

Reaction contact: 

For the top contact (2) , the reaction force along y is :

For the bottom contact (1), i have no reaction force. However, we can see that the penetration is non zero : 

The direction deformation (y) of the green part is :

 

My main question is :

-Why does the displacement continue to increase? I would expect a decay of the oscillations when the green part touches back the initial contact. 

Thank a lot in advance for help and i am here if some information are still unclear. 

[peteroznewman]

Please run a Modal Analysis on the model. What is the first natural frequency?

[glouvincs01]

the first natural frequency is around 2000 Hz from the modal analysis 

[peteroznewman]

Please insert the image of the Modal Mass Participation Summary, which is one of the choices under the Solution Information folder.

I am interested in which mode has the most mass participation in the Y direction and what that frequency is.

[peteroznewman]

The half sine shock has a duration of 8e-4 s.  A half period of the first natural frequency is 3.125e-4 s.

It is recommended that a shock duration be less than the half period of the first natural frequency. Your shock is more than a half period. Can you use a shorter shock duration?

You should add some damping to the system. If you change from Direct Input, you can input a Damping Ratio instead of coefficients.  Use a Damping Ratio of 0.1 at the frequency of 1600 Hz.

[glouvincs01]

Here we have a summary. Only one mode is predominant (on DOF is allowed :y ) 

[glouvincs01]

The half sine shock has a duration of 8e-4 s.  A half period of the first natural frequency is 3.125e-4 s.

It is recommended that a shock duration be less than the half period of the first natural frequency. Your shock is more than a half period. Can you use a shorter shock duration?

You should add some damping to the system. If you change from Direct Input, you can input a Damping Ratio instead of coefficients.  Use a Damping Ratio of 0.1 at the frequency of 1600 Hz.

Yes I can.. I wanted to observe the effect of  the shock duration on the maximum amplfification (max displacement of the green part). I have several studies with different shock durations. 

How can you estimate the damping through the contact ? In general,  was it the damping ratio ? 

What do you think bout the bounce on the contact 2 ? What can be the problem ? 

Thank you Sir 

[peteroznewman]

It's not a problem, it's physics. All you need to do is to understand the physics.

Drag out two more Modal analysis systems and drop them onto the project page. 
Suppress the black body and make the green body have only y direction allowed. 

• On one Modal analysis, put the fixed support on the green body. What is the first natural frequency of the yellow body?


• On the other Modal analysis, put the fixed support on the yellow body. What is the first natural frequency of the green body?

In this model, it doesn't look like there is much sliding of the frictional surfaces. Do you get the same response if you make the contacts frictionless?  If so, the friction is irrelevant.

The contact interface opening and closing does not absorb energy the way sliding frictional contact does. That is why you should have system damping or material damping if you want to see the oscillations die down.

If you do not want to see the oscillations get larger without damping, then you need to follow all the rules in the Energy and Momentum Conserving Contact chapter described in the ANSYS Help system.

https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v201/en/ans_ctec/ctecEnMom.html

Here are instructions for how to use the above URL.

[glouvincs01]

Sir, again thanks a lot for your help. 

On one Modal analysis, put the fixed support on the green body. What is the first natural frequency of the yellow body?

-> The frequency is very high: 77000 Hz

On the other Modal analysis, put the fixed support on the yellow body. What is the first natural frequency of the green body?

-> The frequency is also very high: 12000 Hz

 

In this model, it doesn't look like there is much sliding of the frictional surfaces. Do you get the same response if you make the contacts frictionless?  If so, the friction is irrelevant. 

After the first step ie the bolt pretension, the yellow part starts to lift (punctual contact). I tried to reduce the frictional coefficient near 0. I observed that the response changed a lot. So i think friction is relevant.  

I would like to estimate the energy lost by this friction contact: for me, we have to evaluate mu*Reaction force* sliding distance. It is correct? 

The contact interface opening and closing does not absorb energy the way sliding frictional contact does. That is why you should have system damping or material damping if you want to see the oscillations die down.

Interesting... I read in the help that numerical damping should be low for problems involving frictional or plasticity dissipation,  0.1 is ok ? 

best regards,

 

 

[peteroznewman]

The full system first natural frequency is 2000 Hz when there is frictional contact which becomes Bonded in a Modal analysis.

I expected that when you suppress the fixed Black body and the Green and Yellow bodies are only connected by the Bolt, that the frequency would be lower.  Perhaps I didn't explain the situation well enough.  You have two masses a Green body and a Yellow body, connected by a spring (bolt without pre-tension).

Did you include the pre-tension load so that the Green and Yellow bodies make contact with each other?  That is not what I meant.  I wanted to know if the bolt head is bonded to one body and the bolt nut is bonded to the other body, the bolt shank is the spring that connects the two bodies.

A mass-spring system has a natural frequency. One mass might be heavier than the other mass, so depending on which one is Fixed to ground, there is a different frequency.

[glouvincs01]

Dear Peter, thank you for your help. 

To determine the resonance frequency, I used a simple single degree of freedom system. ma+kx-fo=0  with f0 the preload. 

In my model:

• the mass is the green part 


• the spring is the blue and the yellow parts. 


• the black part is fixed 

Then the green part is puled along y- direction to determine the rigidity of the system. I probe a vertex on the green part and iI measure the force-displacement .

 

Then I apply the classical formula. f0=1/2pi * (k/M)^0.5 to get the natural frequency. 

M is the mass of the green part which is large compare to the other parts. 

Is this approach seem correct for you? 

During the transient analysis, I am trying to reproduce the shock response spectrum of a half-cycle sine.  So, I have a lot of simulations with the same acceleration amplitude but different pulse times. However, when I compare the response curve with research papers, I don't find the same resonance frequency that I computed during the static analysis. 

I don't use a modal analysis because it seems that the rigidity becomes very high because of the frictional contacts. The boundary conditions are clearly not the good ones.  Thus, it impossible to apply the pre-load, and then moving the green part in the analysis, because the preload must have the frictional contact 1 when the y is released during the modal analysis. We have an opened/closed contact.