confusion in definition of SPRING1 value. 1

[jongustav]

I had a post quite recently concerning to find strain energy release in dent growth induced by buckling, I have problems running that/finding a way to get the rates so I am trying another approach with spring elements, SPRING1.
They actually comply alot better with existing theory. Anyway,
In my cofiguration I must have a quite dense mesh to be able to get values as simulating the dent growing, so therefore I use an orphan mesh and then connect the SPRING1 elements to the nodes, now using bias seeding each element as a different area.. would I have to define a K (N/m) for each one of the springs or how does it work?
Also, how is the area the spring ''covers'' defined - 1 spring to 1 element- ?
Hmm quite unclear in the doc.

Thank you very much for your consideration // Jon

 

[bassmanjax]

Hi Jon

I'm not sure I fully understand your problem, but here is something that I have used with spring elements in the past.

You can couple the free node of the spring to a collection of nodes in your mesh using a *COUPLING constraint. All you need to do is collect the mesh nodes into a *SURFACE and set the free node of the spring as the ref node. This allows the spring to exert a force over an area of the mesh (like a distributed load) rather than through a node as a point load.

I use springs quite a lot, but since connector elements were introduced, I tend to use connectors instead of SPRING1 or 2 elements.

Martin

 

[jongustav]

Thanks Martin!

That is a good idea. Anyhow it is not exactly what I need, so i'll redefine the problem.

I'm simulating a plate with an impact damage that is subsequently exposed to a compressive load. So using symmetry, the damage is a quarter circle in one of the corners of the wide plate. This means that when compressing the plate we will see buckling of this circular region (assuming the rest of the bottom area is fixed to ground) which is as expected. Now, instead of fixing the plate to ground I would like to have spring element betwen the plate and ground (SPRING1), this due to that the circular region will in reality start to grow perpendicular to the compressive load once a predefined tensional(plastic, non-linear) value of the spring or whatever core element I use is reached. This means that in using springs at each node along the expected "growing" region, it would be possible to find the energy release countinously as the the circular region is growing/ compressive load increases ( actually I'm using prescribed displacement intead of a pressure load).

Yes, I want to find the energy release rate at the tip of the growing circular region since the criterion I'm investigating is energy based.

(The criterion considers the balance between strain energy released by face sheet buckling and energy consumed in core crushing)

This means that I will have to connect the spring as a point load, as you proposed. But then what will the k-value be defined as, N/m?

This is what I meant by, how is the area that every spring is connected to defined if using "point load/one spring to each node" ?

Hopefully I've brought some light into the problem at hand, and I would very much appreciate any answer/discussion.

// Jon

 

[bassmanjax]

Jon

Yes, the spring constant k will be defined in consistent SI units. Assuming that you are modelling in N,m, sec, then k will be in N/m (load per unit displacement of the spring).

I'd seriously have a look at connector elements, as you can define a whole bunch of other stuff as part of a connector definition, not just stiffness.

As for the load transfer to the mesh, by joining ground to a node using a spring, you are in effect creating a point load at the node (like a *CLOAD in ABAQUS) as you correctly state. The only way to spread or distribute the load over an area is to do as I said previously and couple the free spring node to a *SURFACE.

Could you post a sketch of the problem?

Martin

 

[jongustav]

Martin

Hmm I see, thanks for clearing out the issue with how to assign a "known" area to the spring.

Also I will def. have a look into connector elements, just had a quick view and it looks good!.

Well since I've fibbeled around alot with the cae-inputfile- i woulnd't want to post that, it is also very long.

/ Jon