FEA Adhesive in a Bonded joint 3

[engroma]

Hi

I am modelling a part in which I have to bond with an adhesive a metal block withstanding a load.
To model the adhesive I figured different options

1) merge the nodes in adhesive areas
2) simulate the adhesive with elements
3) using spring elements between the two adherents
4) to use the laminate propriety of Femap and using like core the metal block

What is the best solution in your opinion? Where can I find some example?or do you have better idea?

Thanks

Regards

Engroma

 

[SWComposites]

It depends. So ... what is the purpose of the model? what results do you want to get out of it? what do you expect to do with the model?

Do you have material properties for the adhesive? does it have a nonlinear stress-strain response?

1) merge the nodes in adhesive areas
- generally won't give desired results

2) simulate the adhesive with elements
- generally the best approach, if required, but there are lots of details

3) using spring elements between the two adherents
- can be done, but generally more difficult than 2)

4) to use the laminate propriety of Femap and using like core the metal block
- not sure exactly what you are proposing, but probably won't give desired results

 

[engroma]

I have got two composite panels joined with metal blocks.
These metal blocks are secondary bonded between 4 carbon plies using an adhesive.
I would like to get the adhesive stress due to the load.
I can get the adhesive proprieties from the certification test we carry out on it.
How many details I need for the number 2?

thanks

 

[GBor]

Here's some basic information:

http://femci.gsfc.nasa.gov/adhesive/index.html

And you may want to check

http://www.materials-sciences.com.

I know they've done quite a bit in adhesives modeling. Their SUBLAM software is pretty impressive in this area.

Garland E. Borowski, PE
Engineering Manager
Star Aviation

 

[MechEng2005]

Would it be possible to use two different models? Model everything on one half of the adhesive and determine the reaction forces at connections to the other half. Then, model the other half using the reaction forces as loads.

I'm guessing that your situation is more complex and this isn't the best solution. I'm certainly not an expert on FEA, but figured I'd toss the idea out there.

-- MechEng2005

 

[BWAppliedKinetics]

By the sounds of it you are interested in the stress levels in the adhesive. If this is the case then it would be a good idea to include the adhesive as a solid. There are various methods to accomplish this depending on what software you are using.

Just be careful, for most element types even thin sections require 2 or 3 elements through the thickness in order to get good stress results.

BW

http://www.akeng.com

 

[SWComposites]

engroma -
I would like to get the adhesive stress due to the load.
> and what are you going to do with the adhesive stress?

I can get the adhesive proprieties from the certification test we carry out on it.
> what specific tests? what properties are measured?

As mentioned above, often several or more elements are needed thru the adhesive thickness. Also, nonlinear stress-strain properties are often needed to avoid stress singularities. Then there is the whole issue of failure criteria if you are trying to predict strength.

 

[engroma]

I am going to compare the adhesive stress values got by the part FEA simulations against the adhesive bonded laminate mechanical proprieties test results.
The specimens are made with the adhesive and adherents used in the part.

 

[Onda]

To model the adhesive by using solid elements you need to make very small elements to respect aspect ratio issue.

If your adhesive layers is 0.2 - 0.3 mm thickness, your solid element should not be more then 1 - 2 mm on the longer edge.
If you make a mesh so small you will have probably many problems to manage your model.
Using spring approach you overcome this problem.

To use the spring approach (See Gbor post for explanation) you must be very careful to connect very similar elements, because the stiffness property of each spring depend of area of surrounding elements. if you make different elements you should give to each spring a different stiffness! Not really feasible!
Best way is to mesh the two adherents with same mesh.

Regards

 

[jedward]

Modeling adhesives as bonding layers has been treated by Victor L Genberg (Sigmadyne Inc), Keith Doyle (Optical Research Associates) and Gregory Michels, also of Sigmadyne, in an SPIE short course which provides lots of information regarding the "corrections" for using thin adhesive layers in critical applications.
Check the web and also look for effects due to the high poisson ratio for many adhesives, i.e. elastomers that are incompressible