Correct Material Modeling of FR-4 for an FEA ?

[MechEngr1997]

As I understand it, FR-4 is considered to be an orthotropic material. I want to do a static structural FEA in ANSYS Workbench on a part to be made of FR-4. However, I am having trouble finding all of the material properties I need. The Wikipedia page on FR-4 seems to give the most information that I've been able to find anywhere (

https://en.wikipedia.org/wiki/FR-4).

However, the material properties template in ANSYS Workbench appears to need more.

In the past, when doing FEAs on parts to be made of FR-4, I've really been primarily interested in whether the part is strong enough; so I didn't hassle so much over finding all the material properties. Below yield, stress is only dependent on the forces applied and the physical geometry of the part. I would just compare the maximum principal stresses to the strength numbers I was able to find for FR-4.

Now, I'm a bit more interested in deformation of the part I'm designing. Knowing Young's modulus (YM) and Poisson's ratio (PR) in 3 perpendicular directions seems like it's become more important. Wikipedia gives YM and PR in the lengthwise (warp yarn) and crosswise (fill yarn) directions - but not in the through-plane direction.

What are people out there typically doing to model FR-4 in an FEA? What's the most accepted practice in this situation?

Am I overlooking something? - Is there a way to extract the properties that are missing from the ones that are given?

 

[hendersdc]

I feel your pain as I went through this years ago while trying to model strain in printed circuit boards. The reason there aren't any properties in the Z direction (if the layers are in the X-Y plane) is because performance of the PCB is heavily dependent on the copper layers as well. The prepreg manufactures have no idea about them, can't control them, and they vary not only board to board but significantly from one section of the board to another.

I called a few of the prepreg suppliers when I was looking into this and they basically said that there is no standard property set out there that will adequately model PCB performance. I asked what their other customers were using and they said if they have anything it is proprietary and their customers aren't sharing it with them.

If you absolutely need to get the material data you could mock up some small parts and perform the testing yourself but I don't know what kind of variance you would find from a lot to lot variation. In the end it was easier for me just to be conservative in the model.