modeling polyurethane foam

[jedward]

Low density polyurethane foam is used to fill the space between a series of PCBs which are edge mounted to provide vibrational damping. How do you characterize the modulus for the reticulated foam at the resonant frequency of the assembly, in the 500 to 1000 Hz range?
Material is viscoelastic and may have a negative poisson ratio and appears to have much higher modulus at this frequecy than understatic conditions.

 

[GBor]

If I understand the situation correctly, you could use springs, or ignore vibrational frequencies in this range. I assume a non-linear material vibrational analysis isn't possible?

 

[jedward]

non linear analysis is beyond me.
cannot use springs in this heritage design,(changing design very difficult). I am attempting to provide a model for the assembly with the foam being the primary unknown. Using the static modulus of 30 psi does not give agreement with sine sweep measurements. Better results with about 500 psi, but no winner.

 

[GBor]

I didn't mean physically use springs, I meant approximate the foam in the model with springs...sorry, guess that wasn't very clear. What software are you using? Is there not a way to apply a damping coefficient for the material?

 

[jedward]

I am using DesignStar and have modeled the foam as a solid. Software at this level does not input damping, but I do not expect the inclusion of damping to change the frequency of the modes by much; damping should only decrease the Q of the system and thus decrease the amplitude of the displacement. Many damping materials show a significant increase in the "effective" modulus at high frequencies over the static values. Can not find any specifics on the Scott Polyurethane foam that is used.

 

[GBor]

I understand the dilemma, but have no idea how to compensate in Design Star shy of physcally manipulating the input file, which I also don't know how to do in DesignStar...sorry. Perhaps someone else will chime in at this point...

 

[amorrison]

OT a bit.

Recall that polyurethane foam is an excellant thermal insulator - using it might produce undesired temperatures in your electronic components.

 

[GBor]

Had another thought somewhere along the way. The material is "viscoelastic", which is clearly non-linear, but can you break the analysis down in to "bands" of linearity? Run an analysis with one set of material properties from 0 to 499 Hz, then run a second that covers frequencies from 500 to 1000 Hz, and a third...well, you get the point. You could change the properties for each piecewise linear material set.

 

[jedward]

amorrison---material provides thermal insulation - Yes, but not a problem in our application operating in vacuum

GBor--Deflection is small and therefore "nonlinearity" (whatever that means for viscoelastic materials) is probably not the issue. have run cases with a wide range of values for modulus but do not get agreement with test data on modal frequencies. Model geometry is known to be good.

 

[GBor]

jedward

Using a linear calculation will likely never match test data. Even with small deflection, viscoelastic material damping properties generally change pretty significantly.

Short answer: You are using the wrong material model. This foam may behave more like a 9th order Mooney Rivlin material or a neo-Hookean model. Or any of a dozen other proposed, very non-linear viscoelastic material. A simple, linear elastic modulus will not match test data if this is the case.

 

[hondaknight1]

Gbor is correct, non linear does not mean that the material model only has an effect over large deflections, especially with viscoelastic material models. The speed of loading or strain rate is really important here and this where a viscoelastic material model may well help. You pretty muich answered your own question in your original post "appears to have much higher modulus at this frequecy than understatic conditions". If this isn't a case for using non linear materials models or at least piecewise linear as Gbor suggests then I suggest you revolutionise the FE industry by convicing everyone that non linear isn't neccesary!

You may well need to get your materials tested to give data for a decent material model, which probably means uni-axial and equi-biaxial testing and possibly some variable strain rate stuff aswell.

I'm no expert on dynamics but couldnt you carry out a sine sweep on a simple block of insulating material at the desired frequency and then back calulate what the modulus should be. That might put you in the ballpark..

 

[amorrison]

OT#2 a bit.

Even in a vacuum

1. A "hot" component may be touching the foam and get heat via conduction which will not leave the foam because of its own insulation and the lack of air to produce a convection heat transfer component.
Probably anything above 150 F should get serious consideration.

2. The other components that get heat radiation effects are also in a vacuum and therefore can only get rid of heat by conduction and re-radiation.

 

[jedward]

amorrison
the locations dissapating the most energy get conductively coupled to the case and finally radiated to space

other components have come conductive path to case

the foam provides for more energy transport than a vacumm via radiation to the surrounding at similar temperature.

 

[jedward]

hondaknight1
plans for such a material evaluation test are underway

However: a parametric fem for our geometry could not find agreement with the observed modes found in a sine sweep test -- only the foam modulus was varied

 

[hondaknight1]

Well the problem is that as you change frequency then the modulus will vary too, and probably in a non linear way - looks like a set of physical tests to determine the paramters for a viscoelastic model might be the only way to go. Let us know the result!