Learning Forum

[austinmears]

I have been working on validating adhesive performance under thermal loads and have been seeing unrealistic resultant shear loads on the mating substrates. Referencing Daly and Hawk (2017) or Yoder and Vukabratovich (2015) both utilize some component of mechanical properties of the adhesive and thermal properties of the mating substrate to determine the resultant shear stress of the material, see below. Using 3M's 2216 epoxy adhesive material with a bondline thickness of .025in and a bond area of 0.5in x 1in I was expecting to see a resultant shear response around 260psi at temperature, but Ansys was generating a response around 1000psi at temperature; something wasn't right. I narrowed down the problem to it being the CTE information defined for the 2216 adhesive. Including the CTE info made the resultant shear load blow up and excluding the information resulted in Ansys generating the expected shear load response of the substrates. I cannot explain this phenomena and why CTE information could give false information -- has anyone experienced anything similar or could explain why Ansys acted the way it did?

 

[dlooman]

Probably need a little more information.  You say:

 Ansys gives 1000 psi shear stress vs 260 psi expected. 

Tracked it down to the CTE in the adhesive.

Now shear stress blows up!?

It seems there could be shear strain in the adhesive due to the differential thermal expansion of the mating substratess, but also a shear strain due to the differential thermal expansion of the adhesive itself.  Apparently that second source caused the shear stress to blow up.  What are the coefficients of expansion and the elastic properties of the substrates and adhesive?      

[austinmears]

Thank you for taking the time to look into this. All values were taken at -30C.

Substrate 1 [Zerodur]: CTE=0.006um/m/C, Poisson=0.24, Modulus of Elasticity=90.3Gpa, thickness=0.375in

Substrate 2 [Invar36]: CTE=1.3um/m/C, Poisson=0.26, Modulus of Elasticity=141Gpa, thickness=0.0625in

Adhesive [3M 2216]: CTE=60um/m/C, Poisson=0.43, Modulus of Elasticity=5.57Gpa, thickness=0.025in

How does Ansys look at materials when they're defined with CTE values? Does Ansys take the CTE values and measure each interface individually, i.e metal to adhesive interface to adhesive to glass interface? If the CTE isn't defined for the adhesive does it assume that CTE=0 for the adhesive and then just estimates based on provided elastic performance of the adhesive and assumed metal to glass CTE difference of the substrates (like the journal article estimations)?

[dlooman]

Ansys doesn't look at the interfaces per se, it just shrinks the elements by alpha delta T.  It's easy to see why the shear stress in the adhesive would blow up when its CTE was included.  The substrates have much lower CTE's and are much stiffer than the adhesive so the adhesive shrinkage is entirely constrained at the interfaces.  Since the shrinkage accumulates over the length of the interface this typically results in very high shear stress/strain at the edges and especially corners of the adhesive.  Yes, if CTE wasn't specified for the adhesive then the shear stress would just be based on the relative contraction of the substrates.  

[austinmears]

What is more realistic at this point? Is it the method at which I'm trying to model the interface via Ansys Mechanical or is it the pain of modeling an elastomer in general? The shrinking of material and imparted shear stress certainly make sense especially if the adhesive is shrinking 60 times faster than the mating substrates, but what I still can't grasp is why excluding the adhesive's CTE delivers a more realistic response based on supporting data coming from multiple published articles.

[dlooman]

When I think of an adhesive, I don't think of an elastic body.  I think of a material with very nonlinear properties including creep.  If there are going to many cycles of thermal loading it seems you have to consider the thermal expansion of the adhesive itself because it could eventually fail due to fatigue beginning at the corners.   If this is just a one time load then I wouldn't worry about that.  The shear stress in the adhesive is "self-limiting."  It's not required to satisfy equilibrium.  So if it undergoes some shear deformation it won't hurt anything unless over many cycles it deteriorates.