What are you trying to accomplish by bonding the two materials? If you are trying to bond them so they share load you have to use a strong and stiff andhesive. But this will only work with thin materials. It is impossible to bond 1" thick steel to 1" thick composite and keep them together through thermocycling. There is no adhehsive strong enough. Here you need to use a low modulus adhesive and a thick bond line to allow each tube to expand and contract without tranfering much load to the other. In this case why bond them together? A bushing can do the same thing. Perhaps what you want is a sealant?

Dear CompositePro,

You have a good point there. But I shall explain the problem a bit in more detail. The pipe is clamped on the outside, so on the composite material. On stainless steel (inner) liner, a pull force is exerted. The only load path to lead the pull force from the steel liner into the clamp is via a bond layer between steel and composite. This is done via shear stress, which are quite low compared to the stress that arise due to the difference in CTE. Does this put the situation in another perspective for you?

How long is this pipe or cylinder? It takes a certain length of bond for the differential strains to build up to where is exceeds the shear strain capability of the adhesive. The actual shear strain on the adhesive depends very much on the bondline thickness. You can cut the shear strain in half by doulbling the bondline thickness. But every adhesive will have an optimum bondline thickness for maximum strength. High modulus epoxy adhesive require a 5 to 10 mil thickness and will rapidly get weaker at thicker bondlines. This is because they are brittle and the adherend actually reinforces the adesive that is within a few mils. Chopped fiber additives can help this.

You also have to deal with radial shinkage. On cooling the steel liner will typicaly shrink more than the composite and since the steel is inside this will favor the formation of cracks and gaps - not good. Of course that also depends on what the composite fiber is and its orientation. Curing at room temp. with a slow heat-up rate for post-cure will minimize these stresses. Texture such as knurling can help improve shear strength.

How much shear load do you need to carry? What are the actual dinesions of your pipes? What temperatures will the assembly experience?

The pipe is more than 10 m long, with a diameter of about 0.5 m. It encounters temperatures between -30 and +80 degrees Celsius. I don`t think there are strong influences of the edge on the total bond length capacity, is that correct?
We see unequal radial shrinkage as the maior failure criterion. The orientation of the fibres are mostly close to hoop. So, very different CTE`s in axial/radial/hoop direction. The choice of laminate is dictated by the loads it has to carry (mostly non-axial loads). While it is an epoxy resin, I don`t think curing at RT is an option here.
How would you create the `knurling`?

Regards

By knurling I mean to get the surface as rough as posible for mechanical interlocking between layers. Grit blasting or peening are alternatives. You could  also spot weld a wire screen to your liner but that would be costly. The only other thing I can suggest is that if you are filament winding to use a high fiber tension. This will put a compressive preload on the steel liner so the composite can shrink with the steel when it cools. What you are doing will almost certainly result in delamination of the layers if the residual stresses at any temperature are such that the steel liner is significantly smaller than the shell. An adhesive will not keep them together. I've seen parts like this delaminate between plies of hoop fibers when when the plies are too thick - and there wasn't any steel liner.

Calculate what diameter the steel and composite tubes would be at minimum and maximum temperature if they were not bonded and then calculate how much force it would take to stretch or compress them to be the same diameter. This is what you are trying to do with an adhesive bond. Even if you calculate that the adhesive might be strong enough, if there is even a tiny flaw in the bond then the load that should be carried by the flaw area will add to the load at the edge of the flaw and you can see that the whole bond would quickly unzip.

I was curious on whether you were able to get this to work.