Contact Conductance in Riveted Joint

[MrRogers1987]

I'm trying to work through a thought experiment with a colleague about how to estimate the heat transfer across a riveted joint. The application will be in a vacuum, so there's no interstitial fluid between the joined members to help with conduction. Thus you are highly reliant on surface finish and contact pressure basically causing surface imperfections to yield and increase contact between the joined members. In my industry bolted joints are much more common for the type of connection being examined here, but I don't feel we should treat this the same way thermally. I am not overly familiar with riveted joints though and am struggling to convince myself what the primary mechanism for heat flow will be and what can be neglected.
[ul]
[li]For example, it's not immediately intuitive to me whether the amount of contact pressure from a riveted joint is comparable to a properly preloaded fastener. My understanding is that rivets are primarily for carrying load in shear, and the tensile capability is basically just the rivet head. This makes me think the joint isn't under as much compression as a bolted joint, but again I'm not very familiar with riveted installations. [/li]

[li]Can anyone point to a reference which discusses contact pressure between riveted members?[/li]​

[li]Is there local deformation in the joined members from the riveting process that reduces the contact area in the vicinity? Or does it work in the favor of compressing the members together?[/li]
[li]Assuming the contact force is not significant, is it reasonable to assume that the majority of the heat transfer is through the shank of the rivet and not through contact of the joined members (basically opposite of a bolted joint)?[/li]
[/ul]

Any other insight or considerations that might be helpful in working through this thought experiment is welcome as well!

 

[SWComposites]

I can't think of (at least a first order) reason why a riveted vs bolted joint would behave different thermally. Rivets should clamp the sheets together just like bolts. Maybe the clamping force is different but I can't see why clamp pressure would make any difference to conduction. if you have no clamping pressure, or worse yet a gap, then you have a very bad joint and worse problems than thermal ones.

 

[MrRogers1987]

Conduction across an interface is highly dependent on the clamping pressure. Particularly in a vacuum where there is no interstitial fluid between the clamped plates to aid conduction. Estimating the contact pressure and area is relatively straight forward with a bolted joint since you can control the preload. But I have not come across any method for estimating that same contact force for a rivet. Obviously yes there is some clamping force there, but how to quantify it? It doesn't need to be as high as a bolted joint since rivets are really only meant to take load in shear. So that force could be relatively minimal and the joint still perform adequately from a structural standpoint.

 

[rb1957]

is there fay seal between the plates ? conducting fay seal ??

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[SWComposites]

contact pressure and area is relatively straight forward with a bolted joint since you can control the preload ah, well, you can (sort of) control the torque on the fastener, but the preload is highly dependent on a lot of not so easily controlled variables, friction, etc.

what kind of rivets? driven? squeezed? blind pull type?

if you really need to know rivet clamping force then probably have to put a load washer between two sheets and install the rivet.

 

[MrRogers1987]

@rb - Currently no. I requested we put some sort of material at the interfaces which could fill that purpose, just to make the joint performance easier to predict if nothing else. This was shot down due to "assembly complexity concerns"... I have not given up on that route though.

@SW - Correct, for a preloaded bolt the clamping force can be controlled to a relatively good degree. There needs to be some accounting for uncertainty in actual applied preload but at least there are well documented methods for doing so. Can you make a similar estimate of rivet clamping force from the setting process if you know things like rivet type, size, and plate thickness? I would assume it is a less controlled process so there would need to be a pretty large uncertainty around whatever nominal estimate you could make. Any reference you have for estimating that I'd love to see.

I had thought about requesting some test samples for thermal testing, but I like your suggestion of using a load cell. It won't be quite as accurate for determining the actual interface conduction, but it will be much easier to setup and will at least give some empirically backed number to case calculations on.

 

[SWComposites]

do you actually have thermal conductivity vs joint clamp-up pressure data? Haven't worked spacecraft in a very long time but don't recall ever seeing that kind of data, or worrying about joint contact pressure.

 

[MrRogers1987]

Yes, there are formulas for estimating thermal contact conductance based on clamp pressure, joint member material(s), and surface roughness. A few different methodologies exist, but they largely come to the same values. There is some empirical backing for specific joint types which are more commonly found as well.

I don't typically have to go down that road as there are good estimates you can make for bolted joints just by knowing size and number of fasteners. Unless there is an a-typical joint, such as between metallic and non-metallic components, or with only a few fasteners so you aren't guaranteed solid/unform clamping at the entire interface, or like in this case where the clamping pressure is not from a preloaded bolt.

 

[geesaman.d]

While I'm sure there's an analytical way to approach this, if the riveted joints are consistent enough and frequent enough, you might do well to perform a representative test and treat the riveted joint as a boundary with its a measured transfer coefficient.

David