Modeling of rivet connection in finite elment analysis

[ahkrit]

Hello,

i am currently trying to solve the issue of modelling many rivet connection in one joint using abaqus. With 3D-element rivets, using outer surface for contact is really time consuming. Even if the rivets are modelled with bigger element size.
What is the most reliable way of modeling lap joint rivet connections? Would like to evaluate the local stresses at the holes.

Thanks in advance!

 

[FEA way]

In Abaqus rivets can be modeled with mesh-independent fasteners but if you need accurate results at connections then you have to stick to solid elements. To reduce analysis time utilize symmetry if possible.

There is an example of rivet connection in Abaqus documentatoon, if you haven’t seen it already.

 

[ahkrit]

Thank you for your reply. I have not checked what is in abaqus documentation.
The problem is that i cannot use symmetry and also high number of rivet have to be included in the model,more than 100.
This is undoable with 3D elements and contact iteration.

Is not it doable some other way?
I saw and also tried beam element replacement modells but there are always pretty big differences and stress peaks around the hole edge.

Thanks!

 

[FEA way]

Maybe you could use a simplified model (for example with mesh-independent fasteners or beams) for the whole structure and then utilize submodeling to cut small part of this model and analyze it with solid rivets for better accuracy.

 

[SWComposites]

Calculate the peak stresses around the hole with a closed form hand calc after you get the fastener forces from the FEM using fastener elements.

AND, WHY do you need the local hole edge stresses anyway?

 

[ahkrit]

The stress peaks i need to get the value of the highes loaded connection and to dimension the joint.
What do you mean by "closed form hand calculation"? Anyway i have to do it not only for one load case, but with many loadcases for several rivet connections.

 

[rb1957]

Ahkrit,

I get where you're coming from ... we have FEA, so we can FEA anything ... but there are considerable problems in getting usable data from FEA for the detail you're looking at.

I think you want to know the truth about the stress state of these rivets holes ... and, no, I won't continue with the "few good men" Jack Nicholson line.

To get accurate stresses at the rivet hole you'll need a very fine mesh at the hole, you'll need to account for rivet "swell" (interference from the installation of the rivet) and a bejesus number of other variables and complications (rivet stiffness, rivet head/tail fixed end moment, etc). Oh, and you Have to run material non-linear to resolve the local yielding at the holes.

Fastener modelling should IMHO/IMAO be reserved for key discrete loadpaths, like the bolts that hold the wing onto the fuselage, or engine pylon bolts. "Continuous" fasteners like rivets should not be modelled. The load transfer across a set of rivets should be taken from the element results (shear and normal loads). These can be compared to the rivet allowables (in MMPDS). The holes are stressed but bearing at the fastener, or by (maybe) net section tension.

another day in paradise, or is paradise one day closer ?

 

[SWComposites]

Closed form calcs: go find and read Bruhn, or Niu, etc

I second what rb says above.

 

[ahkrit]

rb1957, SWComposites, to whom it may concern

thank you for your replies. Obviously you have a lot more practical experience calculating riveted structures.
As a newbie to this topic could you please tell me where i could find an exact paper, documentation, or similar about fatigue calculation of the rivet connections?
How is it calculated in the aerospace or construction industry?

General writings i found already a few but a usable documentation about a done calculation example not really.

Thank you very much in advance!

 

[rb1957]

Niu has a good example

another day in paradise, or is paradise one day closer ?

 

[ahkrit]

hello.

i did a bit more research and also checked Niu's book about fatigue calculation of rivets.

After many calculations and thoughts and research about rivet calculation process could you help me with following questions please:

1. for proper rivet lap joint local calculation is it neccessary to build at each rivet a really finemesh model and implement residual stress due to mounting process of the rivet? I doubt that it is done in the praxis especially if there are more hundred rivets in a structure.

2.what one can also find in many papers and studies, is that mid planes of connected plates are meshed with shell elements and the rivets are modelled with beam element with shell element heads connected through GAP elements to the plates. Is it a usable process for calculation? If yes again how is the residual stress taken into account?

3. what are the reasonable safety factors for bearing stresses? what is the fatigue calculation process for that and are there any? as i could see Niu is writing about the tensile stress taken into acount for fatigue.

Overall if anyone could give advice how to handle big structures with many rivet connections including fatigue calculation it is highly appreciated!

best regards

 

[rb1957]

1) "is it necessary to build ..." no (unless of course your boss tells you to do this)
2a) "Is it a usable process for calculation?" ... sure, you could model each rivet, but I wouldn't. I'd derive rivet loads from larger element loads.
2b) "how is the residual stress taken into account?" ... how is residual stress applied to the model ? There shouldn't be any significant residual stress (from assumbling the joint), pieces should fit "properly". In the real world this is done by the full scale tets.
3a) "what are the reasonable safety factors for bearing stresses?" ... what ever your industry uses are standard practice, mine might use 1.15.
3b) "what is the fatigue calculation process for that and are there any?" ... Niu does a good example. yes, he uses tension stress (these cause fatigue), he also uses the shear load (applied as a "pin load")

what business are you in ?

for large scale structures you use what we call a global FEM, a large scale FEM without super small details or mesh refinement. The idea is to pick up the loadpaths. For details like rivet loads ... these can be derived from large element loads by hand calcs. Once you have these detail loads you can get into the detail stress analysis (static and fatigue).

another day in paradise, or is paradise one day closer ?

 

[ahkrit]

hello.
thanks for the answer rb1957.

generally the residual stress i meant for the riveting process itself as it does chnage the stress distribution in the plates and the calculation and results for fatigue and bearing loads, does not it? so how is it taken into count if you do not perform detailed nonlinear analysis?

i am planning light weight riveted frame structures.

So the whole calculation method you do looks like: making a global model with bigger shell elements connecting the nodes at the rivets positions with beam elements(fastener elements).
This gives you the section forces (shear forces) on the rivet beams and with these forces you do the hand calculations according to Niu's book?
Please correct me if i misunderstood or there are any important missing details.

thank you!

 

[rb1957]

"generally the residual stress i meant for the riveting process itself as it does chnage the stress distribution in the plates and the calculation and results for fatigue and bearing loads, does not it? so how is it taken into count if you do not perform detailed nonlinear analysis?" ... never heard of this in nearly 40 years working ! "residual stress" to me is usually from manufacturing (like room temp forming, machining, or other) or from malfitting parts (being "encouraged" to fit).

"i am planning light weight riveted frame structures" ... is that a nod to airplane structures, or buildings ?

"So the whole calculation method you do looks like: making a global model with bigger shell elements connecting the nodes at the rivets positions with beam elements(fastener elements)." ... no, this is the most detailed model I could think of. Applying it to a large scale structure would make the model very large (elements with 1" side ?). Rather I'd model using features of the large scale structure ... fuselage frame 12 to 18" apart, stringers 6" apart, and mote model the rivets at all. Rivets loads are derived from these large element loads.


another day in paradise, or is paradise one day closer ?

 

[SWComposites]

For the previous questions:
1) NO. PERIOD. NO.
2) The residual stresses are built into the joint bearing allowable strengths. Note, for rivets you cannot use sheet material bearing allowables, you need to use specific joint allowables from test specimens with your sheets and rivets.
3) safety factor depends on what this is and how certified. Is it an aircraft? A building? A HVAC duct? …

The answer for fatigue analysis depends on what this is for.

 

[ahkrit]

rb1957,
please see this oink about residual stress from riveting process.

https://www.researchgate.net/public...sses_Around_Solid_Rivets_in_Mechanical_Joints

Have not you heard about it really?

"s that a nod to airplane structures, or buildings ?" these are rather "buildings". structures holding solar panels.

what do you mean by "mote rivets at all"? is it not included at all? how is the correct stiffness taken into acount?

what do you mean by "rivet loads are derived by these large element loads"? (i have no experience with airplane structures. but my problem is still somewhat similar.sheets riveted together.) do you take simply the stress values at rivets and calculate with the additional stress concentration factors and so on?


dear SWComposites,

i am not sure if we are talking about the same. i would like to know the calculation method for prediction of stresses and lifetime of the sheets around the rivets.

For previous answers:
1. if it is a No, then really how are the residual stress in sheets (due to riveting process) taken into account? please see the link above.

2.i would like to know the strength and fatigue lifetime of the sheets. not the rivet.
can you advise any good paper or documentation about that? with aspects of FEM including modellbuilding (element types, discretisation and so on)

3. it is a holding light weight structure.my question here again is the calculation method fir bearing stresses in the sheets at the rivet contact.


So generally again i would much appriciate an overview of these calculation procedure. It is unbelievable that so many rivets are detailed modelled in huge modells, and as you both mentioned you also do NOT calculate with nonlinear calculations for residual stresses in the connected parts, but then how it is taken into account?

could you please answer or send me documentation about the following infos:
1. modellbuilding of large models including riveted sheet plates connections?
2. what stresses are considered for fatigue life time and what is the process for that (a step by step overview example)?
3. what stresses are considered for bearing stress calculation for sheets?


i hope i could describe my problem clearly.
thank you!

 

[rb1957]

your link to residual stress from riveting ... yes, of course rivets swell, fill the holes, and produce "residual stress" ... but in 40 years I've never seen anyone take this into account in any stress analysis. You could take this into account in a fatigue/DT calc, but everywhere I've worked have used the more conservative "open hole". The point for us working in industry is "how do you ensure that every hole has the same interference/"residual stress" when hoels (and rivets) are drilled to a tolerance.

Ok, you're not working on a/c structures by buildings (with solar panels). Maybe your civil design codes are different to my aircraft building methods?

all I can say is that for 40 years in aerospace, rivet loads are typically derived by global FEM elements loads (elements representing a "large" piece of skin (like 18" by 6"). Global stiffness is from the global elements joining up. Yes, the stiffness of the rivets at joints is taken into account in designing joints, but not in global structures.

Model the rivets all you want, fine mesh, CBUSH ... whatever you need to feel comfortable.


another day in paradise, or is paradise one day closer ?

 

[ahkrit]

What do you mean by "open hole"? and why is that more conservative way?

 

[rb1957]

I'm assuming you're primarily concerned with fatigue/damage tolerance. The "open hole" model assumes that the hole is not filled and is a conservative representation of a "filled hole".

As a random thought, one thing we do regarding rivet holes is to consider the "net area" ... increase the tension stress across a line of rivets by (W-D)/W (where W is the rivet pitch and D the rivet diameter).

another day in paradise, or is paradise one day closer ?