I need to model a riveted structure. Especially for the joints the model should give a good representation of the real load bearing behaviour.
In Abaqus there is the option to use mesh-independent fasteners with the *Fastener command. This is not implemented in Calculix (yet).
I just want to avoid modeling each rivet individually in detail. An option would be to use a shell model for global behaviour of the structure and submodelling for critical details.
How would you approach the problem using Calculix?
Of course, the most accurate approach would be with the use of 3D models of rivets but it can be very time-consuming. Submodeling could be utilized to reduce the computational cost.
When it comes to the approximate approach, you can model the rivets using the method typically applied to bolted joints - beam elements representing bolts and coupling constraints attaching beams to surfaces of the parts being connected.
Fasteners are usually designed using analytical calculations. For this purpose, you can use general engineering design textbooks or those written for aerospace engineers. They should cover such topics in detail. Specifically, I could recommend “Analysis & Design of Flight Vehicle Structures” by Bruhn and “Airframe Structural Design” or “Airframe Stress Analysis and Sizing” by Niu. Unless you are interested in structural engineering, there are many books meant for that field as well.
Thank you for the answer. I totally agree with you.
The question is, if fasteners can be easily implemented into Calculix? I am not that deep into the topic to write my own routines here I am afraid.
Also thank you for your literature suggestions, but I am in the civil engineering domain. Thestructure in question is a riveted railway bridge.
So far I have a tendency to use the submodelling approach, but will try it first on a simple testcase.
The approach with beam and “spider” (coupling) elements is very useful but, depending on the number of fasteners, can be quite tedious unless you utilize some tricks such as Python scripts for automation, good preprocessor, or advanced text file editor options.
I may be able to suggest some books for structural engineers that cover rivets:
- “Structural Engineers’ Handbook”. Data for the Design and Construction of Steel Bridges and Buildings" Ketchum (old but good)
- “Guide to Design Criteria for Bolted and Riveted Joints” Kulak, Fisher, Struik
And if you are from Poland (as the name seems to suggest), check out this book: "Projektowanie i obliczanie połączeń i węzłów konstrukcji stalowych tom 1” Bródka, Kozłowski.
yes, it will be a quite tedious process, as there are really a lot of fasteners. AndI will need numerous simulations for a parametric study as well.
That is why I want to give some thought into the model before I start to build it up. As Preprocessor I will work with the Salome Platform, had some good experience in previous projects with it.
I will check out your recommendations (except for the polish one, unfortunately I am not from there).
Maybe a good first question is how much detail do you really need? If deformation at the bearing surface is not important, you could model rivets as springs in a gross sense. AISC in the Instantaneous Center of Rotation method considers bolts as point loads and calculations the ICR based on deformation under load like a spring. It may not be reasonable or necessary to create a 3D model for each rivet and deal with deformation compatibility in a finite element program.
Rivets were driven rather than installed in drilled holes. In the process, rivets expand to fill the hole. This results in great surface contact compared with bolts in 1/16" oversized holes. Rivets were lower strength than bolts however.
interesting with bolt connection simplification, below are my approach in CalculiX to Abaqus *Fastener keyword. some notes regarding to this models are:
- overlapping of bolt and plate areas have mesh continuity, stiffness multiplied by ten to approach rigid areas
- bolt face spitted by two, top and bottom plate groups. it’s connected by tie constraint
- contact between plate surface
- convergence fast (only 22 iteration for 5 number of bolts)
- but, this model could lead to wrong results (as seen on vMises peak stress)
- using shell element for this approach also possible due to similarities.
Very interesting model @xyont , could you share the Prepomax model?
Nice model @xyont. Hopefully, you could share the Prepomax model.
What is the expected value in the peak areas?
this is quick test only, may next day i can improve the study and place the input (CCX) and geometry (STP) files on my GitHub account. more general ways since PMX files cannot be opened by CalculiX user in Linux environment.
regarding result is not about value discrepancy, but in sign contrary. it could be wrong since bolt or rivet resist bearing (compression) only when subjected to in plane force, however the *fastener approach also resist tension side around the hole perimeters.