I have analysed dozens of different approaches, and I keep getting different results for the reactions in the fixing screws What am I doing wrong?
I have a situation where an element is fixed to the building structure, which we can assume is infinitely rigid. The element, for example an angle bracket as shown in the screenshot below, is loaded with a force, and I would like to determine the reactions in each individual fixing screw.
Obviously, there will also be a lever-arm effect acting on the screws, but I’m getting results ranging from 4.7 kN to 0.7 kN for the pull-out force in the same screw, depending on the method I use.
Do you have a reliable way of calculating this?
Until recently, I assumed that if I wanted to check the capacity of a screw, the force acting on it was the bearing reaction directly underneath the screw head.
There are several ways, check other topics about bolt modeling on this forum - you will find a lot of information.
Here are some other threads you can check:
Also see this approach:
To sum up, here are the main ways of getting the bolt forces:
Adding BCs directly or via rigid body constraints (especially when not modeling the structure on the other side - just mounting bolts to the ground). Use this if such a simplified approach is sufficient for you. Then just check the reaction forces.
Modeling the bolts as springs or beams (rather tedious). Checking forces is more tricky.
Modeling the bolts as simplified solids (usually the best approach). Checking forces is easy thanks to the section print feature.
Or try the idea with coupling constraints from the video.
It all depends on the accuracy of the calculations. CompressionOnly supports are useful. In my opinion, it’s easiest to place supports at the edges of the holes.
In the approach from the YouTube video (mentioned at the end of this thread: Quick connection of assembled parts), to avoid overstiffening the model by kinematic coupling constraints (the same may happen with rigid body constraints), leading to wrong load path and incorrect forces in the subsequent rows of bolts, a soft boundary layer is added around the bolt holes. But if you don’t have more bolt rows, rigid body or kinematic coupling (supported in v2.6.0) constraints can be fine. Or just BCs alone.
If you want to model a structure to which the plate/bracket is connected then solid bolts might be the best option. Especially if you also need to account for preload.
The purpose of the calculations is important. Calculations for educational or scientific purposes are usually accurate. For engineering purposes, many simplifications are assumed, as the uncertainty surrounding the installation of this element on the wall is quite large. Standards typically exist that demonstrate the effect of close anchor mounting and the overlap of adjacent concrete cones, which reduces their load-bearing capacity. In such cases, modeling should begin by examining the standards, as it is often assumed that not all anchors will transfer loads. It is worthwhile to model the washer diameter, as this will reduce stresses around the hole. Instead of round holes, holes are usually elongated, like beans. Engineering calculations neglect friction between the base plate and the substrate and bolt prestress.
Only the setup diagrams shown in the document are possible, for example as in:
In my case, I isolated a 15 mm diameter solid around the hole, as this is the area where the screw applies pressure. I then joined the solids as compound parts. At the back, I applied the pressure against the wall as hard contact. The wall is modelled as a solid, with a fixed BC applied at the contact area.
The angle bracket is prevented from shooting off to infinity by using compression-only constraints around the hole and also on the bearing area - again, the 15 mm diameter area.
It seems to me that this may be over-constraining the model, which could explain why I’m getting some strange results.
Compression-only constraints are mostly useful when you don’t model the backplate. For the current approach with two parts, it might be best to model bolts as highly simplified solids (just cylinders). Or use kinematic couplings with local orientations like in the video. However, it requires some keyword edits (for orientations because couplings are supported in v2.6.0). But you will easily obtain the forces and their distribution should be correct if you use that soft boundary layer (or solid bolt models).
Btw. it would be better to mesh the plates with hexahedral elements. For the angle bracket, you could use the Thicken Shell Mesh tool (needs surface geometry as input) or split the bracket into two parts and use extrusion/sweep.
Why model the contact and ground instead of using CompressionOnly on the back across the entire surface? The washer is a fixed support with little elasticity. It’s better to model the diameter of a rigid nut first.
Just make sure that the analysis is nonlinear (otherwise, it will work like normal springs). And you may need to adjust the constraint’s settings if it doesn’t converge. For example, the default stiffness is very high.
So, is that what you meant? I created an imitation of the screw — basically just a shape with a 15 mm diameter head. The “screws” are fixed at the end where they go into the wall, using BC as Fixed. How should I read the forces from this?
It’s usually done using 3 cylinder - shaft, head and nut. Then you don’t need BCs for the bolts themselves, you can apply pre-tension section and read the forces using the section print feature.
Depends on the case, but the Thicken Shell Mesh tool is indeed very useful since it can generate a hex or hex-dominated mesh for pretty much any part with constant thickness and it just needs a representative surface (middle, top or bottom). It can generate multiple layers and works better tnan CalculiX’s internal expansion of shells to solids.
What am I doing wrong?
