Trouble with stress results surrounding fasteners

[JC_13477877]

I'm currently working on trying to evaluate the safety of the screws used for a lift bar of a block but seem to not be able to get reasonable values for the stresses inside the bolts. I believe that the model is over constrained but don't know how to fix it. I'm currently using Creo 11 simulate not the live one. 

The assembly consists of a large steel block (approx. 4,400 lbs), with a T-shaped lift bar bolted to it. The lift bar is mounted using three screws: two on either end of the top of the T and one at the base. The lifting point is centered on the crossbar of the T (via an eyebolt), and in the simulation this point is fixed. Gravity is applied in the -Y direction using -386.1 in/sec².

The fasteners are 1-inch diameter UNRC coarse-thread screws modeled using Creo Simulate's screw fastener feature. They are threaded directly into the block (no nuts), with preload disabled. The materials are high-carbon alloy steels.

• Analysis Type: Structural, nonlinear with large deformation enabled

• Constraints: Fixed at the lifting eye (midpoint of T-bar)

• Loads: Gravity only

• Contact Interface: “Contact” selected between bar and block (surface-to-surface)

• Mesh: Reasonable for the geometry, but refinement hasn’t improved stability

---

Issues Encountered

1. Unrealistic Fastener Forces

   • In some cases, Creo reports axial forces in excess of 8 million lbf on a single screw, which is clearly unphysical.

   • In others, all shear forces on screws are reported as zero.

2. Contact Interface Not Allowing Separation

   • Despite selecting "contact" (not bonded), the lift bar and block behave as if fully bonded.

   • Creo does not present an “allow separation” checkbox — only penetration is available.

   • This seems to artificially restrict behavior and drive up fastener forces.

3. Warnings About Large Rotations

   • I frequently see this message:

     “Running large deformation analysis with advanced rigid links or fasteners may lead to inaccurate results near large rotations.”

   • Unsure if this is related to the instability or a secondary concern.

---

What I’ve Tried

• Disabling preload in fasteners

• Switching between contact types

• Testing mesh refinements and load step reductions

• Reviewing fastener dimensions and thread definitions

1. Is the screw fastener tool suitable for nonlinear large deformation analyses with contact?

2. Should I instead model the screws as solid geometry if they’re threading into the block? i tried this but cant seem to correctly simulate the screws.

3. How can I properly enable separation between contact surfaces i

---

Any help would be greatly appreciated. I’m trying to determine whether the issue lies in the fastener definitions, contact interface limitations, or some deeper constraint/load interaction

[KenFarley]

There are folks on here occasionally who have done a lot more simulation than I, but maybe I can be helpful.

(1) First off, if I were analyzing something like this, I would not include the block. The trouble is it's going to mesh the block and it is a rather large object with respect to the thing you really care about, the "T" bracket. Instead, I'd calculate the mass of the block, then distribute that among the fasteners.

(2) For parts in the assembly, I'd have the "T" bracket and the three bolts. For the contact between the bolts and the counterbores, change the type from bonded to contact or whatever the terminology is. You want the two to be able to slide against each other.

(3) Put the force loads onto the bolts. Maybe add cylindrical surfaces around the bolt representing the length of them that interface with the block holes, so you can apply the load to those areas only.

(4) Put a constraint on the eyebolt hole surfaces to fully constrain the analysis.

This is what I'd try. I hope one of the true experts on this stuff chimes in with better advice if they have it. I'm idealizing the loading as a simple vertical load, perhaps true if the eyebolt is located at the center of gravity.

[JC_13477877]

Thanks for the response!

After many hours of trial and error, I discovered something strange: when using gravity load in Creo, if your units are set to lbm (pound-mass) instead of lbf (pound-force), it ends up multiplying gravity, and a 4000 lb block ends up generating over 100,000 lb of force — way more than expected. Realistically, it should be just 4000 lbf. Once I changed the units, all the screw forces became rational again.

I’m not sure if this is a setting issue on my computer or just a quirk of the software, but switching to a different unit system solved it for me.

As for your suggestion to apply a direct force instead of gravity — that wouldn’t quite work in my case. I’m analyzing a lift bar, and the center of mass isn’t directly under the lift bolt, so the force distribution isn’t symmetric. This is actually the simplified version of what I’m trying to simulate. The full setup involves another block and lift system with 12 bolts and 8 contact surfaces, which seems to be overwhelming Creo. I suspect this is also related to some settings or limitations on my machine, which I’m currently trying to sort out.

Thanks again for trying to help — I appreciate it!

[skunks]

my teaching example

[JC_13477877]

Thanks for the file!

Before diving into the details — I wanted to ask: how long is it taking you to run the simulation you sent? On my end, it takes around 30 minutes just to get results, which makes me think something might be off with my settings or solver configuration.

I have 16 GB of RAM, and my solver RAM allocation is still set to the default 512 MB. I’ve seen others increase this to 8000 MB or more, so I tried doing the same. But when I increased the RAM limit, the model that previously converged fine suddenly refused to converge anymore. Based on what I’ve researched, this might be because Creo switches to an external solver when more RAM is allocated, which could behave differently than the internal one.

This is really frustrating because I’m trying to run a more sophisticated analysis using Creo's built-in fastener restraints, specifically so I can use the results window that displays all the forces. I got one model working perfectly and the results made sense, but now that I’m adding more contacts and screws, it just won’t converge.

The thing is, the new model isn’t that much more complex — it just has a few additional contact surfaces and fasteners. I wouldn’t expect it to take all day or completely fail to solve. But instead of the convergence dropping from 1 to 0, it shoots up to values like 2e+7, which seems completely wrong.

If anyone has advice on:

• How RAM allocation impacts solver behavior in Creo

• Why switching to an external solver would break convergence

• Or any other settings I should check

…I’d really appreciate the help. Thanks again!

[skunks]

The analyses on my PC take 20-25 minutes.

Contact analyses are very difficult and can take several days 😂