Bolt Pattern Restraints
Bolt Pattern Restraints
(OP)
I am using FEA to work out the stiffness of a machined bracket. It is mounted to an actuator by a bolt circle. I need to apply a restraint on the mounting surface. I added a 0.1mm thick pad around each clearance hole and applied the restraint to each of these. My assumption is that the face around the screws is constrained, but much of the rest of the face is not constrained.
I am using M6 screws. My part is aluminium, 100mm thick. I have specified the pads as 16mm in diameter.
Does this sound like a reasonable set of assumptions? Does anybody have a better idea?
JHG
I am using M6 screws. My part is aluminium, 100mm thick. I have specified the pads as 16mm in diameter.
Does this sound like a reasonable set of assumptions? Does anybody have a better idea?
JHG





RE: Bolt Pattern Restraints
I would create at least part of the support structure (which is where you put your boundary conditions on), ten use spring elements of your chioice (Celas/Cbush etc) with correct fastener stiffnesses in to transfer the load.
Depending upon your loading application i'm not convinced the balanced 3-2-1 set up will work (might be wrong), but for something which isn't statically determinate i'm not sure if you can use the method? johnhors hopefully will shed some light??
If your interested in the stiffness of the bracket, does that imply that your going to run normal modes?
There are some good sources about correct bolt modelling one of which is:
Fastener Modelling for MSC NAstran Finite Element Analysis (Rutman, Viisoreanu & Parady - AIAA 2000-01-5585)
RE: Bolt Pattern Restraints
corus
RE: Bolt Pattern Restraints
With the bracket in question, as Luke says this must be in contact with something else (the actuator). If at least part of this actuator is modelled and connected as 40818 suggests via spring type elements to the bracket, then it is still possible to utilise 3-2-1 supports in conjunction with balanced loads. However the attachment must not introduce any mechanism (whereby the bracket could slide or rotate relative to the actuator).
But I think that the OP has a deeper problem, "work out the stiffness of a machined bracket" is very ambiguous ! It depends on what he wants to get out of it. Nastran for instance could output a super element (of just the bracket) , which is just a reduced stiffness matrix connecting prescribed points, which could be used as 40818 hinted at in a simplified model for modal analysis.
I hope this helps !
RE: Bolt Pattern Restraints
I am doing plain, static analysis. This thing will move around slowly and will not be subject to vibration, so I am not interested in modes. My requirement is for a rigid fixture. The stresses will not get anywhere near a failure point. Stresses are only interesting to the extent that I can see where the structure is bending.
I am putting the pads on the side opposite the bolt heads because I am trying to model the interaction with the machined face of an actuator. It is obvious that I could model the entire back face as completely fixed, but I think the clamping only is effective around the bolts.
I am using COSMOSworks with SolidWorks. It would be nice to have a constraint of a flat surface that can be clamped to in one direction only.
The actuator is off-the-shelf, and we are trying to find out how rigid it is. I will design the base. There are no weight constraints, so I can make this very rigid. We are also analyzing fabrication tolerances. Yes, the bracket is part of a larger system with an error budget.
JHG
RE: Bolt Pattern Restraints
RE: Bolt Pattern Restraints
The bracket is a cantilever extending out from the actuator. I am looking at the deflection at a couple of mount points and working out the rotation of an optical device attached to them. I am trying to hold the optical device to within hundreds of microradians.
I am analyzing the structure piece by piece. If I did it all together, I do not think it would not eliminate the problem I described above.
JHG
RE: Bolt Pattern Restraints
The "tool" you are using won't let you model the assembly anyway, as you probably know. Yes, your assumptions are reasonable for the modeller you are using.