contact questions

prop_design

hi victor,

i was experimenting with assemblies and contact. the previous thread inspired the example. i attached it here. it's just a simple bolted connection. the only difference is i didn't use washers to save model size. i had a problem with model sizing being way to big for my computer. so i had to decrease the mesh density of the bolt. i haven't used mecways contact much. i tried it awhile ago and gave up on it. but i tried some things here and got a result that seems somewhat reasonable. i couldn't find anything in the mecway help or ccx help that would guide me. i ended up trying various things. if you look at the model setup, i was wondering what your thoughts are for the inputs to the contact entries. i tried different contact methods and different inputs. do you have some you would recommend or rules of thumb to go by. with ansys, it did all that stuff for you. so i'm not sure what reasonable inputs should be in this case.

all of the geometry dimensions and load inputs are completely made up. i just sketched something applied loads and got lucky that they worked out. it would be better to make a standard fastener dimension and use bolt load calcs as inputs. but this was my first pass.

i also wondered if it wouldn't be better to apply displacements instead of forces in moments. but i haven't looked to see if mecway has a rotational displacement. so that approach may not be possible anyway. using forces and moments is more intuitive from a user perspective. so that approach would just be a fall back.

prop_design

i can't seem to edit the original post to this thread.

i tried the displacement method and it wasn't good. when i tried to reset the model back to the way it was i got different results. they seem better than they were now. not sure why the nut expands more than the washer though. you would think if you define contact between two surfaces they would stay in contact. but nothing i tried seem to get that result. in any event, this is definitely the best i can come up with contact using mecway. like i said it's a lot better, easier, and faster to do this sort of thing with ansys workbench. but maybe it will help someone.

i would still like your input into the stiffness values for contact. i ended up setting all of them to the modulus of the material i used. i suppose if i used different materials you would have to play around and see what works best.

the way i gauged if it was working good was to see if i was getting the motion i wanted and if the stress in the bolt fillet looked good. i achieved both things. but you should do an example comparing a standard bolt to bolt calcs or to possibly other fea software.

i wasn't interested in the bolt thread stress. that is why the threads aren't modelled. i actually did something like this around 15 years ago with ansys workbench. i looked into how the bolt shaft deformations caused bearing rub. the analysis matched the part wear exactly. no one knew why they were getting the wear the way they were until i did the fea analysis of the assembly. i bring it up because it can be useful to model the bolt assembly without the threads. it certainly cuts down on mesh size, solve time, and cad modelling time.

updated model attached

Victor

Yea, ANSYS works out the contact stiffness automatically but here you have to pick a value. There's also a lot of trouble with CCX if a part is only supported by contact (not bonded contact). I'd guess that most of the support emails I get are about contact. The problem hooshsim posted has got me stuck without using bonded contact like you did.

The washer rotating (looks like expansion but it's really rotation) more than the bolt is because the stiffness per unit area is too low on the bonded contact between them. The higher the value, the more correct the solution. I added a couple of zeroes and it came out alright looking. There's effectively a layer of elastic material between the surfaces and that value is it's Young's modulus per unit thickness in the normal direction. Ideally, it would be infinite but that doesn't work so you need a very large value.

prop_design

hi victor,

i started with the default contact and stiffness value from the program then had to experiment with a bunch of different setups and stiffness values. usually the stress was corrupted at high stiffness values and at lower values the deflection was messed up. a lot of the time the end of the bolt was bent which didn't make sense. can you attach your model i would like to look at it.

Victor

Here it is. I changed the last bonded contact's value and remeshed the parts with Fit midside nodes to geometry on all of them.

prop_design

ok thanks. i will take a look. i'm currently working on updating the example to a standard bolt so that i can compare to known values of force and torque. i picked a M16 class 12.9 bolt to work with. so it's of the highest strength allowing max force and torque. i want to see what sort of inputs to the contact stiffnesses are needed.

prop_design

i finished the assembly. it is intended to help anyone wondering how to properly include a bolt assembly in their model. i will post it in a new thread so people can find it easier.