Has anyone figured out how to restrain a surface of a part to have equal deflection in a single direction only? In other words i need a surface to translate in all three directions but the surface must remain parallel to its initial plane. See picture below.
I need the surface where the force is applied to equally deflect in the vertical direction. In the image below you can see the undesired deflection:
not really sure I understand what you are really after but in a restraint you can put in 3 mm in a specific direction when you are applying a restraint and it will force that to happen. I really don't get what you mean by the equal bit though. Equal to what? to 3 mm?
Here is what i mean by equal: I would like all the nodes to have the same y deflection, but i would like simulation to solve for what this value is.
The deflection in the Z-direction causes the Y-direction values to not be equal. Add a constraint to set the Z-deflection values to zero. It's up to you to determine what node, line, or surface to put it on. Not sure of your intent.
I see what your getting at but i dont think thats what i want.
Here is the current deflection:
Your suggesting to straighten the sides:
But what i really want is:
Are you pulling up on the 2 sides with different loads to generate that twist?
There must be another part then that restricts that motion? If you can show us that it might help determine how to represent it as a fixture.
What you need to use is a symmetry fixture between the top and bottom half and on the ends.
I do not understand what you are suggesting. To me it sounds like you are suggesting this (which will clearly not work):
Here is one very unsatisfying solution:
However becuase i will be working with so many different designs similar to this, i would really rather figure out the simpler model shown below (which is currently not providing the correct results):
Here you can see what the deflection is supposed to look like:
I am still confused. Are you trying to simulate the application of an axial load to the full up part? Why not put he loads on the big bit so we can see what you want on the small scale. I suspect that nmaybe cylindrical coordinate system mayb of some help here but it is hard to know what you are trying to do, at least for me.
What are you trying to find out in the sub s.cale that you want know about on the full scale?
I am trying to optimize the geometry to maximize energy absorbsion in the spring while avoiding gross yeilding.
I should mention this is like the fifth design i analyzed and likely will have many more to come.
If I was to try to get a sub scale models response to an axial I would use radial cuts along the section of minimum thickness and isolate 1 stack. And then I would apply symmetry ( nothing normal to the cut face), same restraint on the bottom and load at the top. Since the little elements are all in series you could isolate 3 of these if the idea is to examine the stress and then increase the size of the fillets to attenuate the stress riser at the change of stiffness near the axially oriented section. You would ignore the artificial stress risers at all the restraints which will produce the highest stresses. Hence the need for 3.
Just to make sure the geometry i clear take a look at the close up below. I actually dont care about localized yeilding as long as the hysteresis isn't significant.
Ok that's better. I get it now. You have to do a complete repeating element and if there isn't one you have do the whole thing. Could not see these details in any of the previous pictures. If you have to do the whole thing - use thick shell elements. Auto correct and bad finger aim have gotten the better of me.
The bowtie shape i started with is the repeating element, isnt it? What I am having trouble with is applying the correct constraint to the load surface.
not exactly - since it flips orientation.
it would have to be something that you could build up the part from only using cicular and linear patterns.
I don't think the forum is working very well with iPad. My comments are not going under where I made them.
Well unless there is something you left outbibhave saybthatbloading of the full item under pure axial load will not produce the load path in your good picture. The load will take the stiffest path and across the thin bits isn't it.
You need to get the right load path first. The image way above that you say is what deflection is suppose to look like is what you need. What's wrong with that model with respect to stress?
nothing. it just takes longer to iterate
This just seems like the sort of simplification that solidworks should be able to do. And also if i new how to do this i would do it all the time.
Is there some way to make a rigid mating part which defines a sliding surface and then define some kind of tensile sliding surfaces contact? Like a frictionless vacuum chuck.
Add a displacement to the surface normal to the face of some small value and that should give you what you are looking for as the far the face not angling which I think is the issue. You can get the load from the reactions. You can also use the remote load in the rigid displacement mode to get this to be a restraint that can move axially yet not angle. It would be more convenient but too hard to explain till you are sure the fixed displacement works. Give that whirl and if it looks good I will explain the other.
thx
i tried that and it does work
unfortunately it doesnt save me anytime due to the iterations. Its faster to fun the whole stack.
Thanks for everyone who provided input. I think its safe to assume that solidworks is not capable of restraining as surface to have equal deflection in a single direction.
-Daniel
Ok then just use a remote load acting as a restraint and you are good to go. Are you interested?
wouldnt i still end up in a guess and check situation? Guess the deflection, then see what the resultant is?
No you would apply a load. Put a sketch point In the middle of the face . Use it to put a coordinate system with and axis vertical, in the direction of the load. Add a remote load from under the remote load selection user loads. Pick the displacement rigid connection option, pick th face, pick the coordinate system, leave all the location option at 0. You will have 6 degrees of freedom to specify, set all the rotations to zero, leave the translations free. The add you load and run. Good luck. I have trick or treaters at the door.
Bill:
Using your method above i was able to get it to run. Thanks for the tip, i will definitely use the remote load function in the future:
To bad it makes the face rigid. I this case the rigidity doesnt introduce too much error.
Everything is an approximation. If not rigid then what would you use? You would need to define a planar and angular stiffness and I doubt you would know what they are. The material that is not attached to the face would in efffect keep the planar dimensions of the face very close to the underformed dimensions, we just stopped in from rotating which is what the real material would do. I would play with the restraints in the remote load, restraining all the rotations and freeing all the translations may not be the best way to model it - I didn't think about it all that hard but you probably should. Ciao,
