
Stress Test
Chris Dolejska Dec 24, 2008 2:59 AM (in response to Ben Heath)Cosmos Express would not test your model. So I built a simplified replica with the same 12mm boss and 2.75mm thick piece behind it. Tested easily with full resolution.
Express has Nylon 6/10 but no 6/6, don't know the diff
restrained all the edges of the backing
Force load tested the boss to 35 lbs in sheer straight down.
Looks like a FOS of about 10, so is says it's safe to 350 lbs. (but that seems a little high)
The pic is with it loaded to 15 x the 35 lbs, you can see the red where the boss is starting to sheer off.
TEST4.JPG 122.2 KB


Stress Test
Chris Dolejska Dec 24, 2008 2:38 PM (in response to Ben Heath)Got to thinking (that's always trouble )
You wanted a test of the assembly
so since Express can't do that I tested the metal piece as well
and since the metal piece loads the nylon only at the contact area which is 1mm wide I re tested the nylon piece to load just that 1mm strip
ran the metal at 1020, no 1018 avail. Restrained the metal by the central strip edges
Looks like the metal piece is the weak point although it still takes 4 x the 35 lbs load, so FOS of 4 on the metal piece . Pic is with 8 times load, fail starting at base of strips.
and interestingly the FOS of 13 on the nylon piece under the new test. So loading the nylon in a thinner section was actually better, looks like because of less leverage. Pic is with 20 times load. Fail starting at sheer base of boss.
TEST7.JPG 152.4 KB

TEST6.JPG 135.6 KB

TEST5.JPG 127.3 KB

Stress Test
David Anderson Dec 27, 2008 4:11 PM (in response to Chris Dolejska)keep in mind that nylon is a nonlinear material so the results for a linear static anlaysis are not really valid.
Stress Test
Chris Dolejska Dec 27, 2008 5:06 PM (in response to David Anderson)David Anderson wrote:
keep in mind that nylon is a nonlinear material so the results for a linear static anlaysis are not really valid.
So would the fact nylon is nonlinear make it weaker than this analysis indicates?
I sort of suspected a 12mm nylon boss might not really hold 350 lbs.
Stress Test
David Anderson Dec 27, 2008 11:55 PM (in response to Chris Dolejska)hi chris,
i'll try to explain...
first, plastics have nonlinear mechanical properties and thus are non hookean which means stress is NOT proportional to strain and modulus of elasticity E is not constant. thus a stress strain curve must be supplied with the material property so the stress is known wrt to strain. a metal only requires a single input for E as long as one keeps the stress below the yield strength (the linear region on a stress strain curve) the material will behave linearly meaning it will return to is intital state prior to loading.
the cool thing about working with linear materials is it is not necessary to know ahead of time the "real" load to run a stress analysis. one can simply apply a "unit" load which would equal a value of "1" of the force unit system, thus 1lbf or 1N. Once the analysis is run, it is simple to compute the maximum load since we can scale the results because we know the material behaves linearly. thus for a 1lb load, if we got a stress of 1000psi and the material yield strength is 10,000psi, then the max load is equal to (10,000psi/1000psi) * 1lb=10lb this works for prediciting deflections too.
in a nonlinear material world, such as you have, using FEA, we cannot predict stress and deflections in this manner and have to solve them iteratively using nonlinear solvers.
second, because this material is a plastic, it will most likely undergo large deflections compared to a material made of metal of similar size and shape and these large deflections must be accounted for by using a geometric nonlinear solver.
running a linear static FEA to solve your plastic strength is no more valuable than running a hand crank in this case. i think that this solution will get you in the ballpark but if you cannot run a nonlinear FEA (properly i might add), you should test, test and then test again.
if i understand the problem, you are going to "stake" the head of the nylon boss to permanently deform it such that it captures the metal piece and holds it in both shear and pullout loads. thus you need to know the crossectional properties of not only the shear area, but also the "tensile" area. of the "button" formed by the staking process.
for repeatable results, the staking process, be it hot or cold needs to be well controlled, probably better than "smashed"
sorry for the long reply...guess i got a little carried away.
best,
dave
Stress Test
Chris Dolejska Dec 28, 2008 10:34 PM (in response to David Anderson)David Anderson wrote:
first, plastics have nonlinear mechanical properties and thus are non hookean which means stress is NOT proportional to strain and modulus of elasticity E is not constant.
in a nonlinear material world, such as you have, using FEA, we cannot predict stress and deflections in this manner and have to solve them iteratively using nonlinear solvers.
second, because this material is a plastic, it will most likely undergo large deflections compared to a material made of metal of similar size and shape and these large deflections must be accounted for by using a geometric nonlinear solver.
for repeatable results, the staking process, be it hot or cold needs to be well controlled, probably better than "smashed"
sorry for the long reply...guess i got a little carried away.
how are you using "iteratively" in this context?
are "geometric nonlinear solvers" equations?
Is there FEA software that will calculate nonhookean materials?
Agree with the staking of nylon, it will have to be hot and heat can damage nylon.
no apologies, the longer the reply the better for learning, thanks

Stress Test
David Anderson Dec 29, 2008 6:37 PM (in response to Chris Dolejska)yes there are FEA packages that can solve these types of problems. comosexpress i imagine is only for linear static.
what i mean by iterativeley is that in small discrete steps (determined either by you or the solver), a solution is found, the stiffness matrix is updated and then rsolved again. this happens until a convergence criteria is met, sometimes it is the comparision of the current result with the previous result. if the difference is small enough to be negligeable then the solution ends.






Stress Test
Loic Ancian Dec 30, 2008 12:07 PM (in response to Ben Heath)In solidworks, simulation tools are available in 3 packages :
 Solidworks Simulation (linear analysis like cosmos express but you can work with assemblies, contacts..)
 Solidworks simulation Professionnal (frequency, buckling, thermal analysis)
 Solidworks simulation Premium (large displacements, contacts, non linear materials, dynamic loads analysis).
Simulationxpress (or cosmosxpress in previous versions) is originally for firstpass design checks of very simple SolidWorks parts...