set a tree with traction-flying-torsion,principle help

[meccanicamg]

I tried to make a fem using freecad for the tree at two steps 45/30mm with radius 1 mm and 300nm of torque mopment.
I don't really get your result. you have 300mpa but I would say it's wrong.
If you only have a torque moment, you only have 35.6mpa of tension that is what is also calculated by hand with the theoretical formula.

I took the tree where every diameter is 3 times long, so that it has a homogeneous behavior.

analyzing von mises, which is the most traditional approach that can be done, one has that the theoretical value tension is correct and one can read it quietly on the stretch 3 times the diameter.in the zone of the radius of 1 mm I made a mesh that divides in 4 the radius and the maximum value turns out to be 54.2mpa that is the tension that we expect on the connection radius.

the same assessment was made by evaluating the main stress:also here we have an intensification of the main tension that from zero passes to 60.1mpa which is the localized traction.

I can't understand, if each action has to be applied one at a time, as it is possible to have 300mpa tensions and kt coefficients greater than 2.
I must have missed something.

 

[meccanicamg]

using the worstps chart which is the maximum worst stress we then have as comparison values 60,1/35,6≈1,68 which could be the plausible value of kt made through fem.

 

[meccanicamg]

right to give an idea of how to apply a torque moment to the tree with freecad in the fem module you can look qui.
practically puts a force in the rectangle coordinate system (normal) and then turns the reference system into cylindrical (force of torque).

 

[bulfy98]

I was out of curiosity looking at one of your cases.

two diameter shaft, 40/35 with 1mm radius, 300nm torque.

then I went to take the curves on a specialized manual and then made a couple of accounts with the parameters.

The curves I used are this:View attachment 57295- I can't use it because with d/d=1.5 on the curve r/d=0,033 I don't have the curve... and so how do I invent it? I have to interpolate, invent, prolong etc.

and this:View attachment 57296- I can't use it because t/r can't be more than 4
- when I fell my eye on this I stopped calculating the various coefficients c

but how did you calculate the theoretical value kt?

The latter was exactly the file the teacher gave me. I had also noticed the question t/r>4, but I had to pass on it and I continued the calculations.

 

[bulfy98]

I tried to make a fem using freecad for the tree at two steps 45/30mm with radius 1 mm and 300nm of torque mopment.
I don't really get your result. you have 300mpa but I would say it's wrong.
If you only have a torque moment, you only have 35.6mpa of tension that is what is also calculated by hand with the theoretical formula.

I took the tree where every diameter is 3 times long, so that it has a homogeneous behavior.

analyzing von mises, which is the most traditional approach that can be done, one has that the theoretical value tension is correct and one can read it quietly on the stretch 3 times the diameter.View attachment 57300in the zone of the radius of 1 mm I made a mesh that divides in 4 the radius and the maximum value turns out to be 54.2mpa that is the tension that we expect on the connection radius.

the same assessment was made by evaluating the main stress:View attachment 57302also here we have an intensification of the main tension that from zero passes to 60.1mpa which is the localized traction.

I can't understand, if each action has to be applied one at a time, as it is possible to have 300mpa tensions and kt coefficients greater than 2.
I must have missed something.

I tried to make a fem using freecad for the tree at two steps 45/30mm with radius 1 mm and 300nm of torque mopment.
I don't really get your result. you have 300mpa but I would say it's wrong.
If you only have a torque moment, you only have 35.6mpa of tension that is what is also calculated by hand with the theoretical formula.

I took the tree where every diameter is 3 times long, so that it has a homogeneous behavior.

analyzing von mises, which is the most traditional approach that can be done, one has that the theoretical value tension is correct and one can read it quietly on the stretch 3 times the diameter.View attachment 57300in the zone of the radius of 1 mm I made a mesh that divides in 4 the radius and the maximum value turns out to be 54.2mpa that is the tension that we expect on the connection radius.

the same assessment was made by evaluating the main stress:View attachment 57302also here we have an intensification of the main tension that from zero passes to 60.1mpa which is the localized traction.

I can't understand, if each action has to be applied one at a time, as it is possible to have 300mpa tensions and kt coefficients greater than 2.
I must have missed something.

I do not understand, putting the formula of tau in calculator to me comes in different result: 56,6. The equivalent of von mises is sigma_eq=radq(3)*kt*(56.6) mpa. in my torsion calculations is never an effort of 300mpa.
from the table I have attached it is noted that the sigma of von mises with fem is about 180 mpa, against the 190 mpa calculated analytically despite t/r>4. 10 mpa of difference I do not know if they are many or few in this specific case, but in the traction case it is definitely greater the difference and therefore I care more.

 

[bulfy98]

right to give an idea of how to apply a torque moment to the tree with freecad in the fem module you can look qui.
practically puts a force in the rectangle coordinate system (normal) and then turns the reference system into cylindrical (force of torque).

on ansys I am thinking differently: I am applying the stresses on one of the two basic faces (the one with d minor) and the other extreme framed. in this way the stress should be the same on the whole body (the stress, not the tension).

 

[meccanicamg]

to me they seem high values anyway. You're right, for the torque moment you scored 190mpa... and not 300... but something doesn't come back.
first you should do the calibration....type round diameter 10 mm with 1000n force and see if the result returns. then if it works you switch to more complex models. Maybe there's a basic option that's wrong

 

[meccanicamg]

View attachment 57237same type of tree but with larger dimensions

I only checked the traction because I did the excel sheet only for that... but I don't get back kt and then the traction sigma.
for the fem I have not yet had time to verify it.
check first of all the theoretical values of tensions, so you are sure of value.... otherwise you can go crazy for nothing.

 

[meccanicamg]

How wrong? Good question. as reference document I attach you This is pdf that compares some work done with different programs.
Keep in mind two things:
- error of the fem compared to the theorist around 5% is normal and reliable, if around 10% it is necessary caution, if higher it is necessary to test laboratory and is running something
- each fem produces different results.

on the second point I can tell you something that happened to me at work on an analysis carried out on a carpentry frame under load:
- solidworks symulation made by me = overestimated 80%
- ansys made by customer and ansys germania = overestimated 100%
- freecad made by me = overestimated 40%
- experimental test with frame and comparisons made by me = true value

as you see... a disaster... if there is no comparison with a real test on physical piece.

other thing you have to keep in mind, since I have found no indications about the material.....if you exceed the value of the yield you are no longer in the elastic field and therefore you are obliged to do a non-linear fem analysis, otherwise you will find completely wrong values since linear analysis assumes that sigma/epsilon goes to infinite.

model elastic perfectly plastic and only elastic.

 

[meccanicamg]

I tried to make a fem using freecad for the tree at two steps 45/30mm with radius 1 mm and 300nm of torque mopment.
I don't really get your result. you have 300mpa but I would say it's wrong.
If you only have a torque moment, you only have 35.6mpa of tension that is what is also calculated by hand with the theoretical formula.

I took the tree where every diameter is 3 times long, so that it has a homogeneous behavior.

analyzing von mises, which is the most traditional approach that can be done, one has that the theoretical value tension is correct and one can read it quietly on the stretch 3 times the diameter.View attachment 57300in the zone of the radius of 1 mm I made a mesh that divides in 4 the radius and the maximum value turns out to be 54.2mpa that is the tension that we expect on the connection radius.

the same assessment was made by evaluating the main stress:View attachment 57302also here we have an intensification of the main tension that from zero passes to 60.1mpa which is the localized traction.

I can't understand, if each action has to be applied one at a time, as it is possible to have 300mpa tensions and kt coefficients greater than 2.
I must have missed something.

I realized now that I made a tree with a small diameter of 35mm and not from 30 mm....it is because I don't get the same theoretical numbers.

fact is that we are still out range of application of kt.però non ho capito perché hai 190mpa....I'm not even in the second tree.that I am coming... .
slowly we'll come out of this maze of numbers. Can you please check out what's going on analytical?

 

[meccanicamg]

for the bending I find myself with your theorist for the small tree but I am not for the big tree.
you sayma io col grafico e la her formula ho:It will be the late hour and a little tired but I don't understand where the hippo is.

 

[meccanicamg]

I made a traction fem on the tree 45/30r1....

with von mises I get as below:while evaluating monodirectional efforts according to the axis of the tree.. .while the theory sayswe are about 10% difference between fem and theory...but we know that the theory is approximated.

 

[bulfy98]

I only checked the traction because I did the excel sheet only for that... but I don't get back kt and then the traction sigma.
View attachment 57311for the fem I have not yet had time to verify it.
check first of all the theoretical values of tensions, so you are sure of value.... otherwise you can go crazy for nothing.

I thank you for your patience, very kind.
I find myself in everything except for the value c2 "used", to me comes -2.26 instead of-5.15. this leads to a different kt and therefore a sigma with different carving. I checked but I feel like I've put the formula right

 

[bulfy98]

h

I realized now that I made a tree with a small diameter of 35mm and not from 30 mm....it is because I don't get the same theoretical numbers.

fact is that we are still out range of application of kt.View attachment 57312però non ho capito perché hai 190mpa....View attachment 57313I'm not even in the second tree.View attachment 57314that I am coming... .
View attachment 57315slowly we'll come out of this maze of numbers. Can you please check out what's going on analytical?

I compared your theoretical data to mine for the bigger tree and noticed that I had made a damn calculation error. As for the torsion my stress is 50000nm, maybe we have a different order of magnitude. beyond this by applying the model of von mises tau tensions, in the calculation of the sigma of von mises, must be multiplied by a radq factor(3), at least by making me to the theory of prof. by adopting these changes our values combine.

 

[bulfy98]

update and synthesis:
- net of the correction I made on the calculation of the larger tree, all the theoretical kt values we compared match. However, I will have to point out to the prof that the t/r parameter is out of range in the 30/45 tree.
-I noticed that with regard to the big tree, for twist and bending we have an order of magnitude of difference for stress(I 50000 you 5000),multiplicando for 10 your results combine with mine.(always theoretical)
-in the calculation of the von mises sigma for torsion, the difference between our results should be equal to a radq(3) value which according to the theory of my prof, should appear for tau.
- comparing the fem to traction for the tree 30/45, my on ansys also produces about those results(330mpa) with a mesh quite thick.the theoretical value would be 289.4(both for me and for you).This difference is one of my major scratches and is why the prof made me the demands I sent a few days ago, but I really don't know how to please her.
Could it be a fem/theory limit?
-I have it for the 192/180 traction shaft.655 mpa theoretical against the 684 fem. instead the comparison theory/fem for that tree to bending and torsion is really great because they differentiate of 2-3 mpa.

 

[meccanicamg]

I thank you for your patience, very kind.
I find myself in everything except for the value c2 "used", to me comes -2.26 instead of-5.15. this leads to a different kt and therefore a sigma with different carving. I checked but I feel like I've put the formula right

perfect I found in my excel a wrong sign for c2. now matches -2,26

 

[meccanicamg]

update and synthesis:
- net of the correction I made on the calculation of the larger tree, all the theoretical kt values we compared match. However, I will have to point out to the prof that the t/r parameter is out of range in the 30/45 tree.
-I noticed that with regard to the big tree, for twist and bending we have an order of magnitude of difference for stress(I 50000 you 5000),multiplicando for 10 your results combine with mine.(always theoretical)
-in the calculation of the von mises sigma for torsion, the difference between our results should be equal to a radq(3) value which according to the theory of my prof, should appear for tau.
- comparing the fem to traction for the tree 30/45, my on ansys also produces about those results(330mpa) with a mesh quite thick.the theoretical value would be 289.4(both for me and for you).This difference is one of my major scratches and is why the prof made me the demands I sent a few days ago, but I really don't know how to please her.
Could it be a fem/theory limit?
-I have it for the 192/180 traction shaft.655 mpa theoretical against the 684 fem. instead the comparison theory/fem for that tree to bending and torsion is really great because they differentiate of 2-3 mpa.

traction behavior should be theoretically the best but obviously not so with this kind of carving. You should try to inflate the mesh and see if the value is converging or not.

 

[bulfy98]

traction behavior should be theoretically the best but obviously not so with this kind of carving. You should try to inflate the mesh and see if the value is converging or not.

I initially thought it was a problem of connection radius too small, but the same anomaly is repeated also in the d192 tree with r=4mm. Infixing the mesh the value remains quite overestimated compared to the model, varying only 8/10 mpa compared to the value I found with my initial mesh.

my prof had advised to use a rectangle mesh, but I don't know if this changes effetivament and things.
However I have calculated the nominal values (without carving) with the fem and are practically identical to the theoretical ones (the greater variation is 0.8 mpa), so as to be able to calculate the kt of the fem.

 

[meccanicamg]

probably the speech is that in proportion the connection radius is small and in the curves is in the area where it tends to go to infinite the kt and probably there are difficulties of convergence. Therefore the type of mesh, shape and element, size and algorithm of convergence could give different results.
If you can do the test with rectangular mesh.
I would instead try not to use elements of the second order but of the first order and see if the tensions are lowered. in my practical experience I have seen that the first order despite rougher and rigid often approaches practical feedback.

 

[bulfy98]

probably the speech is that in proportion the connection radius is small and in the curves is in the area where it tends to go to infinite the kt and probably there are difficulties of convergence. Therefore the type of mesh, shape and element, size and algorithm of convergence could give different results.
If you can do the test with rectangular mesh.
I would instead try not to use elements of the second order but of the first order and see if the tensions are lowered. in my practical experience I have seen that the first order despite rougher and rigid often approaches practical feedback.