errore fe analysis

[rapa]

I ask those who use the fe analysis module. Once again I got a discrepancy between calculation and real measure of the shift using this program for deformation verification in carpentry structures.value calculated 1.6mm measured value 9mm! error of 5 times.
I have verified all the data! I don't know what to think! I wanted to know if any of you use this software and if you have found these inaccuracies. Thank you.

 

[ragnol]

If I understand correctly, you're experiencing this difference between how simulated and the real behavior of the structure, right?
can depend on differences between how modeled and the real situation? For example, lowerings due to the arrangement of any connections, perhaps considered indeformable? this regardless of the software used. . .

 

[cacciatorino]

If I understand correctly, you're experiencing this difference between how simulated and the real behavior of the structure, right?
can depend on differences between how modeled and the real situation? For example, lowerings due to the arrangement of any connections, perhaps considered indeformable? this regardless of the software used. . .

besides this much does also modeling (hexaedric mesh, or tetrahedra, linear or square, fixation where it serves etc).

 

[rapa]

If I understand correctly, you're experiencing this difference between how simulated and the real behavior of the structure, right?
can depend on differences between how modeled and the real situation? For example, lowerings due to the arrangement of any connections, perhaps considered indeformable? this regardless of the software used. . .

This is a structure consisting of welded sheets subject to simple bending by hydraulic cylinders. elastic deformation

 

[Kaji]

unfortunately I have never used a fem form; modeling is based on nastran so you could also ask those who use nastran on other cads and maybe they can give you an answer.

 

[Onda]

you should let us understand how you idealized the structure and how it was then realized. Normally such large differences are due to errors, they can be in the type of bond, in the links between the parties, in the way it was loaded. it is not possible to give a priori answer without trying to understand the model. in general an error of more than 5 times is not attributable to the type of elements and mesh, especially when it comes to deformations and not stress.
on deformations, even coarse mesh and unsuitable elements give relatively correct results. then if you look for maximum stress in a detail, it is another speech.

 

[rapa]

you should let us understand how you idealized the structure and how it was then realized. Normally such large differences are due to errors, they can be in the type of bond, in the links between the parties, in the way it was loaded. it is not possible to give a priori answer without trying to understand the model. in general an error of more than 5 times is not attributable to the type of elements and mesh, especially when it comes to deformations and not stress.
on deformations, even coarse mesh and unsuitable elements give relatively correct results. then if you look for maximum stress in a detail, it is another speech.

attach the frame image frame where it supports a raddle with the two symmetric forces due to the cylinders. use press type for wire.mesh matax automatic made without particular compensation or concentration of points but tetrahedra and square precision.

 

[Onda]

what is posted is not enough, you need to see how you have bound, how you applied the loads, and then photos of the real structure, to try to understand a correspondence.

 

[cacciatorino]

excuse can I know the size of the structure? If it's 10 meters long, there's such uncertainty if it's 500 mm long, obviously not.

 

[rapa]

excuse can I know the size of the structure? If it's 10 meters long, there's such uncertainty if it's 500 mm long, obviously not.

Total size in length about 2300mm = 1150mm x2 symmetric

 

[meccanicamg]

if it leans on a raddle it is necessary to delimit the area of the ralla with a surface.... otherwise it also reacts the rest that it does not have.
This may be the mistake. however are errors of bond and application forces.

 

[rapa]

if it leans on a raddle it is necessary to delimit the area of the ralla with a surface.... otherwise it also reacts the rest that it does not have.
This may be the mistake. however are errors of bond and application forces.

in theory the system is symmetrical, if there was a rotation following the ralla it would be seen in the practice.ivece who made the measurements says that there is a small difference of displacement between dx and sx following stiffness (drawing)not just symmetrical

 

[Stan9411]

you have to post a scheme of application of constraints and forces..... to gossip it is impossible to give you a hand.. here we talk about the result of a fem analysis as if we were at the bar.

 

[rapa]

you have to post a scheme of application of constraints and forces..... to gossip it is impossible to give you a hand.. here we talk about the result of a fem analysis as if we were at the bar.

allego schematic design with quotas indication forces and bond and knit size indicated with a small pyramid.I can not indicate more for extreme manic protection of the designs in this company

 

[Stan9411]

the circular surface that in the incastrious analysis, in the reality how is it bound?

 

[meccanicamg]

allego schematic design with quotas indication forces and bond and knit size indicated with a small pyramid.I can not indicate more for extreme manic protection of the designs in this company

to understand the order of magnitude you have to do by hand the calculation of the bending as in these posts qui.
In this way you have a term of comparison between the two fems.
surely one of the two is not correct or is made with very different hypotheses.... material, constraints, linearity or not, mesh, application forces etc.

 

[cacciatorino]

In this way you have a term of comparison between the two fems.
surely one of the two is not correct or is made with very different hypotheses.... material, constraints, linearity or not, mesh, application forces etc.

If I don't remember the discrepancy is not between two fems but between a fem and an experimental measure on the real structure.

 

[Stan9411]

It's morning... ?

 

[Stan9411]

the most sensible thing is that the bond in reality somehow has a much greater yieldability than an ideal ink .. for this I asked how in reality that ralla is fixed to the rest of the world.

if it is not that, you pass to the trivial errors .. one zero less in the module of young, one more in the load, etc etc.

 

[rapa]

the most sensible thing is that the bond in reality somehow has a much greater yieldability than an ideal ink .. for this I asked how in reality that ralla is fixed to the rest of the world.

if it is not that, you pass to the trivial errors .. one zero less in the module of young, one more in the load, etc etc.

Thanks I'll check everything. the software assistance has shocked me to do the calculation with decreasing mesh sizes to see if the result of displacement is different than a few percentages. . I'll try it if I don't.