[hussan]

I would like to know, how Abaqus calculate deflection for different beam profile.
I have calculated the deflection of I and T-beam with same dimension and also analytically but all of the results are different.

Could anybody explain to me why I am getting different results, while comparing analytical and numerical result. Difference is maximum 5%. As for the simple supported T-beam and analytical result, there should be no difference.

Maximum bending stress difference in the case of T-beam analytically and numerical is 2.93%. There should be no difference but I got this difference.

[karachun]

No one knows how you perform your calculations. What mesh you use, what constraints and loads you use, etc.
First give a description of your model. Then you can expect to receive some help.

[hussan]

I have attached the file in which I have done the comparison. You can have a look.

If you have any suggestion/recommendation/solution to this problem, I am waiting for your reply.

Thank you in advance.

[karachun]

Very strange. You got good results using Beam+Plate mesh and not so accurate results using only T-Beam.
I ran a beam model in linear and nonlinear solutions in Nastran (with a distributed load of 5 N/mm) and got max vertical deflection of 404.4 and 399.0 mm. Therefore we can exclude effects of geometric nonlinearity.
Try to double check the beam cross section and load type.
I`m not an ABAQUS user but wher I export my mesh to ABAQUS format using FEMAP I noticed that my model uses beam type B31 while you use B33. Those numbers don't tell me anything but this may be a clue.
I have attached a .inp file. Try to export it, edit some settings to make the model work and try to solve it. Model units are m, s, kg.

[hussan]

Thank you for your reply.

I have attached the .inp file, could you please have a look. B33 (Bernoulli beam), B31(Timoshenko beam). B31 and B33 create a very small difference (for example deflection (with B31) is 418.3 and with B33 is 418.1)

Thank you very much

[DanielSP]

Beam models are approximations that need the Navier hypothesis to be correct (among others). Plate models will deviate from classic predictions on account of the effect of large loads at the supports (stresses and strains will only be linear in a cross section away from concentrated loads) and the fact that plates themselves bend locally.