Abaqus analysis of concrete slab deformation non logical result

[Mlindblom]

I have modelled a concrete slab in Abaqus and i get a big difference between my hand calculations of the deformation (which i am pretty sure are correct) and my deformation results from Abaqus.
I am pretty sure there is something wrong in the model.

Input data:

L: 8m
W: 2.4 m
H: 0.2
density: 2300 kg/m3
Young modulus: 33e+9 Pa
Poisson: 0.2
Uniform load: 3500 N/m2
Boundary conditions: locked in Z at one end and locked in Z and x in the other end

I get a max midspan deformation of about 0.68 m in Abaqus and 0.068 m in my hand calculations. There seem to be an additional multiplier of 10 somewhere in Abaqus, or the correlation between the results may just be a coincidence. Any guidance on this problem would be greatly appreciated.

 

[FEA way]

Can you share a screenshot showing your Abaqus model with boundary condition and load symbols ? The description leaves some doubts.

 

[Mlindblom]

Thank you for your reply.

I attach 3 images showing load and bc symbols and values. Is it something i should change?
My hand calculations showed a max midspan deflection of 3.5 mm, my Abaqus model show 700 mm... Something is horribly off. Any ideas?

 

[FEA way]

I get the deflection of 0.00848 m from analytical calculations and this is in agreement with analysis. You should definitely refine the mesh, especially add more elements in the thickness direction. Also, double-check all the inputs.

 

[Mlindblom]

Can you show me the analytical solution where you get 0.00848? From my analytical solution i get 0.0035 m.

I also tried modelling the floor in 2d with the same properties and i got the exact same deflection as in my analytical solution, 0.0035 m.

I have double checked all my inputs and refined the mesh, thanks for the tip. Now i am getting about 0.009 m in max deflection. Could finer mesh move the deflection closer to 0.0035?

Thanks again for your help. Really greateful.

 

[Mlindblom]

Oh wait,
i forgot to multiply my load with the width of the floor in my analytical solution. Now i changed it to: u=(5*qtot*L^4)/(384*E*I), where qtot is 3500*2.4. With that correction i get 0.00848, which Abaqus now shows with a finer mesh.