Hi, there!

I'm looking to validate an experimental uniaxial stress-strain curve of a direct tension test on a concrete cylinder using the Concrete Damaged Plasticity model in Abaqus. The material data input follows both manual and literature instructions, with a choice of stress x crack-opening displacement relation for the tension stiffening behaviour.

For a '1D' model using bar elements to represent the concrete cylinder as a "pillar", the simulation mean outputs for stress and strain are a near-perfect match to the corresponding stress-strain curve for the input data (-> this curve was derived with Cornelissen et al. exponential law together with the characteristic element length outputted by Abaqus in the message file to go from crack-opening displacements to post-peak strain values).

However, results from 2D and 3D simulations of the 'same' concrete cylinder do not match their corresponding input stress-strain curves in the least.

I have uploaded an image to better show the divergences.

Has anyone encountered something similar and would know of anything that could be the cause of this?

I appreciate any help!

Thanks!