Sheldon Imaoka
Ansys Employee

COMBIN39 is just a node-to-node element. Thus, the node is not directly associated with an area or volume - that is why we use force vs. deflection in its definition. It is just basically a spring, like K*x=F, similar to the FE matrix equation we solve [K]{x}={F}.
When you have solid elements, for example, the integration point for the constitutive relationship is associated with a 'volume', so that is why you end up with stress vs. strain definition instead. From the constitutive relationship (stress vs. strain) and element properties, we go to a form of [K]{x}={F}. So your comment is reversed - we do not go from COMBIN39 force-deflection curve to stress vs. strain but do the opposite; everything is transformed to stiffness [K], deformation {x}, and forces {F} to solve as a matrix equation.
If you have traction vs. separation (bond-slip), did you intend to use something like a cohesive zone model? If your reinforcement is modeled as a beam element (single node describing cross-section), then the COMBIN39 is suitable, and you need to estimate the area associated with a node to convert the bond-slip relationship to force-displacement. If, however, your reinforcement is a solid, then you may want to look at the cohesive zone model instead (see Fracture Analysis Guide, Sections 3.5 and 3.6 on modeling interface delamination). The cohesive zone model is defined as a traction-separation law (units of stress vs. displacement, such as MPa - mm).