This example is going to compute the reaction force on the top half arm at the left end in the

peeling test (Fig. 1). The problem is that the cohesive behaviour seems to be off when I

change the material from neo-Hookean to Holzapfel-Gasser-Ogden. It is so strange because I

think the cohesive behaviour for the contact-surface pair should be independent with the

material model of bulk. Please find the details about the problem as follow.

1. Models

Geometry: width 4, height 1.2, initial crack 0.4.

Boundary conditions: the top and bottom half arm has displacement boundary condition

u_2=4 and u_2=-4, respectively. The right end is fixed.

Materials: incompressible neo-Hookean and Holzapfel-Gasser-Ogden.

Crack propagation is modelled by the surface-cohesive behaviour. The damage criterion is

the maximum nominal stress.

2. Results

Five experiments are done to demonstrate the problem. Here are their features:

Problem 1:

When I change the value for Tc, I would like to get different results, including the deformed

configuration and reaction force. For experiments in group 1 with neo-Hookean material, it

is true (Fig. 1-3). But it is wrong when I only change the material to HGO; I got same result

for Exp 2a and 2b.

What I can think of is that the cohesive behaviour is off in both Exp 2a and 2b. That means

what the simulation in group 2 becomes to compute the bending of two individual beams.

Another supporting observation is that the stresses at the ‘interface’ are always zero (Fig. 4).

Could you please help on it?

Problem 2:

The deformation in Exp 1c is strange (Fig. 5). The crack propagation started from the inside

interface rather from the end of the interfaces, where the crack tip is at. To compare with

that Exp 1a and 1b, this problem is because of bigger value of Tc. But actually, the Tc I used

is still smaller than some references.

Could you please give some suggestion?