Interfacial delamination in polymer coated metal sheet : a numerical-experimental study

An increasing amount of products are nowadays made of polymer coated metal sheet. Polymer coated metal has several advantages compared to traditionally Sn (tin) coated metal, such as costs savings and a more environmental friendly production process. Beverage and food cans are formed by draw-redraw and wall iron processes. These forming processes take place under extreme processing conditions and may result in delamination of the polymer coating. Delamination leads to a loss of attractive and protective properties of the product and must therefore be prevented. If delamination can be predicted, the processing routes, parameters and tooling can be adjusted a priori to prevent it. This work relies on an integrated numerical-experimental approach to achieve these accurate predictive capabilities. A successful numerical-experimental approach calls for an accurate numerical description of the interface and delamination process. Cohesive zones are known to be a suitable tool to describe interface interaction. The often used exponential cohesive zone law of Xu and Needleman is evaluated, whereby further improvements have been revealed to describe a mixed-mode delamination process. An alternative cohesive zone law is proposed to overcome the identified shortcomings and it is shown how the parameters can be determined by dedicated experiments. Experimental observation of the interfacial delamination in polymer coated steel shows that the delamination process takes place by fibrillation. Additionally, large displacements and deformations are present at the interface and in the surrounding bulk materials. Most cohesive zone models require a local coordinate system to decompose the traction and opening displacement into a normal and tangential contribution. The orientation of the local coordinate system cannot be chosen unambiguously and may result in significant differences in normal and tangential contributions when large displacements are present at the interface. Therefore, a cohesive zone model with a large displacement formulation is developed to overcome this ambiguity by defining a direct constitutive relation between the traction and opening displacement vector. Because the polymer and the steel have substantial different material properties, a peel test is found to be one of the few suitable experimental set-ups for characterization of the interface between the coating and the substrate. The coating thickness is measured with an optical imaging profiler and it is shown that there are no residual stresses present in the coating. Full characterization of the interface requires zero and ninety degree peel tests in conjuncture with accurate simulations of the peel tests and an inverse parameter identification procedure. Plane-strain and 3D models, to be solved in a finite element solution context, are built. The peel tests induce large deformations in the polymer coating, which are properly accounted for by a non-linear visco-elastic constitutive model. Attention is given to the intrinsic stress-strain response, the strain softening behaviour and the strain rate dependency of the coating, and the corresponding experimental characterization techniques. A mixed-mode cohesive zone model with a large displacement formulation is developed to discriminate between opening modes and their possible different work-of-separation. Zero and ninety degree peel tests are performed and the interfacial work-of separation is determined by an inverse parameter identification procedure. It is shown that a plane-strain geometry is a good approximation for the modelling of peel tests, which significantly reduces the computational time of the simulations. Deep-draw and redraw forming processes induce large plastic strains in the materials. Plastic strain leads to roughening of the substrate and consequently to a loss of adhesion of the coating. From a surface height profile, the roughness is determined by a height-height correlation function. This results in a root-mean-square roughness parameter and a correlation length, where the latter is an indication of the typical length scale involved in roughening. An orientation dependent correlation function is formulated to provide anisotropic roughness and correlation length parameters. The initial surface topography of unpolished specimens leads to an offset of the roughness, whereas the evolution of the roughness as a function of the strain is similar to the one found in polished specimens. The correlation length is related to the underlying microstructure of the substrate and develops in an anisotropic fashion during pre-straining. The description of the cohesive zone model is extended to account for the loss of adhesion due to substrate roughening by implementing a relation between the work-of-separation and the effective plastic strain. Stress relaxation and ageing take place in the time period between the pre-straining of the specimens and the start of the peel test. Accurate characterization of the interface of pre-strained specimens therefore requires the appropriate modelling of the stress relaxation and the ageing by means of employing a multi-mode version of the constitutive model and the incorporation of ageing kinetics, in addition to the determination of the involved material parameters. With the adopted single-mode model without ageing kinetics, the interface of pre-strained specimens is characterized by assuming a fully aged state of the material before the initiation of the peel test. The combined numerical-experimental method is used to quantify the loss of adhesion due to pre-straining and the results show a clear deterioration of the adhesion as a function of the plastic strain in the substrate. Finally, the experimentally characterized cohesive zones and constitutive models are adopted in an industrial deep-draw model. The simulation results are assessed with an adequate variable that quantifies the interfacial integrity. Several process parameters are varied to study their influence on the deterioration of the interface during deep-drawing. The results point to the most relevant process parameters and the critical areas where the interfacial damage and the boundary conditions may trigger delamination.