Epoxy Adhesives, Correlation between Strain Whitening and Damage Using a Macroscopic Optical Method
MetadataShow full item record
The success of lightweight automotive multi-material assemblies depends on selecting appropriate joining techniques that can provide the expected day-to-day operational strength while delivering occupant protection during crash scenarios and long-term durability performance. Epoxy-based adhesives provide an important joining method to increase structural stiffness and enable the joining of dissimilar materials for multi-material assemblies. However, the design of adhesive joints requires mechanical data to support integration in vehicles and computer aided engineering design. The objective of this research was to address a deficit in the identification and quantification of damage in epoxy adhesive materials under applied loading, which is critical for constitutive models that can be used in the numerical representation of structural epoxy adhesive materials. Three structural epoxy adhesive formulations: a non-toughened single-part epoxy (EC-2214, 3M); a two-part toughened epoxy (DP-460NS, 3M); and a high toughness single-part epoxy (SA-9850, 3M); were tested to failure under tension and shear loading conditions over a range of strain rates (0.002–100 s-1). The measured mechanical data was implemented in constitutive models using two approaches (cohesive zone model and continuum solid element formulation) and verified using finite element simulations of the experiments. This study provided understanding regarding the mechanical response of structural adhesives to loading and the relationship between shear and tensile strength, differences in non-recoverable mechanical response, and the mode of failure for different adhesive formulations (localization and brittle failure, development of strain whitening and ductile failure). Adhesives strength increased with increases in strain rate for all three materials, and limitations in current modeling approaches such as the use of a von Mises yield surface and assuming coupling of strain rate effects between different modes of loading, were identified. Importantly, strain whitening was observed on the surface of the specimens during testing, and exhibited varying intensity and distribution depending on the strain rate and material type. A paucity of damage data for structural adhesives was identified in the literature. This information is necessary to define or enhance failure criteria in finite element simulations, which in turn can improve the physical representation of adhesive materials in numerical simulations. A follow-on study investigated the viability of using Vickers microhardness measurements as a forensic technique to quantify damage in structural adhesives. The study used tensile specimens machined from bulk adhesive and tested to failure over a range of strain rates (0.002-100 s-1). Pre-test reference microhardness measurements were compared to post-test hardness measurements along the gauge length of the test specimens. The changes in microhardness were used to indirectly measure damage in the materials. In general, for toughened epoxies, the damage extended over much of the sample gauge length, while the un-toughened epoxy demonstrated damage localization at the failure location. Increasing strain rate led to an increase in the damage localization for a given material. Out of the three tested materials, the two-part toughened epoxy (DP-460NS) demonstrated the most complex behavior during the straining process including variations in microhardness with strain rate, development of strain whitening with load, and further evolution towards shear banding at high levels of deformation. Although microhardness did provide a reliable method for damage measurement, the procedure was not practical to obtain continuous strain-damage data, as is required for material constitutive models. However, the microhardness data support the premise that strain whitening in the tested specimens was associated with the measured microhardness changes and therefore damage in the toughened epoxies. A follow-up study using the two-part toughened adhesive (DP-460NS) was conducted to further understand the nature of the strain whitening process and its connection to damage through microscopic observations of the material surface during loading. This study also established damage reference values using traditional techniques (changes in stiffness and strength after load-reload) and determined that the observed changes in color (strain whitening) were linked to changes in the morphology of the surface in the strained material. Microscopic observations identified that the morphological changes caused by increases in tensile loading were due to the development of crack opening, cavitation, growth of plastic zones around cracks, and later the development of shear bands. Although the morphological change (21% change in the amount of pixels that describe the morphology of the free surface) was comparable to damage values calculated using traditional techniques (19% from changes in modulus, and 18% using changes in strength), the implementation suffered from the same shortcomings that affected the use of microhardness. That is, impracticality to obtain continuous damage data over the strain history of the material and limitations resulting from the small area observed in the material using the microscope. The previous studies led to the development of a macroscopic optical technique to quantify the evolution of damage in real time. The technique used images captured during tensile testing to assess damage through the change in average color on the material surface with strain. The two-part toughened epoxy was used to asses the implementation. The results were compared against damage data from the previous studies using the same material and damage calculated using other reference techniques. The reference techniques included volumetric strains, changes in modulus of elasticity, and changes in microhardness. Damage measurements from the optical method ranged between 15 and 25% at failure, which agreed (15 to 21%) with the reference techniques (microhardness changes, modulus changes with load-unload and microscopic observations). There was a difference between the damage predicted using changes in volumetric strains (8%) and all other measures of damage. It was hypothesised that the lower value was associated with the volume conserving nature of the shear banding deformation process. In other words, any damage that can occur in parallel with or that can be associated with the shear banding process, was captured by all other techniques (changes in microhardness, changes in modulus of elasticity, microscopic images, change in color) while the volumetric strains fail to capture this contribution to the overall damage in the material. This was due to the lack of detectable changes in volume resulting from the shear banding process. In addition to the numerical agreement between optical damage values with the reference techniques (except for volumetric strains), the implemented optical method can predict the location of the actual fracture zone, quantify the damage level at different locations along the area of analysis, besides providing a continuous strain-damage curve. Damage measurement using optical measurement of changes in average color constitutes an accurate and robust experimental technique for structural adhesives that offers a new method to identify and quantify damage evolution in polymeric materials exhibiting strain whitening. The proposed technique can provide strain-damage curves, which are much needed information for the implementation of constitutive material models for structural adhesives and other polymeric materials. Although the method is limited to strain whitening materials, the measurement can be implemented for testing in tensile loading at any strain rate as long as a suitable image-capturing device is used.
Cite this version of the work
Luis Trimiño (2019). Epoxy Adhesives, Correlation between Strain Whitening and Damage Using a Macroscopic Optical Method. UWSpace. http://hdl.handle.net/10012/14493