Behaviour and modelling of two semi-crystalline polymers over a wide range of rates and temperatures
Doctoral thesis
Permanent lenke
https://hdl.handle.net/11250/3064466Utgivelsesdato
2023Metadata
Vis full innførselSamlinger
Sammendrag
Through this thesis, I seek to explore how to characterise and simulate the strain rate and temperature dependent material behaviour of semi-crystalline polymers. To this end, a cross-linked polyethylene and a polypropylene are subjected to quasi-static and dynamic conditions in a combined experimental and numerical study. Uniaxial tension tests have been performed over a wide range of strain rates and temperatures utilising universal test machines, split-Hopkinson bars and climate chambers. The test set-up has been investigated through finite element simulations replicating the physical experiments. Two material models, which are fundamentally different, have been applied and compared to determine the degree of complexity which is needed for finite element simulations of polymer components. The thesis consists of three closely related parts that explore an experimental test set-up for high-rate tension testing of polymers, the behaviour of two materials and possible material models for simulating polymers. Preceding the main parts, the overall motivation, objectives and structure of the thesis are provided. Then, tying the work together, concluding remarks and suggestions for further work are provided. Finally, an appendix is included, incorporating an extended version of one of the material models.
Part 1 of this thesis presents a combined experimental and numerical study of a small-diameter steel split-Hopkinson tension bar. The purpose of the study is to investigate the feasibility of using a finite element model as a digital twin of a complex physical experiment to investigate potential sources of uncertainty. The split-Hopkinson bar is the most common apparatus for performing high rate material characterisation tests, but the application to polymers is associated with some challenges that are well suited for analysis with such a finite element model. Both the cross-linked polytethylene and the polyproplylene are subject to this study, but the former is reported in greater detail since it is a more challenging test case due to its comparably low strength and high ductility. The finite element model, which incorporates the input bar, a test specimen and the output bar, showed good correspondence with physical tests. The model could thus be used to evaluate details otherwise not available solely from the experiment. Combining the experimental data and the numerical model it is demonstrated that the influence of measurement noise in the force recording was quantifiable and acceptable. Furthermore, the test specimen is shown to be in dynamic equilibrium for the majority of the test, despite contrary indications. Following this study, the split-Hopkinson bar could be used with confidence for experiments in the subsequent studies.
In Part 2 the split-Hopkinson tension bar and a universal test machine are utilised to perform uniaxial tensions test with the cross-linked polyethylene and the polypropylene at strain rates ranging from 10−2.5 s−1 to 103 s−1 and temperatures ranging between −30◦C and 80◦C. Transparent climate chambers employing liquid nitrogen and thermal elements were used to obtain lowered and elevated temperature respectively. The transparency of the chambers allowed for local surface deformation measurements using digital image correlation which was essential due to the early onset of necking in the tensile specimens. A complex strain rate and temperature sensitivity were observed for the flow stress of the material, which was found to increase at high rates or low temperatures. This trend was significantly better represented by a two-process Ree-Eyring flow model than a Johnson-Cook rate and temperature dependency, which is more often used for models that are available in commercial finite element codes.
Part 3 of the thesis presents a comparison study regarding the performance of two material models for simulating the large strain behaviour of the cross-linked polyethylene under monotonic loading over a wide range of rates and temperatures. The first material model serves as a pragmatic but efficient option that was originally developed for metals, while the second is a physically motivated model that was developed specifically for polymers. The input parameters of the material models were calibrated from the uniaxial tension test data presented in Part 2. These tests were then compared to finite element simulations, serving as a first-step evaluation of the rate and temperature dependent features of the two models. A series of pipe compression tests were then performed experimentally and numerically at different rates and temperatures to serve as a validation case independent of the material calibration. This study showed that the physically motivated model did indeed perform better than the phenomenological model at a wider range of test conditions, but also that the latter could become an effective alternative with only minor modifications.