Structural safety components of motor vehicles are designed to absorb the impact energy released during crash events in order to minimise the destructive consequences of unavoidable collisions. Due to the remarkable combination of price, density, and toughness, mineral-filled polymers have become the candidate material to manufacture automotive safety components. To predict their dynamic and inelastic mechanical response, experimentally and computationally expensive constitutive models have been implemented in finite element solvers, i.e. SAMP1 material model in the explicit finite element solver LS-DYNA. However, they struggled to be used in automotive applications due to their extensive experimental and numerical costs. The research project aims to develop an efficient and robust material model that is capable of simulating the mechanical behaviour of mineral-filled polymers when subjected to multi-axial loads at different strain-rates. To achieve this goal, the mechanical properties of a widely-used injection-moulded talc-filled Polypropylene compound were experimentally investigated and numerically simulated in LS-DYNA at quasi-static and dynamic strain-rates. In order to obtain accurate numerical simulations, reliable mechanical properties need to be extracted from material characterisation tests. The repeatability of uniaxial tensile tests was found to be improved when dog-bone shaped specimens were machined from injection-moulded plaques using the milling machine. Since no standardised methodology exists for non-fibre reinforced polymers, the repeatability of the in-plane shear test was also investigated. The modified Wyoming Iosipescu test exhibited the most reliable stress-strain results due to the consistent shear strain uniformity across the specimen during the experiment. A novel post-processing methodology was developed to accurately extract the stress-plastic strain curve that does not introduce numerical instability in the simulation of the uniaxial tensile test. A new definition of equivalent plastic strain was introduced in the constitutive equations to take into account the large volume dilatation that characterises mineral-filled polymers. Compared to SAMP1, the new constitutive model improves the numerical prediction of the material flexural behaviour by 20 % while reducing the simulation running time by 60%. When the bi-axial properties of the material are introduced in the constitutive equations using a reverse engineering methodology, the material model improves the numerical accuracy of the perforation test by 75 % for the quasi-static test and 65 % for the dynamic test. Since it reduces the simulation running costs by 60% for both tests, the new material model constitutes an efficient tool for CAE design engineers that want to simulate the dynamic and multi-axial response of structural components made of isotropic materials.