Cardiovascular diseases are major contributors to mortality, with valvular disease being a primary precursor. A valve failure is common and can lead to a decline in pumping function, potentially resulting in cardiac arrest. Conducting in-vivo experiments on human heart valves, whether native or artificial, poses significant challenges. However, computational fluid dynamics (CFD) combined with fluid structure interaction (FSI) provides an effective and cost-efficient means of investigating the haemodynamics of both natural and artificial heart valves. This approach offers valuable insights for diagnostic and clinical applications. In this study, we developed a haemodynamic model of the mechanical heart valve using smoothed particle hydrodynamics (SPH). The model was successfully validated against traditional finite volume methods (FVM) and experimental data. We demonstrated that SPH is well-suited for simulating heart valve function due to its Lagrangian description of motion, which is particularly advantageous for FSI. Furthermore, to enhance the clinical relevance of the model, we have conducted three case studies of the mechanical valve malfunction due to leaflet restriction resulted by pannus and thrombus formation. The proposed SPH/FSI technique presented here provides a unique and valuable tool for accurately modelling the transient haemodynamic behaviour of the malfunctioned bi-leaflet heart valve under different velocity phases.