This contribution presents ongoing research taking place at the AS2M Department of the FEMTO-ST Institute, specialized in micro-/nano-manipulation and microrobotics. Within the context of the PEPR O2R project, our work aims to develop physically meaningful and scalable computational models for liquid actuators and mechanical joints (i.e., two solids linked together by a fluid drop), with a special interest in deformable microrobotic applications. At the microscale, surface tension becomes the dominant force that dictates the dynamics. Following previous experiences with a quasi-static surface solver (Surface Evolver), our team identified the need for a more versatile simulation method capable of capturing dynamic behaviours in order to simulate and optimize new design ideas developed by our group. These include liquid transistors, liquid manipulation tips, liquid pistons and compliant joints. The Smoothed Particle Hydrodynamics (SPH) method has been chosen as the simulation technique for some noteworthy reasons. Firstly, SPH naturally handles both the mechanical behaviour of solids, soft bodies and fluid dynamics phenomena, promising effective coupling between different forces and constraints. Secondly, the meshless nature of SPH allows for topology changes and large deformations, convenient for soft microrobotics. Lastly, SPH provides flexibility to introduce diverse models for different phenomena concerning surface tension: cohesive forces, contact angles, contact angle hysteresis, triple line forces, etc. This modularity is particularly useful, given that it allows customizing the simulation to capture specific effects relevant to the system at hand. Following a literature review, our current work focuses on implementing and benchmarking various selected models within the open-source simulation software SPlisHSPlasH. The first goal is to replicate relatively straightforward theoretical and numerical results to establish a baseline. Subsequently, we intend to challenge the static and dynamic results of these models with more complex scenarios at various scales and resolutions, which will likewise be recreated experimentally for validation.