The commonly used grasp simulators such as GraspIt! and OpenRAVE use wrench space formulations and grasp quality metrics such as ϵ and v to identify stable grasps in dynamic conditions. However, wrench space formulations are derived based on static mechanical equilibrium equations, and the physical attributes of the object such as stiffness and mass are also not considered for grasp analysis. Above all, these grasp analysis frameworks cannot be experimentally validated, thereby resulting in grasps that are not reliable. In this paper, an experimentally validatable Finite Element (FE) based grasp analysis framework is proposed for evaluation of robotic grasps in dynamic conditions. By applying standard solutions of Hertzian contact theory to a few robotic grasps, Finite Element Method (FEM) is validated. A real-world grasp situation is then simulated using FEM, and the FE model is validated based on the contact area measured in real-time using a pressure sensor. By applying dynamic perturbations to the validated FE model, the stability of the grasp is evaluated, and the most stable grasp is identified using the contact area based metric, π. It is observed that FE simulations agree with the analytical solutions and experimental results, with a relative error of not more than 7%.