Soft pneumatic actuators (SPAs) are widely employed to drive soft robots. However, their inherent flexibility offers both benefits and challenges. This property reduces their output force/torque and makes them hard to control. This paper introduces a new design method that enhances the actuator's performance and controllability. The complex structure of the soft actuator is simplified by approximating it as a cantilever beam. This allows us to derive a mechanical equation between input pressure to output torque. Additionally, a dynamical model is explored to understand the correlation between the natural frequency and dimensional parameters of the SPA. The design problem is then transformed into an optimization problem, using the mechanical equation as the objective function and the dynamical equation as a constraint. By solving this optimization problem, the optimal dimensional parameters are determined. Prior to fabrication, preliminary tests are conducted using the finite element method. Six prototypes are manufactured to validate the proposed approach. The optimal actuator successfully generates the desired force/torque, while its natural frequency remains within the constrained range. This work highlights the potential of using approximated models and optimization formulation to boost the efficiency and dynamic performance of soft pneumatic actuators.