Abstract
Development of constant force non-linear softening (CF-NLS) springs has recently gained attention in the literature due to their energy storage potential in many applications including robotics, biomechanics, machining, etc. These springs are typically designed by using computationally exhaustive topology optimization techniques which have shown to produce stress concentrations significantly reducing the operational life of the spring. Furthermore, current design methodologies including spline based optimizations are not exploiting the design space efficiently. There is a need of a computationally efficient design methodology capable of producing mechanisms without stress concentrations while more fully exploiting the design space. A new graph based design methodology is proposed by using a modified-depth first search (M-DFS) algorithm with 2D finite element analysis. A detailed parametric study is also conducted to design a mechanism by maximizing of the stored strain energy on a general stiffest-softening behavior. The proposed CF-NLS is validated experimentally and compared with mechanisms in the literature. It is observed that the proposed CF-NLS is providing more displacement capacity without stress concentrations and with less expensive materials due to improved exploitation of the design space. This proposed CF-NLS is beneficial for dynamic activities like walking, running and jumping robots or biomechanics where high energy storage capacity is required.
