The long-term goal of this study is to improve the ability of designers and prosthetists to match the mechanical characteristics of prosthetic feet to patient specific parameters, including, needs, abilities and biomechanical characteristics. While patient measures of performance are well developed, there is a need to develop a practical method by which non-linear and time-dependent mechanical properties of the prosthetic component can be measured. In this study, testing methodologies were developed that separately evaluated the elastic and time-dependent properties. Three styles of feet were tested to span the range of designs of interest:​ a standard solid ankle cushioned heel (SACH) foot, two energy return feet for active users and a new prosthetic foot designed to provide partial energy return.

The first testing regime involved mechanically characterizing prostheses under conditions similar to gait. The heels and toes of four sample feet were loaded to peak forces based on their design mass at a series of angles and forces that the prosthetic system would go through during the gait cycle, based on the waveform in ISO 22675. Tangential stiffnesses of the samples were determined using numerical differentiation. The force-displacement responses of prosthetic feet reflect increasing stiffnesses with increasing loads and a decreasing pylon angle. Key features reflecting foot design are:​ the relative stiffness of the heel and toe and the displacement gap at midstance. Stable feet tend to exhibit lower heel stiffnesses and higher toe stiffnesses, whereas dynamics energy return (DER) feet tend to exhibit higher heel stiffnesses and lower toe stiffnesses. The differences in heel and toe loading at midstance suggest that DER feet can aid in the transition from heel to toe, providing a smooth rollover whereas SACH feet provide greater stability.

A second testing regime examined the time-dependent properties of the heel and toe. A three-parameter reduced relaxation response of the form L(t) = A + (1− A) exp(−tτ ) − Bt was able to capture the force-relaxation characteristics with RMS differences ranging from 0.0006 to 0.0119. In this model, A is the initial decay, B is the decay coefficient, a linear decay term, and τ is a time constant. While the model is practical for comparing various prostheses at a single load level, a fully non-linear model is required to model the time-dependent response at all loading level