The tendon-driven architecture of a prosthetic hand is, essentially, a synthetic copy of the human exor system: inextensible tendons routed through annular pulleys transmit the force of the exor muscles to the phalanges. The anthropomorphic load path comes with a cost, though. The synthetic analogue inherits the two failure modes that affect its biological counterpart under chronic loading: cyclic fatigue of the tendon element at each pulley bend, and abrasive wear of the pulley surface. Both mechanisms are examined here for a cable-driven hand prosthesis with Nylon 6/10 phalanges whose cables are tensioned by DC gear motors to reproduce six degrees of freedom of the human hand. Applying the modified Goodman criterion together with Archard's law at the cable-channel contact exposes a degradation loop that static finite-element analysis cannot detect: wear reshapes the geometry that governs fatigue, and fatigue damage accelerates as the wear groove develops. At nominal 30 N loading, the fatigue safety factor is adequate (nf = 3. 45) but the estimated life is limited to 2. 1×106 cycles - about 210 days of intensive daily use - and cumulative channel wear of 0. 53 mm raises the local stress concentration over that same interval. Finite-element simulation points to the medial phalanx of the middle finger as the critical fatigue location, with a peak von Mises stress of 72. 5 MPa (52% of the elastic limit). A 0. 5 mm, 100 mm long Nitinol wire, acting as an artificial muscle in parallel with the existing structure, is shown to deliver the same 30 N grip force with a thermomechanical life beyond 10⁷ cycles - a fivefold improvement - while removing the abrasive cable-channel contact from the actuation phase altogether. A trade-off remains: the SMA heating time (3-8 s) is slower than DC motors (< 1 s), and passive convective cooling adds further delay. A hybrid scheme is proposed in which NiTi artificial muscles hold grip force while DC motors handle rapidpositioning. The present article is a predictive framework; the fatigue, wear, thermal and integration models formulated here motivate a subsequent experimental validation campaign on a dedicated finger test rig, which is outlined as future work.
Jair Leopoldo Loaiza Bernal (Thu,) studied this question.