Structural health monitoring of a bicycle stem: fatigue life evaluation from strain measurements and finite element analysis

The acquisition of mechanical strain signals from operational components is widespread across various applications. Their applicability for durability assessment is inherently dependent on the evaluation chain that links them to local material damage. This work demonstrates such a chain for a bicycle stem. Four full-bridge strain gauge configurations, calibrated including cross-sensitivities, synchronously capture bending and torsion during real riding. A linear-elastic finite element model, validated against analytical and experimental strains with deviations below 3.2 %, serves as the transfer function of the measurement chain. The acquired signals are converted into dimensionless load factors and projected as multiaxial stress histories across the entire component surface, encompassing failure-critical notches inaccessible to strain gauges. Fatigue life is evaluated with a Findley critical-plane criterion, rainflow counting, and Palmgren–Miner accumulation. A sensitivity analysis quantifies the accuracy requirements: due to the S–N exponent of 9.55, a ±10 % scaling error in the dominant bending channel changes the predicted life by −60 % and +175 %, while torsion channel precision is negligible for the investigated urban riding spectrum. The choice of the S–N data is even more decisive: literature-based material curves shift the predicted fatigue life by almost three orders of magnitude. A stress range threshold up to 5 MPa leaves the prediction unchanged and allows substantial data reduction in the embedded measurement system.