Physics-based simulation and statistically optimized assessment of multi-crack and mass-variation effects in a Jeffcott rotor

Rotating machinery is vulnerable to catastrophic failure when shaft cracks and rotor mass variations remain undetected, yet systematic studies of their combined influence within a controlled benchmark rotor remain limited. This work develops a physics-based computational–experimental framework that integrates finite-element modelling in ANSYS, a Jeffcott rotor test rig and a Taguchi L9 design with ANOVA to quantify how disk mass, crack number and crack depth affect the first-mode vibration frequency. A nonlinear breathing-crack FE model with time-varying stiffness was formulated and then reduced to an equivalent static notch representation for efficient frequency prediction, and its natural-frequency predictions were validated against experiments on nine crack–mass configurations. Taguchi response tables and ANOVA identify disk mass as the statistically dominant factor governing frequency, while crack parameters primarily manifest through 2X harmonic content and orbit distortion in the experimental spectra. The optimized configuration yields a predicted first-mode frequency near 24.7 Hz with a mean absolute experimental error of about 0.5 Hz and regression fits with R 2 ≥ 0.9856, indicating highly reliable estimation over the investigated design space. Rather than proposing a new rotor model or crack signature, the study provides a validated benchmark-level framework for quantifying interacting fault effects in a Jeffcott rotor and supports future extension towards more realistic multi-disk industrial systems, condition-based maintenance schemes and structural health monitoring strategies.