The prevention of torsional vibration-induced fatigue damage to turbo-generators requires determining natural frequencies by either
field testing or mathematical modelling. Torsional excitation methods, measurement techniques and mathematical modelling are active
fields of research. However, these aspects are mostly considered in isolation and often without experimental verification. The objective of
this work is to compare one dimensional (1D), full three dimensional (3D) and 3D cyclic symmetric (3DCS) finite element (FE) methodologies
for torsional vibration response. Results are compared to experimental results for a small-scale test rotor.
It is concluded that 3D approaches are feasible given the current computing technology and require less simplification with potentially
increased accuracy. Accuracy of 1D models is reduced due to simplifications but faster solution times are obtained. For high levels of
accuracy model updating using field test results is recommended.