TURBINE-GENERATOR SHAFT TORSIONALS
9-15
handling and manufacturing process, lack of homogeneity in the shaft material, porosity,
impurities, forging tears etc. It is evident that the predicted value of the number of cycles to
failure can vary considerably depending on whether the upper or the lower curve is used. The
uncertainty is more pronounced in the high cycle fatigue domain. There is considerable
difference in opinion in the definition of shaft failure. Failure can be defined as an outright
failure, or as reaching the point where cracks appear, necessitating inspection, resurfacing or
replacement. The definition of failure accepted by the industry seems to be crack initiation.
100% fatigue implies that crack initiation has been reached.
Any transient event that produces shaft torques exceeding the high cycle fatigue strength will
result in some loss of life. The cumulative effect of the number of cycles exceeding the fatigue
strength will determine the fatigue damage for the whole event. The fatigue damage for each
cycle can be determined using a fatigue life expenditure curve such as the one shown in Figure
9.4. Corresponding to the peak value of the torque during the torque cycle, the number of cycles
to failure is read from the curve. The inverse of this number multiplied by one hundred is the
percent loss of fatigue life for that torque cycle. The cumulative life expenditure for the
completed incident is obtained from the sum total of the life expenditures for the individual
cycles. The procedure is to be repeated for each shaft section.
A representative assortment of disturbances based on historic data could be considered for the
assessment of the shaft torsional stresses and the accompanying loss of fatigue life during the
operating life of a turbine-generator. Due to the many uncertainties involved and the lack of a
thorough understanding of the subject of material fatigue, the life expenditure curves supplied by
the manufacturers can be expected to be conservative. Therefore, if the estimated loss of life is
well below 100%, then, barring extraordinary events, shaft failure can be considered unlikely. In
a marginal situation or where shaft failure is indicated, the case should be studied in more detail
with the manufacturer.
Effect of damping
The amount of damping present in the torsional oscillations determine the number of fatigue
cycles experienced by each turbine-generator shaft section following a transient event. It has
been recognized that the decay rates of torsional oscillations at natural frequencies are very
small. Due to the complexities of the damping mechanisms, it is difficult to reliably predict the
damping values from design data. Different damping values have been observed in tests
conducted on nominally identical units under similar operating conditions. Tests conducted on a
number of turbine-generators have shown the decay time constants of the various oscillation
modes to be in the range from 5 to 15 seconds. Tests have also shown the damping values to be
functions of generator power output and network configuration -- in general, damping tends to
go down as the power output is reduced.
Since the oscillation frequencies are high and the damping is low, the initial peaks of the
torsional oscillations are not affected by damping. However, the amount of damping present
determines the number of cycles the shafts experience before the amplitudes fall below the high
cycle fatigue levels. Estimates of shaft fatigue life expenditure will therefore be critically
dependent on the damping assumptions. Since the damping values cannot be precisely
determined at the design stage, manufacturers recommend that station tests be performed to
accurately measure the damping values, if torsional response evaluations show significant fatigue
duty for disturbances that occur relatively frequently on the system.