Courtesy by SIEMENS
Courtesy by DOOSAN



A variety of different measurement techniques have been successfully employed in the past to measure the torsional vibration characteristics of coupled shaft systems. These methods are subject to continuous improvement; therefore, which is most appropriate for a specific application will depend on several factors. This section provides further background to some of these measurement techniques. It is emphasized that these are not the only available methods and there are some other equally applicable methods developed by the different OEMs.


The following devices may be used as torsional vibration transducers:

a) eddy current probes, inductive probes, lasers, etc. (non-contacting transducers);

b) strain gauges;

c) optical decoders;

d) accelerometers positioned circumferentially at specified angles (preferably at 0° and 180°).

Other methods may be used by agreement between the customer and set supplier.


Depending upon the method of measurement, it is recommended that the following be included in the test report:

a) rotational velocity of the shaft system;

b) turbine generator set power output;

c) torsional vibration magnitude;

d) torsional strain;

e) ambient temperature of test site;

f) torsional natural frequency;

g) speed range over which measurements are carried out.

An additional parameter that can influence the torsional vibration is

h) blade-disc coupled nodal diameter frequency.

Other parameters can be measured on agreement between the customer and set supplier.

Courtesy by GE
Courtesy by ANSALDO


“Modal tests” may be performed on rotors in their static (i.e. non-rotating) configuration in the factory. This is one way of measuring mode shapes and natural frequencies under static conditions.
They provide verification of predicted behaviour and, therefore, help to calibrate shaft system models for fundamental rotor body and overhang modes under stationary conditions. In recognition of the fact that boundary conditions influence the final outcome of the test, it is important either
to carry out a free-free test or to support test rotors on bearing journals with hard rubber or similar supports. These supports provide little resistance to the impact energy path at the contact areas so that the relevant torsional modes of the rotor under test are properly captured in the
frequency spectrum.

When blades are mounted on rotors, a perfect contact between blade roots and the main rotor body may not always be achievable when the rotor is stationary, due to the design of root employed or manufacturing tolerances. As a result, the impact energy imparted by a test hammer to the structure may be disrupted at the blade-rotor body contact areas, making it difficult to capture blade-disc and rotor coupled frequencies and their associated modes. Even if such modes are captured in the test, they are less useful because the blade-disc frequencies will continuously change with speed. 
In other words, the blades dynamically couple with the rotor and this will continue until rated speed is reached. This coupling effect, along with stress stiffening due to speed, creates new sets of torsional frequencies that are different from individual blade and disc alone frequencies under
stationary conditions. Similar difficulties exist when carrying out static tests on generator rotors due to the influence of the copper windings and associated wedges.

Therefore, although static “modal tests” are helpful to calibrate rotor body models, they do not generally provide an accurate assessment of the frequencies of either the blade-disc coupled system or the generator body modes that vary with speed. Full speed (dynamic) factory or site tests are necessary to assess these effects.
Figure below shows the arrangement for a typical factory static rotor test.


Depending on the type of mode that is critical, a field test could involve measurements at a few locations or be performed on a more elaborate scale. The choice of measurement positions is normally determined by examination of the predicted mode shapes, but in most cases the measuring of torsional vibration magnitudes at two locations, such as the turbine to generator shaft region and the permanent magnet generator, is sufficient to capture the important turbine and generator coupled rotor and blade modes. However, if more detailed mode shapes are required, it can be necessary to measure torsional response at other locations on the shaft or at the blade tips.

In order to measure torsional natural frequencies at a power plant site, it is necessary to provide a means of detecting torsional vibration at one or more locations along the shaft. This can be done by various means: toothed wheels and magnetic or proximity transducers, painted or taped stripes and optical transducers, permanent magnet generator (PMG) voltage signals, strain gauges, etc. The demodulated torsional signals are then usually displayed on a spectrum analyzer to determine their various frequency components. Although it requires more effort, utilizing multiple measuring locations is an advantage because they can potentially identify more torsional modes and because their relative magnitudes at each natural frequency enable the measurement of mode shapes.

When a turbine generator is assembled at a power plant site, it becomes a challenge to measure torsional natural frequencies, particularly those in the vicinity of twice line frequency. 
Sub-synchronous (below-line frequency) torsional natural frequencies can typically be easily measured with the unit on-line, as they are usually excited to measurable levels by random power system fluctuations. If necessary, synchronization or line switching tests can be done to excite
these sub-synchronous torsional modes. Such techniques, however, may not be successful in exciting and measuring super-synchronous (above-line frequency) torsional natural frequencies with much confidence. This is why in the mid-1970s the off-line torsional frequency response test
was developed.

An off-line torsional frequency response test involves exciting the generator in a controlled manner using oscillatory torque developed from unbalanced currents flowing in the generator stator.
These unbalanced currents are obtained by the application of a line-to-neutral short circuit test connection on the high-voltage side of the generator main step-up transformer (or, alternatively, a line-to-line connection at the generator terminals) while the unit is shut down and not connected to the grid. With the turbine generator running at various speeds, a small amount of field excitation is applied to the generator, which induces unbalanced or negative sequence current in the generator. The field excitation is applied in a controlled amount small enough not to exceed the negative sequence current heating limits of the generator but large enough such that the induced negative sequence currents can excite shaft system natural frequencies to measurable levels at resonance.

The generator air gap torque induced by these negative sequence currents occurs at a frequency equal to twice the electrical frequency of the generator. By changing the speed of the generator during this test (possible because the generator is off-line), the electrical frequency and thus
the frequency of the air gap torque also changes. Thus, by slowly ramping the speed, a slow sweep of the air gap torque frequency is obtained. In this manner, torsional natural frequencies can be detected by the occurrence of resonant peaks in the torsional response as the speed is ramped.
Later in the test, the speed can be held at each of these resonant peaks in order to measure precisely the torsional natural frequency, and also to confirm that the measured response is indeed due to torsion, by removing the field excitation (and thus the air gap torque) and observing that the response magnitude changes accordingly.

For several years, the off-line frequency test was the only reliable method of accurately identifying the torsional natural frequencies of the fully installed unit at site. However, with the increasing sophistication of signal analysis techniques, it is now possible to detect all frequencies of interest by measuring the effects of small transient disturbances that occur under normal operation. These disturbances, which are a consequence of the minor random disturbances that are inevitably present on all electrical networks, cause transient excitation of those natural modes that can be excited by the generator. The advantage of this technique for the customer is that, other than the time required to install the measurement equipment, there is no impact on the normal operation of the power plant.