0% found this document useful (0 votes)
47 views10 pages

Rectifier Diode Fault Detection Using FFT Analysis On Exciter Current

Power generation and electromechanical energy conversion

Uploaded by

Gabriel Ouverney
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
47 views10 pages

Rectifier Diode Fault Detection Using FFT Analysis On Exciter Current

Power generation and electromechanical energy conversion

Uploaded by

Gabriel Ouverney
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

2024 Paris Session

10123
A1 Rotating Electrical Machines
PS2 Evolution and development

Rectifier diode fault detection using FFT analysis on exciter current

Marc FLORES
EDF Hydro / DTG
France
[email protected]

Luc TEMPLIER
EDF Hydro / DTG
France
[email protected]

SUMMARY

Brushless excitation systems use rotating diode rectifiers to supply power to the main rotor of a
generator. A.C. exciters with rectifiers, are mainly made of a field-controlled AC excitation generator
which armature coils output currents supply the main generator’s rotor through a non-controlled diode
rectifier. The pros of such technology are a limited maintenance and smaller power parts of the excitation
system than with static exciters (mainly a thyristor bridge and three excitation transformers). The cons
are that it doesn’t allow an easy access to the rotor and then means missing measurements of the rotor
current and voltage. The difficult access to the rotating parts of excitation also means that it is difficult
to detect a rotating diode failure.

While on some large generators (output power> 300 MW), excitation systems could be equipped with
complex and costly detection systems for rotating diode failures (for instance using Hall effect
stroboscopic monitoring of diodes), EDF’s Hydro Power Plant generators fleet use smaller and cheaper
devices for this monitoring, essentially based on a cost-effective measurement of the exciter’s inductor
current.

The most implemented solution is based on an analog relay that monitors the ratio between the waveform
of the current signal and its mean value. Other ones use bandpass filters and monitors the amplitude of
the resulting signal.

The aim of this article is to present an alternative way of monitoring a diode fault using a Fast Fourier
Transform (FFT) of the raw current signal. In most cases the current is measured with a measuring shunt.
The shunt’s nominal voltage is often 100 mV. The FFT-based method presented here identifies the
presence of a component at the frequency of the exciter synchronous machine. This frequency is specific
for the rotating exciter synchronous machine and its typical value is between 10 Hz and 300 Hz.

Furthermore, simulations, real on-site test recording for a large number of generators (for the open
diode) and some real on-site failure recordings (for the short circuited diode) give us a large enough

1
feedback in order to define thresholds that could be used to find out if the fault is an open diode of a
short-circuited one. These thresholds can be defined with standard p.u. values which can be tuned with
on-site tests or simulations. For the moment, we assume that a ratio of at least 2 % means an open diode
and more than 10 % should be a sort-circuit. These ratios are given for reference purposes. What matters
in this article is the relative order of magnitude between the ratios obtained with and without faults.
There are different ways of calculating the FFT, for instance removing the DC component or not in the
signal etc… The calculation method will thus have an impact on the obtained numerical indicator. It is
also important to emphasize that the ratio is independent of the current amplitude and therefore of the
operating point of the generator. This observation applies to all types of faults and no-fault operation.
The main goal is to have a reliable protection device with separate setups for the different thresholds.

The next step consists in making a real-time embedded device (based on a digital signal processor),
implementing the FFT monitoring of such failures, installed in parallel of the actual analog protection
device and generating only alarms. The goal of this demonstrator is to show the feasibility and reliability
of the device.

KEYWORDS

Excitation systems, rotating diode rectifier, brushless excitation, diode rectifier fault protection, AVR,
measured waveform analysis.

2
1. QUICK REMINDER ON BRUSHLESS EXCITERS WITH ROTATING RECTIFIERS

One of the types of excitation systems is the brushless exciter. It is mainly made of a field-controlled
AC excitation generator which armature coils output currents supply the main generator’s rotor through
a non-controlled diode rectifier.

The figure below shows a typical brushless excitation system as it can be found on almost 30 % of EDF
hydro power plant (HPP) generators. [1] gives typical transfer functions for stability studies of such a
system.

Figure 1 : Excitation system based on brushless excitation

The pros of such technology are a limited maintenance and smaller power parts of the excitation system
than with static exciters (mainly a thyristor bridge and three excitation transformers). The cons are that
it doesn’t allow an easy access to the rotor and then means missing measurements of the rotor current
and voltage. The difficult access to the rotating parts of excitation also means that it is difficult to detect
a rotating diode failure.

2. ROTATING RECTIFIER DIODES FAULTS

Severe electric transients for the HPP are mainly due to grid short-circuits or grid-synchronisation
malfunctions that can create overvoltages on the generator rotor that can lead to a diode failure if the
voltage is higher than the diode breakdown voltage.

Overheating can also lead to a diode breakdown. This can occur for instance in case of an excess of dust
pollution or if the diode are used in an oil polluted environment.

In any case the failure of a diode can be of two kinds:


- a short-circuit,
- the opening of the diode.

3
NB : The opening of the diode is generally the consequence of the short circuit.

For both failures, the diode fault can be detected on the measurement of exciter’s inductor current (Iex)
in which a component at the frequency of the rotating exciter appears. This component is always present
in the Iex signal, even without diode fault. This can be explained by the armature reaction of the
brushless synchronous machine and to the diodes themselves switching each period with discontinuities
due to their voltage drop. When a fault occurs, the discontinuity in the output current of the exciter
increases the amplitude of this component.

3. EDF RECTIFIER DIODE FAULT PROTECTION DEVICE

While on some large generators (output power> 300 MW), excitation systems could be equipped with
complex and costly detection systems for rotating diode failures, EDF’s Hydro Power Plant generators
fleet use smaller and cheaper devices for this monitoring, essentially based on a cost-effective
measurement of the exciter’s inductor current.

The most implemented solution is based on an analog relay that monitors the ratio between the ripples
of the current signal and its mean value [2]. Other ones use bandpass filters and monitors the amplitude
of the resulting signal.

The aim of this article is to present an alternative way of monitoring a diode fault using a Fast Fourier
Transform (FFT) of the raw current signal. In most cases the current is measured with a measuring shunt
(as it can be seen in red on figure 1). The shunt’s nominal voltage is often 100 mV.

4. FFT FAULT SIGNATURE


4.1. REMARK ON THE SHUNT MEASUREMENT ENVIRONEMENT
The measuring shunt is generally in the power equipment cubicle of the AVR. That means
that, as shown on figures 2 and 4 above, the recorded signal is very noisy.

4.2. NO FAULT MEASUREMENTS

First of all, we can have a look on the example of a recorded excitation current with no rectifier fault.
The figures below show the signal. On the first one we can overlook the whole signal and notice that
this shunt measurement is very noisy. On the second one a specific zoom is shown at t=1 second.
The 100-mV voltage image the current is measured on the terminals of the shunt with a Dewetron data
acquisition system with a 5 kHz sampling frequency. The precision of the measurement is approximately
of 0,3 %.

4
Figure 2 : Excitation current measurement

Figure 3 : Excitation current measurement: zoom on the ripples without fault

Figure 4 : Excitation current measurement: zoom on the noise

The FFT of this signal on this raw measurement of the current (without filtering) is shown below. 3000
points sampled at 5 kHz are used.

5
Figure 5: FFT of the measured current between 0 and 1 kHz

The green mark at 0 Hz shows the amplitude of the DC component. In this example, the nominal
frequency of the exciter synchronous machine is 100 Hz. We can notice that there is nothing but noise
at this frequency (red circle).

A 600 Hz component can be seen, due to the operation of the three phases diode rectifier operating on
the 100 Hz currents of the exciter. We can also notice a small amount of 150 Hz and 300 Hz which are
harmonics of the 50 Hz frequency of the main generator.

6
4.3. OPEN DIODE FAULT MEASUREMENTS

The figure below shows the recording of the exciter’s inductor current (Iex) during no-load generator
tests with one arm of the rotating rectifier bridge open.

Figure 6: Excitation current measurement with an open diode fault


The FFT of the current is shown below.

Figure 7: FFT of the measured current with an open diode fault between 0 and 1 kHz

In this example, the nominal frequency of the exciter synchronous machine is 83 Hz. We can notice that
there is a component of the current at 83 Hz (and also harmonics of this frequency). The ratio between
the amplitude of the component at the exciter frequency and the DC component is of 2,9 %. This

7
information is provided for reference purposes. What matters in this article is the relative order of
magnitude between the ratios obtained with and without faults. There are different ways of calculating
the FFT, for instance removing the DC component or not in the signal etc… The calculation method
will thus have an impact on the obtained numerical indicator.

It is also important to emphasize that the ratio is independent of the current amplitude and therefore of
the operating point of the generator. This observation applies to all types of faults and no-fault operation.

4.4. DIODE SHORT-CIRCUIT FAULT MEASUREMENTS

“Chance” made that the exciter current behaviour during a real short-circuit diode fault has been
recorded. This was during the commissioning of a new AC exciter and a diode was in short circuit when
the excitation was switched on at the first time. Unfortunately, the manufacturer hadn’t yet received the
protection and decided to start his tests without it.

The figure below shows the recording of the exciter’s inductor current since the startup of excitation
with the AVR in manual mode (current control).

Figure 8: Excitation current measurement with short-circuit diode fault

Nb: the measurement saturates at 100 A (6.4 p.u.) because it wasn’t expected to have such values given
a nominal current of 15 A.

The FFT of the current is shown below.

8
Figure 9: FFT of the measured current with an short-circuit diode fault between 0 and 1 kHz

In this example, the nominal frequency of the exciter synchronous machine is 23 Hz. We can notice that
there is a huge component of the current at 23 Hz (and also harmonics of this frequency). The ratio
between the amplitude of the component at the exciter frequency and the DC component is of 42 %.

5. CONCLUSIONS

The FFT of exciter current shown in §4 below shows that monitoring the AC component at the nominal
frequency of the exciter synchronous machine makes it possible to detect a rectifier fault. This frequency
is specific for the rotating exciter synchronous machine and can be from 10 Hz to 300 Hz.

Furthermore, simulations, real on-site test recording for a large number of generators (for the open
diode) and some real on-site failure recordings (for the short circuited diode) give us a large enough
feedback in order to define thresholds that could be used to find out if the fault is an open diode of a
short-circuited one. These thresholds can be defined with standard p.u. values which can be tuned with
on-site tests or simulations. For the moment, we assume that a ratio of at least 2 % means an open diode
and more than 10 % should be a sort-circuit. These ratios are given for reference purposes. What matters
in this article is the relative order of magnitude between the ratios obtained with and without faults.
There are different ways of calculating the FFT, for instance removing the DC component or not in the
signal etc… The calculation method will thus have an impact on the obtained numerical indicator. It is
also important to emphasize that the ratio is independent of the current amplitude and therefore of the
operating point of the generator. This observation applies to all types of faults and no-fault operation.

The next step consists in making a real-time embedded device (based on a digital signal processor),
implementing the FFT monitoring of such failures, installed in parallel of the actual analog protection
device and generating only alarms. The goal of this demonstrator is to show the feasibility and reliability
of the device.

BIBLIOGRAPHY

[1] “Rotating Electrical Machines – Part 16: Excitation systems for synchronous machines, chapter
2: models for power system studies”, IEC 60034-16-2, 1991
[2] M. G. McArdle, D. J. Morrow, “Noninvasive detection of brushless exciter rotating diode failure”,
IEEE, 2004

9
[3] H. Gras, “Modélisation dans le domaine du temps d’une excitatrice synchrone à diodes tournantes
et analyses des modes de défaillance”, école polytechnique de Montréal, 2014
[4] J. Sjölander, “Non-Invasive Technologies for Condition Monitoring of Synchronous Motors”,
Uppsala University, 2013

10

You might also like