Patent Application
20170254696
The invention relates to a method for operating a machine plant having a shaft train.
A machine plant, such as for example a power plant for generating electricity, has an electrical generator and at least one turbine, the turbine driving the generator. For this purpose, the turbine and the generator are arranged on a common shaft train. It is possible to provide couplings in the shaft train in order to couple individual turbines into the shaft train or decouple them from the shaft train.
During the operation of the power plant, the generator feeds electricity into a power grid. Electrical disturbances in the power grid may have the effect that a torsional oscillation of the shaft train is induced by the generator. In the worst case, the shaft train may be induced to oscillate at one of its natural frequencies. Being induced to oscillate at a natural frequency has the effect of high torsional stress loading of the shaft train, which may lead to a reduction in its service life or to the formation of stress cracks. It is expected that the disturbances in the power grid will occur more frequently in the future, because the structures of the power grid are changing in the course of the transition to new forms of energy.
Conventionally, the loading of the shaft train is measured, by its rotational speed being measured with high resolution at a number of axial positions of the shaft train. The loading of the shaft train is then concluded from the differences in the rotational speeds at the different axial positions. However, in this method the measuring technique and the associated evaluation are complex. It is alternatively possible to measure oscillations in the electrical energy delivery of the power plant.
In the event that potential oscillations are identified in the energy delivery, no conclusions as to the actual loading of the shaft train can be drawn however.
DE 40 32 299 A1 describes a method for monitoring a rotatable component in which the oscillation excursion of the component in the radial direction in relation to the component and the rotational position of the component are measured time-dependently. EP 0 585 623 A2 describes a method for the early detection of a crack in a rotating shaft. DE 27 34 396 A1 describes a method for monitoring torsional oscillations on shafts in which the instantaneous power of the generator is measured.
An object of the invention is to provide a method for operating a machine plant having a shaft train in which loads on the shaft train caused by torsional oscillations can be easily detected.
The method according to the invention for operating a machine plant having a shaft train has the steps of: a) computationally determining the natural frequency of at least one torsional oscillation mode of the shaft train and computationally determining the mechanical stresses of the shaft train occurring during an oscillation period of the torsional oscillation mode; b) determining in each case a correlation for each torsional oscillation mode between a first stress amplitude, which occurs at a position of the shaft train that is susceptible to the formation of torsional stress damage, and a second stress amplitude, which occurs at a measuring location of the shaft train, on the basis of the stresses computationally determined for the respective torsional oscillation mode; c) establishing a maximum first stress amplitude for the position; d) determining a maximum second stress amplitude, corresponding to the maximum first stress amplitude, for the measuring location on the basis of the correlation; e) measuring the stress of the shaft train while the shaft train is rotating at the measuring location in a time-dependent manner; f) determining a stress amplitude at each natural frequency from the measured stress; g) emitting a signal in the event that, at one of the natural frequencies, the stress amplitude determined from the measured stresses reaches the maximum second stress amplitude.
By determining the correlations between the stress amplitudes occurring at the position and the measuring location from the computationally determined stresses, stress amplitudes can be advantageously determined at measuring locations of the shaft train that are not accessible for measurement. A single measuring location on the shaft train is sufficient for determining the stress amplitudes at the position, while any location can be chosen as the measuring location. As a result, the method can be carried out advantageously easily. By determining a correlation for each torsion mode, the stress amplitudes at the position can be advantageously determined with a high degree of accuracy.
In response to the signal, measures can be taken to reduce the torsional oscillations of the shaft train. If the machine plant is a power plant for feeding electricity into a power grid, a desynchronization between the oscillation-inducing factors from the power grid and the torsional oscillations of the shaft train may for example be performed. This may take place for example by changing the natural frequencies of the shaft train by decoupling a turbine or a plurality of turbines from the shaft train. It is also possible to decouple the power plant from the power grid by disconnecting electrical lines.
Advantageously, in step e) the stress is measured by means of a magnetostrictive sensor and/or by means of a strain gage, the stresses measured by means of the strain gage being transmitted by means of telemetry. The magnetostrictive sensor generates a high-frequency magnetic field by means of a transmitting coil and floods the surface of the shaft train with the magnetic field. The magnetostrictive sensor has one or more receiving coils, which with the transmitting coil form a magnetic circuit and are arranged specifically for the detection of torsional oscillations. With the specific arrangement of the transmitting coil and receiving coils, the magnetic susceptibility at the surface of the shaft train can be determined. Because the magnetic susceptibility of a body depends on its mechanical stress as a result of the Villari effect, the signal detected by the receiving coils also depends on the mechanical stress of the surface of the shaft train. The magnetostrictive sensor emits an analog output signal that is proportional to the torsional stress of the shaft train. As a result of the proportionality of the output signal in relation to the torsional stress, the data processing is advantageously of computationally low intensity, as a result of which the method can be carried out easily and quickly. Both the magnetostrictive sensor and the strain gage determine the stress of the shaft train directly, whereby the method can likewise be carried out easily.
It is advantageous that in step f) the stresses measured in step e) are filtered by means of a bandpass filter, which allows a frequency band arranged around the natural frequency to pass, and/or that a Fourier transformation, in particular a fast Fourier transformation, of the stresses measured in step e) is carried out. This allows the stress amplitudes of the different torsional oscillation modes to be determined advantageously easily. In the event that the bandpass filter is used, in step g) the signal can be emitted just a short time after reaching the maximum second stress amplitude. In the event that the Fourier transformation or the fast Fourier transformation is used, the transmission of the signal takes place after completion of the calculation.
It is advantageous that in step e) the stress is measured at a plurality of measuring locations arranged at different axial positions of the shaft train. Here it is possible to achieve a high degree of measuring accuracy of the natural frequencies by arithmetic averaging of the natural frequencies determined in the plurality of measuring locations. Moreover, it may happen that one of the measuring locations coincides with an oscillation node of one of the torsional oscillation modes, as a result of which the measuring location would not be able to detect the torsional oscillation mode. By providing the plurality of measuring locations, the probability of all the oscillation nodes for a torsional oscillation mode lying at the measuring locations is very low. Consequently, all the relevant torsional oscillation modes can be detected with a high degree of probability.
Advantageously only the measuring location that has the highest mechanical stress, computationally determined in step a), for a selected torsional oscillation mode is used in step f) for the selected torsional oscillation mode. As a result, a high signal-to-noise ratio is advantageously achieved in the determination of the stress amplitudes. It is advantageous that in step f) there is an assignment of the stress amplitudes to one of the natural frequencies, in which an experimental ratio of the stress amplitudes and/or oscillation phases, determined in step f), at the plurality of measuring locations is formed, a computational ratio of the computationally determined stress amplitudes and/or oscillation phases at the plurality of measuring locations is formed and the experimental ratio is compared with the computational ratio. The comparison of the experimental ratio with the computational ratio allows the torsional oscillation modes in which the shaft train vibrates to be advantageously determined correctly with a high degree of probability.
It is advantageous that the natural frequencies computationally determined in step a) are either lower or equal to two and a half times the rotational frequency of the shaft train or, in the event that the machine plant is a power plant for generating electricity and the grid frequency is higher than the rotational frequency, lower or equal to two and a half times the grid frequency. These natural frequencies result in particularly strong stress loads on the shaft train, and so monitoring of these frequencies is particularly advantageous. In step a) the centrifugal loading that the shaft train has during the operation of the machine plant is advantageously taken into account in the computational determination of the natural frequency and the mechanical stresses. This allows the first stress amplitudes to be determined with a high degree of accuracy. The correlation between the first stress amplitude and the second stress amplitude that is determined in step b) is advantageously a proportionality.
In step c) an additional third maximum stress amplitude, which is higher than the first maximum stress amplitude, is advantageously established for the position, in step d) a maximum fourth stress amplitude, corresponding to the maximum third stress amplitude, is determined for the measuring location on the basis of the correlation and in step g) a stop signal is emitted in the event that the stress amplitude at the natural frequency that is determined in step f) reaches the maximum fourth stress amplitude, the stop signal initiating a shutdown of the machine plant. In this way, excessive stress loading of the shaft train can be advantageously automatically prevented.
The method advantageously has the step: d1) coarse calibration of the magnetostrictive sensor and/or the strain gage on a shaft that is loaded with a defined stress by means of a testing machine. It is advantageous that the machine plant has a generator and that the method is carried out with the step: d2) fine calibration of the magnetostrictive sensor and/or the strain gage during the rotation of the shaft train by using a stress that is determined at the measuring location by using the power delivered by the generator. The torque that the generator exerts on the shaft train can be concluded from the power delivered by the generator. The stresses occurring in the shaft train can then be concluded from the torque. This advantageously allows the calibration to be performed on the actual shaft train and during the operation of the machine plant.
It is advantageous that in step b) correlations are determined for a plurality of positions of the shaft train that are susceptible to the formation of stress damage, in step c) a respective maximum first stress amplitude is established for each position, in step d) a corresponding maximum second stress amplitude is determined for each of the positions and in step g) the signal is emitted in the event that one of the maximum second stress amplitudes is reached. This advantageously allows a plurality of critical locations of the shaft train to be monitored with just a single measuring location. The maximum first stress amplitudes at the various positions may in this case be the same or different from one another. It is also possible to establish third maximum stress amplitudes for the plurality of positions, it also being possible for these to be the same or different from one another.
It is advantageous that the measuring locations on the shaft train are arranged in the region of an exposed portion of the shaft train, in the region of a portion of the shaft train that is accessible for the measuring technique and/or in the region of a computationally determined stress maximum. The positions are advantageously arranged in the region of bearings, couplings and/or blade roots. These are advantageously positions that are particularly susceptible to stress damage. Advantageously, in step a) the natural frequencies and the stresses are computationally determined by means of a finite element method.
The method according to the invention is explained in more detail below on the basis of the accompanying schematic drawing, in which:
As can be seen from
Likewise represented in
In the method, at least one position of the shaft train 2 that is susceptible to the formation of stress damage is identified. The stress damage may take the form of stress cracks and/or forced ruptures. These positions may for example lie in the region of bearings, couplings and/or blade roots. For each of the positions, a maximum first stress amplitude is established, chosen such that it is not reached during normal operation of the machine plant in which no resonant disturbances occur. It is also chosen such that, when it is reached, still no stress damage occurs at the position. The maximum first stress amplitude is in this case the same for all the torsional oscillation modes 14 to 17. It is also possible to establish for the position a maximum third stress amplitude that is higher than the maximum first stress amplitude. The third maximum stress amplitude is in this case chosen such that, when it is reached, still no stress damage occurs on the shaft train.
On the basis of the computationally determined stresses, a correlation between the stress amplitude at the position and the stress amplitude at at least one of the measuring locations 7 to 11 is determined for each torsional oscillation mode, the correlation being a proportionality. On the basis of the correlation, a second maximum stress amplitude can be determined for the measuring location 7 to 11, the second maximum stress amplitude occurring at the measuring location 7 to 11 when the first maximum stress amplitude occurs at the position. On the basis of the correlation, a fourth maximum stress amplitude can also be determined for the measuring location 7 to 11, the fourth maximum stress amplitude occurring at the measuring location 7 to 11 when the third maximum stress amplitude occurs at the position.
The mechanical stress of the radially outer-lying surface of the shaft train 2 is measured at each measuring location 7 to 11 with in each case at least one magnetostrictive sensor 18. For reasons of redundancy, a plurality of magnetostrictive sensors may be provided at each measuring location 7 to 11. The magnetostrictive sensor 18 has a transmitting coil, by means of which a magnetic field is generated. The surface of the shaft train 2 is flooded with the magnetic field. The magnetostrictive sensor 18 has a receiving coil or a number of receiving coils, which are arranged in a magnetic circuit together with the transmitting coil and the flooded surface of the shaft train 2 in such a suitable way that the magnetic susceptibility of the surface of the shaft train 2 can thereby be determined. As a result of the Villari effect, the magnetic susceptibility of the shaft train 2 depends on its mechanical stress. The signal emitted by the magnetostrictive sensor is in this case proportional to the torsional stress of the magnetic field. As can be seen in
In the method, the mechanical stresses are measured in a time-resolved manner by means of the magnetostrictive sensor 18. The measured stresses can then be filtered by means of a bandpass filter, which allows a frequency band arranged around the natural frequency of the torsional oscillation mode to pass. It is alternatively possible to evaluate the measured stresses by means of a Fourier transformation, in particular a fast Fourier transformation. As soon as this filtered stress or the stress amplitude obtained by means of the Fourier transformation reaches or exceeds the second maximum stress amplitude, a signal is emitted. In the event that the natural frequencies lie close together, it may be necessary to provide a bandpass filter that allows natural frequencies for a plurality of torsional oscillation modes to pass. In this case, the signal is emitted when the filtered stress reaches or exceeds the lowest of the second maximum stress amplitudes of the plurality of torsional oscillation modes. In the event that a maximum fourth stress amplitude is provided, a shutdown of the machine plant 1 may be provided in the event that the filtered stress reaches or exceeds the maximum fourth stress amplitude.
It is possible to use only a single measuring location 7 to 11 for a specific torsional oscillation mode 14 to 17. In this case, the measuring location 7 to 11 that has the highest computationally determined stress for the specific torsional oscillation mode 14 to 17 may be used in order to achieve a high signal-to-noise ratio. For this purpose it is possible to identify in
It is likewise possible to use the stresses measured at a plurality of measuring locations 7 to 11 in order to identify the torsional oscillation mode 14 to 17 in which the shaft train 2 oscillates. For this purpose, a computational ratio of the form Ar1:Ar2: . . . :Arn is formed for each torsional oscillation mode 14 to 17, where Arx is the computationally determined stress amplitude at the xth measuring location and n is the number of measuring locations. An experimental ratio of the form Ae1:Ae2: . . . :Aen is formed from the stress amplitudes determined from the measurements, where Aex is the stress amplitude determined from the measurements at the xth measuring location. By comparing the computational ratio with the experimental ratio, the torsional oscillation mode in which the shaft train 2 oscillates can be uniquely assigned even when the natural frequencies of the various torsional oscillation modes lie close together.
In step c the stresses of the shaft train 2 are measured and in step d the measured stresses are displayed. In step e the measured stresses are filtered by means of the bandpass filters in order to determine the stress amplitudes at the respective natural frequency. In step f the filtered stress amplitudes are displayed.
In step g it is enquired whether the filtered stress amplitude exceeds the limit value σSTORE. If this is the case, in step h the measured stresses are stored. It is alternatively possible to store the measured stresses all the time at great time intervals and in step h to store the measured stresses at small time intervals.
In step i it is enquired whether the filtered stress amplitude exceeds the limit value σW. If this is the case, in step j a warning signal is emitted. In step k it is enquired whether the filtered stress amplitude exceeds the limit value σA. If this is the case, in step l an alarm signal is emitted. The alarm signal may have the effect that an operator of the machine plant shuts it down manually. Alternatively, the alarm signal may have the effect that an automatic shutdown of the machine plant takes place.
Although the invention has been described and illustrated in more detail by way of the preferred exemplary embodiment, the invention is not restricted by the disclosed examples and other variations can be derived herefrom by a person skilled in the art without departing from the scope of protection of the invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 14183042.2 | Sep 2014 | EP | regional |
This application is the US National Stage of International Application No. PCT/EP2015/069642 filed Aug. 27, 2015, and claims the benefit thereof. The International Application claims the benefit of European Application No. EP14183042 filed Sep. 1, 2014. All of the applications are incorporated by reference herein in their entirety.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2015/069642 | 8/27/2015 | WO | 00 |
