TECHNICAL FIELD

The present invention relates to a method and auxiliary apparatus for balancing a rotor of a gas turbine .

BACKGROUND ART

As known, a gas turbine rotor in an industrial system for generating electricity generally comprises a plurality of bladed discs and one or more spacer elements, which are aligned along an axis and frontally coupled. The discs and spacer elements are sandwiched by a central tie rod. The discs are provided with respective arrays of blades and each defines a compressor or turbine rotor stage.

The rotors of the gas turbines must be made and assembled with high accuracy to ensure near perfect balancing. Given the masses and the high rotation speeds (normally 3000 rpm or 3600 rpm, according to the standards of the various countries), even minimum faults may cause dangerous vibrations and usually cause the early aging of some components.

However appropriately balanced a gas turbine rotor leaving the factory is considered to be, the weight of the rotor itself, the working temperatures and the loads generated during the rotation may cause mass symmetry alterations with respect to the axis over time. Indeed, in use, the weight force tends to deform the rotor, which is supported by bearings at the ends only.

It is often necessary to intervene on the rotor of a gas turbine in service to restore the appropriate balancing conditions which, although not ideal, are in all cases within tolerated limits.

In some cases, relatively low invasive interventions are possible, such as the addition of balancing weights in appropriately provided housings.

Other times, instead, the unbalance cannot be compensated only with these operations and it is necessary to disassemble the rotor and modify the relative orientation of the groups of discs, so that the alignment errors may tend to be cancelled.

The operations of disassembling and reassembling the rotor are extremely costly, particularly because they imply very long machine downtime, even of the order of several weeks. It is thus apparent that reducing the time required for this type of interventions as much as possible is of primary importance.

Unfortunately, the problem of identifying which discs are to be rotated and by what angles is extremely complex, and not uncommonly the solutions identified on paper are inadequate in practice because of various unknown factors which cannot be taken into account. For example, the contact surfaces between continuous discs may have deformations or faults caused by use, which make the effect of a relative rotation unpredictable. If the theoretical solution does not offer the expected results and the check reveals that rotor balancing either has not been achieved or is insufficient, the procedure must be repeated with great waste of time.

DISCLOSURE OF INVENTION

It is the object of the present invention to provide a method and auxiliary apparatus for balancing a rotor of a gas turbine which allow the described drawbacks to be avoid or at least attenuated.

According to the present invention, a method and auxiliary apparatus for balancing a rotor of a gas turbine are provided as defined in claims 1 and 10, respectively .

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described with reference to the accompanying drawings, which show some non-limitative embodiments thereof, in which:

- figure 1 is top plan view of a gas turbine rotor, taken in section along a horizontal axial plane, and shows a simplified block chart related to an auxiliary apparatus for balancing the rotor of a gas turbine according to an embodiment of the present invention;

- figure 2a is a diagrammatic representation of an eccentric portion of the rotor in figure 1 ;

figure 2b is a chart which shows first polar coordinate eccentricity measurements related to the rotor in figure 1 in an initial configuration;

- figure 3 is a simplified flow chart, related to a method for balancing the rotor of a gas turbine according to an embodiment of the present invention;

- figure 4 is a more detailed flow chart related to the steps of the method in figure 3;

figure 5 is a chart which shows eccentricity components extracted from the first measurements in figure 2;

- figure 6 is a chart which shows first quantities used in the method according to the present invention;

- figure 7 is a chart which shows second quantities used in the method according to the present invention;

- figure 8 is a more detailed flow chart related to further steps of the method in figure 3;

figure 9 is a chart which shows second polar coordinate eccentricity measurements related to the rotor in figure 1 in an initial configuration;

figure 10 is a chart which shows eccentricity components extracted from the second measurements in figure 9;

figure 11 is a chart which shows third polar coordinate eccentricity measurements related to the rotor in figure 1 in an actual modified configuration; and

figure 12 is a chart which shows eccentricity components extracted from the second measurements in figure 11.

BEST MODE FOR CARRYING OUT THE INVENTION

With reference to figure 1, a rotor of a gas turbine of a system for generating electricity is indicated as a whole by reference numeral 1 and comprises a plurality of discs 2 aligned along an axis A and sandwiched by a central tie rod 3. A first group of discs 2, provided with respective first rotor blades 5, defines a compression section la of rotor 1, while a second group of discs 2, provided with second rotor blades 6, defines a turbine section lb of rotor 1. The compression section la and the turbine section lb are separated from each other by a disc 2 free from blades acting as a spacer element in practice and substantially cylindrical in shape. In use, an annular combustion chamber (not shown) of the gas turbine may be arranged about the spacer disc 2.

A front bearing coupling portion 7 and a rear bearing coupling portion 8 are obtained in a first end disc and in a second end disc, respectively, also referred to as front or frontal hollow shaft and rear hollow shaft and indicated herein by reference numerals 2a, 2b.

For convenience, hereinafter "discs 2 " will indicate as a whole the front hollow shaft 2a, the rear hollow shaft 2b, the discs 2 of the compression section la, the discs 2 of the turbine section 2b and the spacer disc 2, unless otherwise specified.

Figure 1 also shows an auxiliary apparatus for balancing the rotor of a gas turbine, indicated by reference numeral 10 and comprising a processing station 11 and a measuring instrument 12.

The processing station 11 receives measuring data from the measuring instrument 12 and, according to the received data, determines a modified configuration of rotor 1, in which one or more discs 2 are rotated with respect to the initial configuration so as to reduce the total unbalance of rotor 1.

The measuring instrument 12 is made so that it can be displaced along the axis A of rotor 1 mounted on the floor of the lower half-casing of a turbine (not shown) or on a lathe (of which only a steady rest 13a, which supports the front bearing coupling portion 7, and a spindle 13b, which supports the rear bearing coupling portion 8, are illustrated in a simplified manner; the zero eccentricity centering with respect to the rotation axis is obtained by accurately positioning the clamp) .

The measuring instrument 12 is made so as to acquire information related to the phase φ of rotor 1 (e.g. by an encoder or a revolution and angular speed detector) and to the axial position PA at which the measurements are performed.

The measuring instrument 12 may thus determine data related to eccentricity of each disc 2 with respect to the actual rotation axis AR of rotor 1 supported by the bearings of the machine or by the steady rest 13a and by the spindle 13b of the lathe (also see figure 2a: the actual rotation axis, in the presence of alignment faults, does not perfectly coincide with the theoretical axis A of the undeformed rotor 1 and with the center C of the discs 2; furthermore, the actual rotation axis AR may also be determined by parallelism faults in the coupling to the steady rest 13a and spindle 13b, when the measurements are performed on a lathe, as in the described example) . The eccentricity is determined by rotating rotor 1 on the lathe (or, alternatively, on its bearings in the lower half-casing of the machine) and determining the oscillations of the surface of each disc perpendicularly to the actual rotation axis AR at a location. In practice, a periodical run-out signal is recorded. The magnitude and phase of the first harmonic of the run-out signal define the entity of the deviation of the center C of disc 2 under invetigation with respect to the actual rotation axis AR (also see figure 2a) and the direction of deviation with respect to a reference direction of rotor 1. Instead, the second harmonic of the run-out signal defines the ellipsoidality of the component under examination. Furthermore, in the case of a perfectly circular disc 2 and with actual rotation axis skewed with respect to the axis of gravity (which is typical for a gas turbine rotor) , the second harmonic contributes to estimating the bending of the corresponding portion of the axis A of rotor 1. In one embodiment, the run-out data (and thus eccentricity data) related to each disc 2 are determined by performing measurements indicated during a plurality of rotations of rotor 1 and by performing an averaging operation on the determined measurements. In one embodiment, for example, the processing station 11: determines the Fourier transform of the portion of the run-out signal related to each rotation of rotor 1 during which the measurement is performed for the disc 2 under investigation; checks that the determined eccentricity peak-peak value (on N revolutions) is within the repeatability range of the measuring instrument; determines a mean run-out signal from the mean of the transforms of the run-out signal in bands corresponding to the first harmonics (e.g. the first four harmonics) present in each portion of the run-out signal; and determines the inverse transform of the mean run-out signal. The mean calculation operation allows to reduce and estimate the measurement errors.

Figure 2b shows, in an example, the set of eccentricity measurements, represented in polar coordinates as eccentricity vectors El, EP with magnitude and phase of all the discs 2 of rotor 1 in the initial configuration. The processing station 11 is configured to perform the procedure described below with reference to figure 3.

Firstly, the processing station 11 checks the coherence of the data received from the measuring instrument 1 and corrects any acquisition errors (block 100) . For example, non-zero eccentricity values at the front bearing coupling portion 7 or at the rear bearing coupling portion 8 are indicative of an imperfect alignment of rotor 1 on the lathe (not shown) . The alignment error influences the eccentricity measurements of all discs 2. In this case, the processing station 11 determines, for all discs 2 of rotor 1, an eccentricity contribution caused by the inaccurate coupling to the lathe and subtracts the contributions thus obtained from the data supplied by the measuring instrument 12.

Once the errors have been eliminated, the measuring station 11 checks whether rotor 1 requires a balancing, e.g. by comparing a balancing index obtained from the data supplied by the measuring instrument against a threshold value (block 110) . If the balancing is not necessary (block 110, output NO) , the procedure ends (block 120) .

Otherwise (block 110, output YES), the processing station 11 identifies critical areas of rotor 1, i.e. pairs or groups of discs 2 between which the variations of eccentricity are high (block 130) . The procedure related to the identification of the critical areas will be described more in detail later on with reference to figure 4.

If there are at least two critical areas (block 140, output YES), the processing station 11 determines corrective actions of rotor 1 (block 150), otherwise (block 140, output NO) the procedure is ended (block 120) .

When a corrective intervention can be performed, the actions are defined by the processing station 11 as relative rotations between one or more pairs of contiguous discs 2, which remain angularly fixed with respect to the other discs 2 upstream and downstream, respectively. In an embodiment, the respective rotations are defined about a gravity axis of the disc 2 involved in the rotation and arranged closest to the front bearing coupling portion 7 (figure 1), in particular a gravity axis perpendicular to the faces of the disc 2 itself. In this manner, two sections of rotor 2 are rotated with respect to each other in practice. The sections are each delimited by a respective one of the two disc 2, between which the relative rotation is set.

The processing station 11 then determines the configuration resulting from the corrective actions (block 160) and evaluates whether the resulting configuration corresponds to a sub-optimal solution, i.e. a solution which at least locally optimizes an objective function indicative of the balancing of rotor 1 (block 170) .

If the resulting configuration does not correspond to a sub-optimal solution (block 170, output NO) , the processing station 11 excludes the identified resulting configuration and determines a new corrective action and a new resulting configuration (block 180) .

If, on the contrary, the resulting configuration is a sub-optimal configuration (block 170, output YES), the processing station 11 also evaluates the compliance to balancing criteria (e.g. if the expected balancing index associated with the resulting configuration is lower than the threshold value, block 180) .

The procedure is ended without solution (block 180, output NO; block 120) if the resulting configuration is not in compliance. If, instead, the test is passed (block 180, output YES), the identified solution is considered reliable and the corresponding corrective action is actually implemented (the rotor is dissembled and reassembled in accordance with the identified resulting configuration) .

Once the corrective actions have been performed, rotor 1 is measured again with the measuring instrument 12 and the processing station 11 checks the final balancing using updated data (block 190) .

The procedure for identifying critical areas is based on the observation that a rotor free from faults and supported by bearings near the ends has a substantially regular bending due to weight and tends to be arranged according to an arc. The discs which compose it have a given eccentricity, which increases from the ends towards the center of the rotor. The variation of eccentricity between contiguous discs is low and the line which joins homologous points, e.g. the centers, in sequences of consecutive discs, may also be approximated in satisfactory manner by a straight line. In the procedure described here, all the groups of N consecutive discs 2 (N may be conveniently chosen from 3, 4, 5) are identified. For each group, an interpolating function and a tolerance region with respect to an ideal case are determined.

The interpolating function is chosen from a class of functions which describe the configuration of rotor 1 in the absence of symmetry faults (ideal case) .

The tolerance reason is determined with respect to the interpolating function taking into consideration the asymmetries related to measurement inaccuracies and the acceptable machining and coupling tolerances between adjacent discs 2, including, for example:

the eccentricity between the self-centering housings on opposite faces of each disc 2 with respect to the outer edge of disc 2 where the run-out signal is detected (e.g. 10 μπι between edge and housing of each face) ; parallelism faults between self-centering housings on opposite faces of each disc 2.

The acceptance criterion is based on whether or not the eccentricity measurements related to a group of consecutive discs 2 belong to a corresponding tolerance region. If the eccentricity measurement of at least one of the discs 2 of the group under investigation lays outside the tolerance regions, the same group of discs 2 under investigation is labeled as a critical area.

More in detail, the critical areas are identified as described hereinafter with reference to figure 4.

Firstly (block 200), the processing station 11 determines, by a vector sum, a resulting eccentricity, indicated by ER in figure 2 (for simplicity, the resulting eccentricity ER is represented by way of example only and not in scale) . Then, the processing station 11 determines the projections in the direction of the resulting eccentricity ER of all the vectors indicating the eccentricity of the individual discs 2. The eccentricity components ER1, ERP thus obtained are shown in figure 5 in the positions of the respective discs 2 along axis A.

All the groups of N consecutive discs 2 present in rotor 1 are identified (block 210), i.e. in practice: the front hollow shaft with the two immediately successive discs; the three discs immediately downstream of the front hollow shaft; and so on, up to the last group which comprises the last two discs of the turbine section and the rear hollow shaft. In one embodiment, each group comprises three discs 2.

For each group of discs 2, the processing station 11 determines an interpolating function (F in figures 6 and 7; figure 4, block 220) and a respective tolerance region (R in figures 6 and 7; figure 4, block 230) .

The interpolating function F may be, for example, a polynomial function up to the fourth degree or a moving average of appropriate period. The degree of the interpolating function F may be chosen according to the number N of discs 2 belonging to each group. The higher the number N of discs 2 belonging to each group, the greater the degree of the interpolating function F. In the example in figures 6 and 7, the number N of discs 2 in each group is 3 and the interpolating function F is a polynomial function of first degree, i.e. a straight line .

The interpolating function F may be determined using the Ordinary Least Squares method on the eccentricity measurements of the respective group of consecutive discs (2) (ERJ, ERJ+1, ERJ+2 in the case in figure 6; ERK, ERK+1, ERK+2 in the case in figure 7) .

Furthermore, in one embodiment, the tolerance region R is a band of amplitude W which extends and is centered about the interpolating function F.

The processing station 11 checks whether the eccentricity measurements ERQ, ERQ+1, ERQ+N-1 related to all discs 2 of a same group lay within the respective tolerance region R (block 240) .

If the eccentricity measurements of all the discs 2 of the group lay in the respective tolerance region R (block 240, output YES) , the group of discs 2 is labeled as regular (block 250; as in the case of the eccentricity measurements ERK, ERK+1, ERK+2 in figure 7) .

If, on the contrary, at least one of the eccentricity measurements of the discs 2 of the group lays outside the respective tolerance region R (block 240, output NO) , the group of discs is labeled as critical area (block 260; as in the case of the eccentricity measurements ERJ, ERJ+1, ERJ+2 in figure 6) .

In the example in figure 5, the groups labeled as critical areas are identified with an ellipsis.

The selection of the interpolating function F and of the tolerance region R and the check that the interpolating function F belongs to the tolerance region R are repeated for all the identified groups of consecutive discs 2.

With reference to figure 8, in order to determine the corrective actions of rotor 1, the processing station 11 initially excludes the discs 2 in the groups labeled as critical areas (block 300) and identifies a further group of candidate discs 2 (block 310) between two subsequent critical areas. The group of candidate discs 2 comprises a minimum number of consecutive discs 2 , e.g. five .

For each pair of discs 2 of the group of candidate discs 2 (block 320) , the processing station 11 determines a plurality of modified configurations, determines the corresponding expected eccentricities for all the discs 2 of rotor 1 (block 330; also see the examples in figures 9 and 10, where the expected eccentricities are represented by eccentricity vectors El', EP' and by the respective components ER1',

ERP' along the direction of the resulting eccentricity ER' ) and calculates, for each modified configuration, a merit parameter indicative of the expected unbalance (block 340) .

For example, for each pair of discs 2, the processing station 11 determines 180 modified configurations with pitch of 1° in the 180° ± 90° range with respect to the initial configuration.

In one embodiment, the merit parameter is a residual momentum weight PRM, which is defined as the magnitude of the total momentum weight of the rotor and considers the eccentricity of its components and their mass. In particular, the residual momentum weight is defined as:

where Mj is the mass of the 2 J-th disc (possibly provided with respective blades) and EJ is the vector which defines the eccentricity of the 2 J-th disc.

Alternatively, the merit parameter may be an expected unbalance index IS, defined as:

where ||£7|| is the magnitude of the vector which defines the eccentricity of the 2 J-th disc, I _NOM is the nominal momentum of inertia of rotor 1 and C is a constant (e.g. 10 ⁹ ) .

Alternatively, the merit parameter may be the maximum cumulative eccentricity on the quadrant MEQ, which is a heuristic parameter and is calculated as follows :

a circular sector of a given amplitude (e.g. 60°) is defined about the actual rotation axis AR;

the magnitudes of the eccentricity vectors El, EP in the circular sector are summed up;

the circular sector is rotated in constant pitches about the entire circumference, adding the magnitudes of all the eccentricity vectors El, EP present in the sector one at a time;

the sector in which the sum of magnitudes of all the eccentricity vectors El, EP present in the sector is maximum is identified.

The cumulative eccentricity on the quadrant MEQ is a merit parameter to be optimized.

The indicated parameters allow a quantitative evaluation of the unbalance of a real rotor before and after a corrective intervention and the expected unbalance associated with configurations resulting from planned corrective actions.

In the ideal case of a perfectly symmetric and balanced rotor, the residual momentum weight PMR, the unbalance index IS and the maximum cumulative eccentricity on the quadrant MEQ are zero. Higher values represent pejorative balancing situations up to a threshold beyond which the rotor is not considered in compliance .

In a further embodiment, the merit parameter to be optimized is a combination of residual momentum weight PMR, of the unbalance index IS and of the maximum cumulative eccentricity on the quadrant MEQ normalized, made adimensional and weighed. A merit parameter of this type is robust because it summarizes different aspects individually represented by the residual momentum weight PMR, by the unbalance index IS and by the maximum cumulative eccentricity on the quadrant MEQ.

Once the processing station 11 has determined all the modified configurations and the corresponding values of the merit parameter, the configuration which optimizes the merit parameter (in this case, the modified configuration which minimizes the residual momentum weight; figure 8, block 350) is selected.

The result of the check once the selected modification actions have been performed with the described procedure is shown in figures 11 and 12.

In particular, updated eccentricity measurements of the discs 2 in the actual modified configuration are acquired. The updated eccentricity measurements of the discs 2 are represented by the polar coordinates as eccentricity vectors El", EP" in figure 11, while figure 12 shows updated eccentricity components ERl", ERP" along the direction of the resultant ER" .

In one embodiment, the processing station 11 is configured to identify several groups of candidate discs 2, each group being defined by a set of consecutive discs 2 between two respective critical areas of rotor 1. In this case, the processing station 11 determines corrective actions, substantially as described, with the possibility of also providing more than one relative rotation between continuous discs 2.

The method according to the invention allows a modified configuration of the rotor to be determined, which very probably leads to a satisfactory balance while measuring the benefits. Thereby, the risk of needing to repeat the complex operations of disassembling and reassembling the rotor several times is avoided or at least drastically reduced. The maintenance intervention times are thus minimized with considerable saving of costs. Furthermore, with a relatively low computing weight, the method allows a class of potentially satisfactory configurations to be delimited and an optimal solution to be identified in the class.

It is finally apparent that changes and variations can be made to the described method and apparatus without departing from the scope of the present invention as defined in the appended claims.