A METHOD FOR THE ULTRASONIC INSPECTION OF METAL COMPONENTS BY A SYNTHETIC APERTURE FOCUSING TECHNIQUE

TECHNICAL FIELD The present invention relates to a method for the ultrasonic inspection of metal components by a synthetic aperture' focusing technique. BACKGROUND ART

As it is known, in many industrial fields, it is fundamental to have non-destructive inspection techniques that allow to check the structural integrity of metal components, which form critical parts of a system subjected to particular stress. For instance, in the field of electric power plants, the large size rotors typically employed in machines such as steam and gas turbines or alternators, need to be subjected to inspections not only before set-up, but also after determined operation times, in order to prevent breakdowns which, although infrequent, may have disastrous effects.

Inspection techniques based on the use of ultrasounds (so-called "endosonic" or "boresonic" instrumentation) , which allow the analysis of the integrity in an absolutely non-invasive manner, have been developed for this purpose.

Substantially, a scan of the metal component is carried out using ultrasonic probes mounted on mobile heads. The echo signals captured by the probes are

processed in order to identify in the material possible discontinuities associated to defects or breaks .

Techniques directed to the elimination or at least the reduction of the contribution of the background by exploiting correlations between measurements performed from different positions have been designed, as well as processes that allow to identify "macroscopic" elements detectable, for example, by the envelope of the received signals. Specifically, the synthetic aperture focusing techniques (SAFTs) provide that a scan of the examined metal component be carried out by performing measurements at distances displaying a constant length along a scanning path. At each position on the path, the captured signal is sampled and stored in an indexed vector, which is normally termed A-SCAN. A predetermined number of vectors A-SCAN corresponding to adjacent positions defines a synthetic array. The correlations between adjacent vectors A-SCAN are dependant on the fact that the radiation diagrams corresponding to side positions in the synthetic array are at least partially overlapped to one another and to the radiation diagram corresponding to the central position. Substantially, the same points are therefore visible from several positions of the array itself, although from different angles and distances. Therefore, the same discontinuity causes echoes which are sensed with different delays according to the position of the probe in the array and to the depth at which the discontinuity is located.

The synthetic array is focused at a determined depth, i.e. the vectors A-SCAN are temporally realigned by introducing delays which are dependant on the position of the probe in the synthetic array and on the focusing depth (substantially, a translation of the samples of each vector A-SCAN is performed by an index transformation) . Subsequently, a mean operation is performed allowing to eliminate the reciprocally inconsistent background components, whereas the useful signal contributions are summed and are revealed. The vector resulting from the mean operation is associated to the central position of the synthetic array.

A problem associated to the use of the SAFT techniques is the selectivity of the focusing. The realignment renders the mean operation significant only at around the focusing depth, whereas at different depths the samples of the vectors A-SCAN having the same index

(i.e. the samples that are summed in the mean operation) are not actually temporally correlated. As it is impossible to extract useful information at depths other than the focusing depth, the conventional SAFT techniques may in practice only be exploited to obtain a better definition of defects already detected by other methods. For computational load reasons it is instead not advantageous to extensively apply the known SAFT techniques, for example, for the purpose of identifying previously unknown defects. Indeed, for each position of the synthetic array along the scanning path, the entire

focusing process (alignment and mean, as well as possible further accessory operations) would have to be repeated for different depth values and would therefore require excessively long processing times. DISCLOSURE OF INVENTION

Therefore, it is the object of the present invention to develop a method for the ultrasonic inspection of metal components by a synthetic aperture focusing technique, which does not display the above described drawbacks.

According to the present invention, there is developed a method for the ultrasonic inspection of metal components by a synthetic aperture focusing technique according to claim 1. BRIEF DESCRIPTION QF THE DRAWINGS

The present invention will now be described with reference to the accompanying drawings, which illustrate some non-limitative examples of embodiment, in which:

- .figure Ia is a longitudinally sectioned simplified side view of a rotor of a steam turbine;

- figure Ib is a front view of the rotor in figure Ia, sectioned along the plane of line IB-IB in figure Ia; figure 2 is a diagram schematically showing signals corresponding to the method according to the present invention and vectors deriving from the sampling of such signals; figures 3 and 4 are diagrams showing signals corresponding to the method according to the present

invention and correspondences between such signals and positions in the rotor in figures Ia and Ib; figure 5 is a diagram schematically showing visible volumes of an ultrasonic probe in different detection positions;

- figure 6 is a simplified flow chart corresponding to the method according to the present invention;

- figures 7a and 7b schematically show a simplified radiation diagram of an ultrasonic probe employed in the present method;

- figure 8 is a more detailed flow chart of a first portion of the flow chart in figure 6; figure 9 is a diagram schematically showing signals corresponding to the method according to the present invention;

- figure 10 shows a portion of the view in figure Ib, which is enlarged and provided with position references;

- figure 11 is a more detailed flow chart of a second .portion of the flow chart in figure 6;

- figure 12 is a diagram showing sizes corresponding to the method according to the present invention;

- figure 13 is a more detailed flow chart of a third portion of the flow chart of figure 6; - figure 14 shows a portion of the view in figure Ib, which is enlarged and provided with position and angle references; and

- figure 15 shows a portion of the view in figure Ia, which is enlarged and provided with position and angle references.

BEST MODE FOR CARRYING OUT THE INVENTION In the following description, reference will specifically be made to the use of the invention for the inspection of a hollow axis rotor of a steam turbine. This should although not be considered limitative, because the invention may advantageously be exploited for the inspection of metal components of any kind, especially alternator axes and gas turbine disks.

Figures Ia, Ib schematically show a rotor 1 of a steam turbine, which is not shown in detail here. An inspection bore 2, having radium Ro, extends around an axis A of the rotor 1 and is accessible from outside. Ultrasonic probes 3, which are controlled by an external control unit 5, are introduced in the inspection bore 2 by means of mounting elements, which are known and not shown, and are displaced through a plurality of detection positions along a predetermined scanning path. For the sake of simplicity, reference will be made to a single probe 3 in the following of the description, without this being considered limitative.

In the example described herein, the probe 3 covers a helicoidal path in contact with the surface of the inspection bore 2 at an increasing distance from the entrance of the inspection bore 2 itself. The detection positions along the path are uniformly angularly

distanced and are identified by respective detection angles θo, θi, θ2, ..., θi (figura Ib) .

At each generic detection position θi, the probe 3 is guided by the control unit 5 so as to alternatively emit ultrasonic inspection signals Si towards the body of the rotor 1 and receive return signals S _R , possibly containing echoes reflected by discontinuities 10 in the rotor 1 (substantially, after the transmission of inspection signals Si, the probe 3 is set in reception for a predetermined time) . The return signals S _R are sampled ^" with a sampling frequency Fs and stored in a control unit 5 in the form of indexed vectors A-SCANi (see figure 2) . More specifically, a vector A-SCANi is defined by the series of samples acquired during a measurement at the detection position θi and the generic element of the vector A-SCAN ₁ , obtained at the K-th sampling time K/F _s , is indicated by A-SCANi(K). The vectors A-SCANi are therefore temporally sorted and index K, which defines the position of each element A-SCANi (K) in the series forming the respective vector A-SCANi, is also representative of the depth the possible echo signal is referred to. For the sake of simplicity, in figure 2 the origin of the time axis coincides with the inspection signal emission time Si. In figure 2, the origin of the abscissa axis K (K = 0) coincides with the time at which the first sample of the return signal S _R is taken (at time T = T ₀ and at level R = Ro corresponding to the interface of the bore 2 of the rotor) . The vectors A-

SCANi may all be preliminarily acquired and processed later, or they may be processed as they are acquired.

In the presence of discontinuities 10, return echoes are contained in the vectors A-SCAN _x acquired from detection positions, from which the discontinuities 10 are visible (substantially, the vectors A-SCANi comprise waveforms indicative of discontinuities in the rotor body 1) . The starting index K of the echo provides information concerning the distance from the discontinuity to the probe 3, because the ultrasonic sounds are propagated at a constant and known speed. By way of example, figure 3 shows a discontinuity 10 positioned at a depth Ri from the axis A of the rotor of the rotor 1 (which is not shown here) as well as a vector A-SCANi and a vector A- SCANi ₊ j acquired by the probe 3 in a position θi, aligned with discontinuity 10, and in an offset position θi _+J , respectively. The vector A-SCANi contains an echo Ei, which starts at a time identified by an index Ki. In the vector A-SCANi ₊ j, instead, an echo Ei' starts at a subsequent time, identified by an index Ki' . The delay τi between the index Ki and the index Ki' is due to the difference in the path (2Di) of the return signals S _R captured by the probe 3 in the detection positions θi ₊ j. In figure 4, a discontinuity 10' is located at a depth R ₂ greater than the depth Ri. In this case, the delay τ _∑ is different from the delay τ _lr because the difference in the path (2D ₂ ) is also different.

As shown in figure 5, a synthetic array SAi centred on a generic detection position G ₁ is defined by 2M-1 vectors A-SCAN ₁ -( _M - _D , ..., A-SCANi, ..., A-SCANi _{+ (M} -i) , acquired by the probe 3 located in respective detection positions θi_( _J j- _I ), —, θi, ..., θ1-1 ₊ (M- _D • For the sake of simplicity, the central detection position will be indicated as detection position θi and the other side detection positions will be indicated as detection positions θi ₊ j. Furthermore, the central vector of the array SA _x , i.e. the vector corresponding to the detection position θi, will be indicated by the notation A-SCAN ₁ ; the other vectors A- SCAN or side vectors A-SCAN of the synthetic array SA ₁ will instead be indicated by the notation A-SCANi _+J , J being an integer comprised between -(M-I) and M-I and J≠O.

Each detection position θi _+J defines an element of the synthetic array SAi. The number 2M-1 of elements of the synthetic array SAi is determined so that in any detection position θi _+J of the synthetic array SAi, the volume V visible by the probe 3 (schematically shown in figure 5) at least partially covers an examined portion of the .rotor 1, which radially faces the central element of the synthetic array SAi (substantially, the probe 3 in the central detection position θi) . With reference to figure 6, the SAFT technique described here provides that, after an initialization

step (block 100), the synthetic array SAi is positioned in the generic detection position θi (block 110) .

Therefore (block 120), the vectors A-SCANi ₊ j corresponding to the elements of the synthetic array SAi are acquired. As previously pointed out, all of the vectors A-SCAN _I+J may be acquired preliminarily to the processing and stored.

An apodisating phase is then performed (block 130) , in which the vectors A-SCAN _I+J are normalised in order to consider the radiation diagram RD of the probe 3 (shown in a simplified manner in figures 7a, 7b) , which is is not isotropic, is characteristic of the type of probe 3 used and is known. Substantially, the elements of the vectors A-SCAN _I+J are weighted so as to compensate the dependence of the radiation diagram RD upon the angle α with respect to the pointing axis of the probe 3. The central vector A-SCANi does not require apodisation, because it corresponds to the maximum of the radiation diagram RD. In order to eliminate the effects of amplitude non-linearity in the measurement of the probe 3, the vectors A-SCAN ₁ -( _M - _D , ..., A-SCANi-i, A-SCANi ₊ i, ..., A- SCAN _I+ ( _M -D are further normalised with respect to the so- called DAC curve ("Distance-Amplitude Curve") of the probe 3 itself. The DAC curve is as well a known feature of the probe 3.

The side vectors A-SCAN ₁ - _(M -D , ..., A-SCANi-i , A-SCANj+i ,

..., A-SCANi ₊ ( _M -D (vectors A-SCANi _+J with J≠O) are then processed and phase-translated so as to obtain a multiple focusing (block 140) , according to the modes described hereafter. Here and in the following, "to apply a phase translation to an element of a vector" ^" has at least the following meanings:

- to perform a mere time translation of the element, assigning the corresponding sample a new position (index) in the vector;

- to substitute the sample corresponding to the element at issue with a new sample of the signal represented by the vector, the new sample having a given phase difference with respect to the original sample (consider, for example, a vector V, the elements V(K) of which are obtained from the sampling of a signal X(T) at times KT, being K = 0, 1, 2, ...; the substitution of the sample X(KT), corresponding to the element V(K), with a new sample X(KT + T*) is defined as a phase translation T* applied to the element V(K)).

In the second case the new sample and the original sample may obviously have a different value.

In this step, substantially, realigned vectors A- SCANRi-(M-i), ..., A-SCANRi, ..., A-SCANRi ₊ ( _M -D (from here on concisely indicated as vectors A-SCANRi _+J ) are generated, the elements A-SCANRi _+J (K) of which are indicative (index K being the same) of the continuity condition of a same

portion of the rotor 1 (substantially, of a portion of rotor 1 aligned to the probe 3 in the central detection position θi and corresponding to the element A-SCANi(K) of the central vector A-SCANi) • A central vector A-SCANR ₁ coincides with the central vector A-SCANi.

Then, the mean of the realigned vectors A-SCANRi ₊ j is calculated (block 150) and a focused vector A-SCANSi is determined (block 160), the elements A-SCANSi(K) of which are the mean of the corresponding elements A-SCANRi ₊ j(K) of the realigned vectors A-SCANR _I+J . The focused vector A- SCANSi is associated to the central detection position θi of the synthetic array SA ₁ .

A preliminary inspection procedure (hereinafter described in greater detail) is then performed to identify in real time the focused vectors A-SCANSi containing echoes, which may be associated to possible discontinuities 10, 10' in the rotor 1, and store them

(block 165) . In this step, the focused vectors A-SCANSi which do not contain echoes are eliminated instead, thus avoiding the storing of considerable amounts of unnecessary data.

If the scan of the rotor 1 has not already been completed (i.e. if the synthetic array SAi has not yet reached the end of the scanning path, output NO from block 170) , the synthetic array SAi is repositioned and centred on a detection position θi ₊ i following the detection position θi which has just be analysed (block

180) . The calculation is resumed with the acquisition of the vectors A-SCAN ₁ -M ₊₂ , ..., A-SCANRi ₊ i, ..., A-SCANRi _+M corresponding to the new position of the synthetic array SA ₁₊₁ (block 120) . If, instead, the scan of the rotor 1 is completed (output YES from block 170) , an identification procedure is performed to identify, from the stored focused vectors A-SCANSi, echoes El and El' associated to discontinuities 10, 10' in the rotor 1 (block 190; also see figures 3 and 4) . Such a procedure will be described in detail hereinafter.

The realignment and focusing operations (block 140 of figure 6) are performed as follows. With reference to figures 8 and 9, the central vector A-SCANi is subdivided (block 200 in figure 8) in a plurality of sections Si, S ₂ , ..., S _N (figure 9) , according to how many are the different focusing depths selected.

For each vector A-SCANi _+J of the synthetic array SAi, a maximum scanning depth EOSi _+J is then defined (block 210, figure 8) . In the case of the central vector A-SCANi of the synthetic array SAi, the maximum scanning depth EOSi is used to define a range δINT = EOSi/N within which the focusing depth is assumed to be constant. A generic section S _L (with 1<L<N) therefore comprises a number δ of elements of the central vector A-SCANi defined by values of the index K in the range from (L-I) δ to Lδ-1. For each

section S ₁ ,, a focusing depth of section R ₁ is then defined, obtained from

where V is the propagation speed of the ultrasonic signals in the material forming the rotor 1 and is known. Substantially, it is considered that the elements A- SCANi(K) of the section S _L have the same focusing depth of section R ₁ .

Subsequently (block 220) , the distances Di _+J , _L are calculated between the points of the internal surface of the rotor 1 corresponding to the detection positions θi _+J

(except for the central detection position θi) and the points corresponding to the focusing depth R _L associated to each section S _L (see figure 10) .

The distances Di _+J , _L depend, besides, upon the type of probe employed. In the embodiment described here, a so-called planar probe is used and the distances D _I+J , _L are obtained from:

D _I+J , _L = V(Ro + R _L ) ² + RQ -2R ₀ (R ₀ +R _L )cos(θ _I+J -G ₁ ) ( 2a)

Ih the case of a CW ("clockwise") angle probe, there applies:

D _ϊ+J/L = V ^R _L + 2R ₀ ² (1 -A)- 2R _L R _{0 λ} /2(1 -A)B (2b)

where A = COs(Gi _+J -O ₁ ) and B = cos [ (θ _I + _J -θ _I ) /2+φ] and φ<π/2 represents the inspection angle characteristic of the specific probe (figure 14).

In the case of a CCW ("counterclockwise") angle probe, the same relation (2b) still applies, with φ>π/2 (figure 14) .

For a RW ("rearward" with respect to the axis A of the rotor 1) angle probe, there applies:

Di _+J(L = V( ^F " ^G f + ( ^{H ~ M} ) ² + (zi + N - z _I+J ) ^{2 (} 2c ⁾

where F = (Ro + R _L sin φ) sin θi, G = Ro sin θi _+J , H = (Ro + R _L sin φ) cos Q ₁ , M = R ₀ cos θi _+J , N = R _L cos φ, with φ>π/2 (in this case, the axial coordinate Z must also be taken in account, along axis A, figure 15) .

For a FW ("forward" with respect to the axis A of the rotor 1) angle probe, the same equation (2c) is used, with φ<π/2 (figure 15) .

The distances D _I+J , _L are therefore used to calculate delays τi _+J/L (J = -(M-I),..., M-I; L = I, 2, ..., N) to be applied to the elements of the vectors A-SCANi _+J in order to obtain a phase correction that allows to separately and simultaneously focus all of the sections Si, S ₂ , ..., S _N . Specifically, the delays ti ₊ j, _L are obtained from

τi _+J ,L = 2 ( D _{I+Jr L} - R ₁ ) /V ( 3 )

and keep into account the difference in the path the ultrasonic pulses need to cover to reach a point in the rotor 1, which corresponds to an element of the vector A- SCANi of a section S _L from the different detection positions θi _+J . For each section S _L , the delays T _I + _J , _I , allow to identify corresponding sections S _I + _J , _L of the side vectors A-SCANi ₊ j (J≠O) containing information corresponding to the same portion of the rotor 1 explored by the elements A-SCAN ₁ (K) of the section S _L of the central vector A-SCANi (figure 9) .

The calculated delays T _I + _J , _L are then applied to the respective elements A-SCAN _I+J (K) to perform the multiple focusing, so that the same value of the index K is associated to the elements containing corresponding information.

Substantially, an element A-SCANi _+J (K+τi ₊ j, _L ) of each side vector A-SCANi _+J would correspond to an element A- SCANi(K) of a section S _L of the central vector A-SCANi (whereby (L-I) δ < K < Lδ-1) . As an effect of the non- linearity reciprocally connecting the distances Di _+J , _L , although, the times K+X _I+J , _L do not correspond to actual sampling times and, furthermore, the sections of the side vectors A-SCAN _I + _J normally contain a number of samples

other than the number of the corresponding sections Si, S ₂ , ..., S _N of the central vector A-SCANi.

In order to perform the multiple refocusing, the vectors A-SCAN _I+J are interpolated (figure 8, block 230) and subsequently resampled (block 240) , considering the delays ti ₊ j _/L . The realigned vectors A-SCANR _I + _J are thus generated (block 250) . Specifically, the interpolation and resampling are performed so as to have δ elements for each section Si _+J , _L in each of the realigned side vectors A-SCANRr ₊ j (J≠O) and so that the information associated to the elements A-SCANRi _+J (K) of the realigned side vectors A-SCANR _I + _J corresponds to the information associated to the element A-SCANRi(K) of the realigned central vector A-SCANRi having the same index K (the realigned central vector A-SCANR ₁ coincides with the central vector A- SCANi) • The interpolation and the resampling in practice perform the phase translation:

separately and independently for each section S _L . The resampling is obviously always possible, substantially without the loss of information, provieded that the sampling frequency F ₃ is selected in accordance with the well-known Nyquist-Shannon sampling theorem.

The realigned vectors A-SCANRi _+J determined in this

manner, are then used for the calculation of the mean and the generation of the focused vectors A-SCANSr (blocks 150 e 160 of figure 6) .

Substantially, according to the process described, the sections S _L of each vector A-SCANi ₊ j are realigned and focused separately, so that the contributions of useful signal contained therein are consistently summed during the mean operation. The background components, which are inherently not consistent, are, instead, substantially eliminated or, in any case, reduced by the mean operation, rendering the useful signal more easily distinguishable. A single mean operation is sufficient to obtain focused vectors A-SCANSi regardless of the depth. The process is therefore light enough from a computational point of view to be extensively applied along the entire scanning path. In this manner, it is possible to also detect minor structural defects, which would normally be covered by the background and would not be detected. Figure 11 shows the preliminary inspecting step of the defects 165 in figure 6 in greater detail. At first

(block 300), a negative detection threshold TH _N and a positive detection threshold THp are defined, which need to be exceeded for the identification of waveforms E associated to discontinuities in the body of the rotor 1 (also ^" see figure 12) . For this purpose, a series of

reference vectors A-SCAN is detected in a test portion of the rotor 1 having no defects. The acquired signals in the absence of defects will be zero-mean radio frequency signals , usually having a non-constant amplitude as the depth varies along the pointing axis of the probe. The mean and both the positive and the negative envelope of the reference vectors A-SCAN are calculated. The negative detection threshold THn and the positive detection threshold TH _P are defined by the respective negative envelope and positive envelope fractions, to which the possible offset is summed.

Therefore, a focused vector A-SCANSi that must be analysed (block 310) is selected, and a threshold check is performed (block 320) . Specifically, the selected focused vector A-SCANSi is considered significant if the following conditions are fulfilled:

- the focused vector A-SCANSi contains a first element A-SCANSi (Ki) and a second element A-SCANSi(K ₂ ) such that A-SCANSi (Ki) >THp and A-SCANSi (K ₂ ) <TH _N ; and - Ki<K _EF and K ₂ <K _E F (K _E F is the index at which the background echo is detected and it may be determined either empirically or by previously knowing the thickness of the examined section) .

If these conditions are not fulfilled, the focused vector A-SCANS ₁ is not indicative of possible defects and is ignored (output NO from block 320, block 330) and the

procedure ends (block 360) .

The focused vectors A-SCANSi which have passed the threshold check are subjected to a frequency check (output YES from block 320, block 340) . Substantially, the return signals S _R related to discontinuities in the rotor 1 have a cisoid trend with oscillation frequency substantially equal to the fundamental frequency Fo of the inspection signals S ₁ emitted from the probe 3. In this phase, the frequency with which the return signal S _R contained in the focused vector A-SCANSi exceeds both the positive detection threshold THp and the negative detection threshold TH _H , is checked (figure 12) . If such a frequency is compatible with the fundamental frequency F ₀ (output YES from block 340) , the focused vector A- SCANSi is subjected to a time analysis (block 350) , otherwise it is ignored (output NO from block 340, block 330) and the procedure ends (block 360) .

The time analysis comprises the following operations: - extracting local maximums and minimums contained in the focused vector A-SCANSi;

- checking the compatibility of the frequency of the local maximums and minimums with the fundamental frequency Fo. Alternatively, the time analysis comprises calculating a correlation index between the maximum and,

respectively, minimum elements A-SCANSi(K) of the positive detection threshold THp and the negative detection threshold TH _N and samples of the response to the impulse of the probe 3. The criterion to fulfil is that the correlation index be higher than a predetermined threshold.

If the time analysis is not passed, the focused vector A-SCANS ₁ is ignored (output NO from block 350, block 330) . If, on the contrary, the time analysis is also successful, the presence of a likely discontinuity or defect and its position are stored (output YES from block 350, block 355) . In both cases, the procedure ends (block 360) .

Figure 13 shows in greater detail the identification procedure (which is performed in a post-processing step, even off-line) of discontinuities 10, 10' in the rotor 1

(block 190 of figure 6) . Firstly, a planar consistency check is performed comparing focused vectors A-SCANSi corresponding to adjacent detection positions Q ₁ and verifying the consistency of the information contained

(block 400) . Specifically, the planar consistency check corresponding to a detected defect is passed if:

- different adjacent focused vectors A-SCANSi, corresponding to detection positions θi (reciprocally contiguous in an observation angle α' ) , contain indications of a defect;

- the indications of defect in adjacent focused vectors A-SCANSi are compatible as depth is concerned, i.e. the difference between maximum depth and minimum depth is smaller than a predetermined value; and - the observation angle α/ of the defects is compatible with the amplitude of the radiation diagram of the probe 3.

If the planar consistency check fails (output NO from block 400) , the focused vectors A-SCANSi containing incongruent defect indications are ignored (block 410) and the procedure ends (block 420) , otherwise, a volumetric consistency check is performed (output YES from block 400, block 430) . Specifically, there are checked: - the consistency of the indications of defects in adjacent focused vectors A-SCANSi in axial direction; and

- the correlation for small defects (smaller at the aperture of the probe) between "scattered" intensity and amplitude' of the intersection between defect and radiation diagram.

If the volumetric consistency check fails, the defect indications and the corresponding focused vectors A-SCANS ₁ are ignored (output NO from block 430, block 410), otherwise the defect indications are validated (output YES from block 410, block 440) . In both cases, the procedure ends (block 420) .

It is finally apparent that modifications and variants may be made to the method described, without departing from the scope of the present invention, as defined in the accompanying claims. Specifically, the method according to the invention may be advantageously exploited for the inspection of any type of metal component, not only for hollow axis rotors. For instance, it is possible to analyse gas turbine disks, even though the geometry is clearly different. As previously indicated, it is also possible to use a plurality of probes of various types both simultaneously and alternatively.