method for monitoring the condition of a tank intended to contain a liquid

Abstract

A method comprising the steps of: acquiring by a receiver a signal representative of acoustic waves propagating in a tank containing a liquid; preprocessing producing a signal of interest from the acquired signal, the signal of interest being selectively representative of horizontal shear waves of mode SH0 and / or SH1 among the acoustic waves; and processing comprising evaluating the existence of a tank defect and / or locating a tank defect, based on the signal of interest. Figure for the abstract: 5

Description

Title of the invention: method for monitoring the state of a tank intended to contain a liquid FIELD OF THE INVENTION

[0001] The present invention relates to a method for monitoring the state of a tank intended to contain a liquid, and a device for carrying out such monitoring. STATE OF THE ART

[0002] Methods are known from the state of the art using acoustic waves that have propagated in a structure to evaluate the state of a structure. A receiver placed on the structure acquires a signal representative of such acoustic waves. A defect and its position on the structure has an influence on the characteristics of the acquired signal. Thus, a signal acquired by the receiver while the structure is healthy and a signal acquired by the receiver while the structure is damaged may present differences that can be used to detect whether the structure has a defect, and where appropriate, to locate such a defect. Statement of the invention

[0003] The inventors have found that a particular problem arises when the structure to be monitored is a tank intended to contain a liquid. Signals representative of acoustic waves having propagated in a tank are sensitive to the liquid level inside the tank, and are also sensitive to the orientation of the tank (given that the liquid moves by gravity according to this orientation). Thus, when the liquid level changes inside the tank, or when the tank changes orientation, the signals acquired by the receiver also change, and this change can be unduly attributed to a deterioration in the condition of the tank. Thus, such a change can lead to false alarms.

[0004] An aim of the invention is to detect or locate more reliably a defect on a tank intended to contain a liquid.

[0005] This aim is achieved by a method comprising the following steps: acquisition by a receiver of a signal representative of acoustic waves having propagated in a reservoir containing a liquid; preprocessing producing a signal of interest from the acquired signal, the signal of interest being selectively representative of horizontal shear waves of SH0 and / or SHi mode among the acoustic waves; and processing comprising an evaluation of the existence of a reservoir defect and / or a location of a reservoir defect, from the signal of interest.

[0006] The horizontal shear waves SH0 and SHi are very insensitive to a variation in the level of liquid present in the tank and to the orientation of the tank. Thus, these specific waves carry the same information whether the tank is full, almost empty, standing or lying down. As a result, the detection or location of defects carried out by signal processing is made more reliable in the long term.

[0007] The proposed method may also comprise the following optional features, taken alone or in combination whenever technically possible.

[0008] Preferably, the preprocessing comprises steps of filtering the acquired signal into a filtered signal of the same duration as the acquired signal, the filtered signal consisting of a first temporal half and a second temporal half subsequent to the first temporal half, and of extracting the signal of interest in the second temporal half of the filtered signal.

[0009] Preferably, the signal of interest is at the end of the filtered signal.

[0010] Preferably, the acoustic waves are guided waves emitted by a transmitter positioned on the reservoir based on an excitation signal, and the signal of interest has a duration which is a function of a duration of the excitation signal, for example proportional to the duration of the excitation signal.

[0011] Preferably, the preprocessing produces, from the acquired signal, a signal of interest being selectively representative of horizontal shear waves of SHi mode, and the processing comprises an evaluation of the existence of a reservoir defect from this signal of interest.

[0012] Preferably, the preprocessing produces, from the acquired signal, a signal of interest being selectively representative of horizontal shear waves of SH0 mode, and the processing comprises a location of a reservoir defect from this signal of interest.

[0013] Preferably, the processing comprises the following steps: determining a score indicating an amount of energy transported by the signal of interest; comparing the score to a reference score range; and generating a detection result indicating the presence of a defect in the reservoir when the score is not included in a reference score range, and indicating the absence of a defect in the reservoir when the score is included in the reference range.

[0014] In one embodiment, the acoustic waves are waves emitted by a transmitter positioned on the tank. In another embodiment, the acoustic waves are not emitted by a transmitter.

[0015] Preferably, the signal of interest consists of a sequence of samples having respective amplitudes. When the acoustic waves are waves emitted by a transmitter positioned on the tank, the score is preferably a sum of the squares of the amplitudes. When the acoustic waves are not emitted by a transmitter, and the score is preferably a maximum amplitude among the amplitudes.

[0016] Preferably, the method further comprising steps of: measuring a pressure inside the tank, the measured pressure being associated with the acquired signal; and selecting the reference score range from a database containing score ranges respectively associated with pressure ranges, the score ranges having been determined from previous signals representative of acoustic waves having propagated in the tank while the tank was in a healthy state, and the pressure ranges having been determined from pressures measured inside the tank while the tank was in the healthy state.

[0017] Preferably, the acquisition is repeated by the receiver so as to acquire a succession of signals representative of acoustic waves having propagated in the tank, the succession of signals comprising: at least one signal originating from guided waves emitted by a transmitter positioned on the tank and at least one signal originating from acoustic waves not being emitted by a transmitter. The preprocessing and processing steps can then be applied to each signal forming part of the plurality of signals.

[0018] Preferably, the method further comprises steps of: determining a state of a machine intended to use hydrogen contained in the tank; implementing the acquisition by the receiver, the pre-processing and / or the processing provided that the state indicates that the system is in operation.

[0019] Preferably, the method further comprises steps of: determining an electromechanical impedance associated with the receiver; implementing the acquisition by the receiver, the preprocessing and / or the processing provided that the electromechanical impedance associated with the receiver is included in a range of reference impedances indicative of normal operation of the receiver.

[0020] When the acoustic waves are waves emitted by a transmitter positioned on the reservoir, the location of a reservoir defect can be implemented from a Fourier transform of the acquired signal and a group velocity of the horizontal shear waves of mode SH0 or SHi among the acoustic waves.

[0021] When the acoustic waves are not emitted by a transmitter, the location of a reservoir defect may comprise a hyperbolic triangulation carried out between three signals comprising the acquired signal and two additional signals acquired by two additional receivers, the additional signals also being representative of acoustic waves having propagated in the reservoir.

[0022] Preferably, the tank contains liquid hydrogen.

[0023] Another aspect of the present disclosure is a device comprising: a receiver configured to acquire a signal representative of acoustic waves having propagated in a tank for storing a liquid, and a signal processing module configured to implement preprocessing and processing. The preprocessing produces a signal of interest from the signal acquired by the receiver, the signal of interest being selectively representative of horizontal shear waves of SH0 or SHi mode among the acoustic waves. The processing includes an assessment of the existence of a reservoir defect and / or a location of a reservoir defect from the signal of interest.

[0024] Preferably, the receiver is a piezoelectric translator having a dielectric constant equal to d24.

[0025] Also provided is a system comprising a reservoir for storing a liquid and the foregoing device. DESCRIPTION OF FIGURES

[0026] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which:

[0027] [Fig.1a] schematically illustrates a system for monitoring a liquid tank, according to one embodiment.

[0028] [Fig.lb] comprises three views of a tank according to one embodiment, including a front view, a profile view and a back view.

[0029] [Fig.2], [Fig.3], [Fig.4], [Fig.5], [Fig.6] and [Fig.7] represent flowcharts of steps that may be implemented by the system of [Fig.1a].

[0030] Throughout the figures, similar elements bear identical references. DETAILED DESCRIPTION OF THE INVENTION

[0031] 1) Description of a system according to one embodiment

[0032] With reference to [Fig.1a] and [Fig.1b], a system 1 comprises a reservoir 2 for liquid, a machine 4 intended to use the liquid contained in the reservoir, and a device 6 for monitoring the state of the reservoir.

[0033] The reservoir 2 has an elongated shape along a longitudinal axis. The reservoir 2 comprises a wall having an internal surface delimiting a liquid storage cavity, and an external surface opposite the internal surface.

[0034] The wall comprises an annular portion extending around the longitudinal axis, and two end portions connected together by the annular portion. The internal surface thus has a substantially cylindrical general shape.

[0035] The wall of the tank 2 is formed by assembling three parts: a first part 20 forming the annular part, a second part 22 forming one of the end parts, and a third part 24 forming the other end part. The first part 20 is closed on itself around the longitudinal axis by one or more longitudinal weld beads 26 (therefore extending parallel to the axis Ion- longitudinal). The first part is fixed to the second part by means of a transverse weld bead 28 extending around the longitudinal axis, and is fixed to the third part by means of another transverse weld bead 30 extending around the longitudinal axis. The two transverse weld beads 28, 30 extend perpendicular to the longitudinal axis.

[0036] The wall of the tank 2 is made of metal, for example aluminum or steel.

[0037] The tank 2 may also comprise a mounting armor 32 extending over the two extreme parts of the wall.

[0038] It is noted that the thickness of the tank wall is measured between the internal surface and the external surface.

[0039] In a particular application, the tank 2 is configured to contain liquid hydrogen. As a reminder, liquid hydrogen is dihydrogen cooled below its condensation point, i.e. 20.28 kelvins (-252.87°C) at atmospheric pressure (101,325 Pa). It then has a density of 70.973 kg / m3.

[0040] The machine 4 is fluidically connected to the reservoir 2. Thus, liquid contained in the reservoir 2 is capable of being transferred to the machine 4.

[0041] The machine 4 is capable of taking two states: an activated state and a deactivated state. In the activated state, the machine 4 draws and / or uses all or part of the liquid contained in the tank 2. The use in question depends on the liquid. For example, when the tank 2 contains liquid hydrogen, this liquid hydrogen can be used by the machine 4 as clean energy to power fuel cells. In the deactivated state, the machine 4 does not use the liquid contained in the tank 2.

[0042] The machine 4 may be a vehicle, such as a land vehicle (automobile, train, etc.) or an aircraft (plane, helicopter, etc.).

[0043] The machine 4 may contain the device 6 and / or the reservoir 2. Alternatively, the monitoring device 6 and / or the reservoir 2 is outside the machine 4, as is schematically represented in FIG. 1.

[0044] The monitoring device 6 comprises a receiver 8 and a signal processing module 10.

[0045] The receiver 8 is configured to acquire signals representative of acoustic waves which have propagated in the tank 2, in particular horizontal shear waves of SH0 and / or SHi mode.

[0046] The receiver 8 is fixed to the tank 2, for example on the external surface of the tank 2.

[0047] The monitoring device 6 may comprise a transmitter 12 configured to emit acoustic waves into the tank 2, for example ultrasonic guided waves, so that these waves can be received by the receiver 8 after having propagated in the tank 2, and in particular through the wall of the tank 2.

[0048] The receiver 8 can also act as a transmitter 12.

[0049] Preferably, the transmitter 12 is capable of simultaneously emitting horizontal shear waves having both SH0 and SH1 modes.

[0050] Thus, the receiver 8 can receive: acoustic waves initially emitted by a transmitter 12 of the device (the device is then said to operate in active mode), and / or acoustic waves which are not emitted by such a transmitter 12, but which can be generated by the tank 2 itself or come from equipment not controlled by the device (this is then referred to as passive mode). In this regard, it should be noted that the monitoring device 6 may not include any transmitter 12; in this case the device 6 cannot operate in active mode.

[0051] Preferably, it is ensured that the following relationship is satisfied between the length of each receiver 8, denoted Le, a frequency f to which the receiver 8 is sensitive, and the thickness e of the wall of the tank:

[0052] Lc - 1 SH^ | SH^ ​​e)

[0053] In this equation, SH1 denotes the wavelength of the guided mode SHi, and SH0 denotes the wavelength of the guided mode.

[0054] Two variants of the active mode are also distinguished: the “pulse-echo” variant, in which a receiver 8 receives waves that it has itself emitted as a transmitter 12, and the “pitch-catch” variant, in which a receiver 8 receives waves emitted by a transmitter 12, 12 distinct from the receiver 8.

[0055] Preferably, the receiver 8 is a piezoelectric translator, that is to say a receiver 8 which converts the acoustic wave which it receives into an electrical signal (or vice versa, when this receiver 8 also plays the role of transmitter 12).

[0056] The piezoelectric transducer preferably has a dielectric constant equal to d24 and / or is suitable for operating at cryogenic temperatures. The PTZ4-d24 type transducer is a transducer which satisfies these two conditions cumulatively, and which can therefore advantageously be used as a receiver 8 or transmitter 12.

[0057] The monitoring device 6 may comprise several receivers 8 (and where appropriate several transmitters 12) having different positions on the external surface of the tank 2. Each receiver 8 conforms to the preceding description. In particular, it may be provided that each receiver 8 is in reality a transmitter-receiver, that is to say a translator playing the role of transmitter 12 of acoustic waves or receiver 8 for acoustic waves.

[0058] In the non-limiting example of [Fig. 1b], the reservoir 2 comprises nine receivers 8, including two receivers 8 arranged on one of the end portions of the wall, five receivers 8 arranged on the other end portion of the wall, and two receivers 8 arranged on the annular portion of the wall.

[0059] Each receiver 8 positioned on the annular part of the wall is close to the weld beads, in order to be more sensitive to defects in the weld beads. In particular, it may be provided that a distance between two receivers located on the annular part of the wall is less than or equal to twice the distance between an emitter and the nearest weld bead.

[0060] The signal processing module 10 is configured to apply processing to signals representative of acoustic waves which have been received by the or each receiver 8.

[0061] When the monitoring device 6 comprises one or more transmitter(s) 12, the signal processing module can also control the emission of acoustic waves by the or each transmitter 12.

[0062] The signal processing module 10 has a conventional structure. It comprises a processor configured to perform calculations based on data from signals to be processed.

[0063] The signal processing module 10 is connected to the machine 4, such that the module 10 is capable of knowing whether the machine 4 is in the activated state or the deactivated state.

[0064] The monitoring device 6 further comprises a memory 14 adapted to store data provided by the or each receiver 8 and / or by the signal processing module 10.

[0065] The monitoring device 6 further comprises a pressure sensor 16 adapted to measure a pressure inside the tank 2.

[0066] The monitoring device 6 may also comprise an output device for providing information to a user (for example in visual form, by means of a display screen or indicator lights) and / or a communication interface for transmitting such information to a remote user terminal.

[0067] 2) Description of a method according to one embodiment

[0068] A method implemented by the system 1 comprises two phases: a preliminary learning phase, and a subsequent monitoring phase. 2.1) Learning phase

[0069] During the learning phase, reservoir 2 is in a supposedly healthy state, and each receiver operates normally.

[0070] With reference to [Fig. 2], the learning phase comprises the following steps.

[0071] In a step 101, a transmitter 12 emits acoustic waves into the tank 2 on the basis of an excitation signal having a duration w, called the emission duration. These acoustic waves are ultrasonic guided waves. The excitation signal is for example a sinusoidal signal windowed by a Gaussian, Hamming, Hanning function or by another function which attenuates the secondary lobes of the cor- respondent.

[0072] Preferably, the excitation signal has a frequency f chosen as a function of the thickness e of the tank wall and as a function of the material from which this wall is made. When the tank wall is made of aluminum and its thickness e is equal to 5 mm, then the frequency f used is equal to 359 kHz. For example, when the tank wall is made of aluminum, the product fe is equal to 1.65 MHz.mm, and when the tank wall is made of steel, this product fe is equal to 1.8 MHz.mm.

[0073] The emitted waves propagate in the tank 2, and, in particular, pass through the wall of the tank 2. The emitted acoustic waves comprise the SH0 mode and / or the SHp mode

[0074] In a step 102, a receiver 8 acquires a first signal y(t) representative of such acoustic waves, after their propagation in the tank 2. In other words, the device 6 operates in active mode. The first acquired signal y(t) generally has a duration, called reception duration, which is greater than the excitation duration w. The receiver 8 transmits the first acquired signal y(t) to the signal processing module 10.

[0075] In a step 104, the pressure sensor 16 measures a pressure in the tank 2 temporally associated with the first signal y(t). By “temporally associated”, we mean that the measured pressure corresponds to the pressure inside the tank 2 during the propagation of the waves leading to the signal y(t). In practice, the pressure measurement step can be carried out during the acquisition of the signal y(t) by the receiver 8, or in a temporal vicinity close to this acquisition (before or after). The pressure measurement is stored in the memory 14.

[0076] In a step 106, the signal processing module 10 applies to the first acquired signal y(t) a preprocessing producing a signal of interest. The signal of interest is selectively representative of horizontal shear waves of mode SHi among the acoustic waves. This implies that the preprocessing eliminates any other mode possibly present in the acquired signal y(t), in particular the modes Ao, Ai, So and also the mode SH0.

[0077] The preprocessing 106 comprises a modal filtering of the acquired signal, so as to produce a filtered signal f (f) of the same duration as the acquired signal y(t). During this filtering, any mode other than the SHi mode is eliminated. The modal filtering may comprise the following sub-steps: • calculation of a Fourier transform of the acquired signal, which we denote by Y(w), • application to the signal Y(w) of a backpropagation of the guided waves to the position of the transmitter 12 which emitted them, so as to calculate the signal Y((w) = Y(o?)where A is a distance between the transmitter 12 and the receiver 8, is the wave number associated with the SHi mode, • i is the imaginary unit, • calculation of an inverse Fourier transform of the signal Kj(w), which we denote by • Time windowing of the time signal (f) so as to obtain a signal windowed yf(t) as follows: Q t > w yf(t) =y (f)W(f) , W(t) = • calculation of a Fourier transform of the windowed signal, which we denote by yf^ • application to the signal yf( of a propagation of guided waves at the receiver position 8, which amounts to calculating the signal calculation of an inverse Fourier transform of the signal £ (, 1st result of this calculation constituting the filtered signal Ç

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] The filtered signal £ ) is of the same duration as the signal y(t) initially acquired by the receiver 8. The filtered signal consists of a first temporal half and a second temporal half after the first temporal half. It was previously indicated that the preprocessing 106 carried out by the signal processing module 10 produces a signal of interest selectively representative of horizontal shear waves of mode SHi among the acoustic waves represented in the signal y(t). The signal of interest may be the filtered signal. However, the inventors were able to observe that the filtered signal is richer in information on the SHi mode in its second temporal half than in its first temporal half. Consequently, the signal of interest is advantageously a portion of the filtered signal extracted in its second temporal half. Most preferably, the signal of interest is at the end of the filtered signal. In other words, the signal of interest ends with the end of the filtered signal. The duration of the signal of interest can be determined as a function of the duration w. For example, we can provide that this duration is proportional to w. In a step 108, the signal processing module 10 calculates a first score indicating a quantity of energy transported by the signal of interest. Since the signal of interest consists of a sequence of samples with respective amplitudes, the first score can be calculated as the sum of the squares

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] of the respective amplitudes of these samples. The first score f is stored in memory 14 in association with the pressure measurement measured in step 104. Steps 101, 102, 104, 106 and 108 form a set of steps 100. This set of steps 100 is repeated over time, so as to obtain a succession of first scores respectively associated with pressure measurements in the tank 2, these pressure measurements possibly being of different values. In a step 110, the signal processing module 10 determines different ranges of first scores [éa g^] associated with different ranges of measured pressures. For this, the signal processing module 10 can divide the first scores into different groups according to the pressure measurements associated with them. For this, the module can divide the interval formed by all the measured pressure values ​​into distinct pressure ranges, for example of the same length. Each group is made up of the first scores associated with respective pressure measurements contained in one of the pressure ranges. Then, the signal processing module 10 can calculate, for each group, a range of first scores [g^ g£*] from the mean Ç and the standard deviation of the first scores constituting the group, as follows: ' & = + ,4 = with In these formulas, Na denotes the number of first scores in a group, and the first scores in the group are denoted f ..., N . The first score ranges] are stored in memory 14 in association with the corresponding pressure ranges. As will be seen later, these data are representative of the healthy state of reservoir 2, and can serve as a reference to determine whether reservoir 2 deteriorates in the monitoring phase following the learning phase. With reference to [Fig.3], the preliminary phase further includes the following steps. In a step 202, the receiver 8 acquires a second signal representative of ultrasonic waves having propagated in the tank 2. These waves are not emitted by a transmitter 12 fixed on the tank 2, but may simply result from the vibrations to which the tank 2 is subjected. In other words, the monitoring device 6 operates in passive mode. The duration of the second signal acquired in step 202 is predefined.

[0097] In a step 204, the pressure sensor 16 measures a pressure in the tank 2 temporally associated with the second signal. The information provided previously regarding step 104 is also applicable to step 204.

[0098] In a step 206, the signal processing module 10 applies to the second acquired signal a preprocessing producing a signal of interest. The signal of interest is selectively representative of horizontal shear waves of mode SHi among the acoustic waves represented in the second signal. Step 206 is similar to step 106.

[0099] In a step 208, the signal processing module 10 calculates a second score $pa indicating a quantity of energy transported by the signal of interest. The index “pa” of the second score refers to the passive mode in which the monitoring device 6 operates.

[0100] The signal of interest being made up of a sequence of samples having respective amplitudes, the second score £pa can be the maximum amplitude among these respective amplitudes.

[0101] The second Spa score is stored in memory 14 in association with the pressure measurement measured in step 204.

[0102] Steps 202, 204, 206 and 208 form a set of steps 200. This set of steps 200 is repeated over time, so as to obtain second scores respectively associated with pressure measurements in the tank 2, these pressure measurements possibly being of different values.

[0103] In a step 210, the signal processing module 10 determines different ranges of second scores associated with different pressure ranges. measured.

[0104] For this, the signal processing module 10 can divide the second scores into different groups according to the pressure measurements associated with them. For this, the signal processing module 10 can divide the interval formed by all the measured pressure values ​​into distinct pressure ranges, for example of the same length. Each group consists of the second scores associated with respective pressure measurements contained in one of the pressure ranges. Then, the signal processing module 10 can calculate, for each group, a range of second scores [from the average and the standard deviation of the second scores constituting the group, as follows:

[0105] = ^+3<7pa ( Spa = fpa ~ ^pa

[0106] With:

[0107] _ IJ y JVpa , ~ ~T — y Npa V pari ” pa /

[0108] >- i Çpa Npa^Tpï pctp

[0109] In these formulas, Npa denotes the number of second scores of a group, and the second scores of the group are denoted A „..., A ,T . b 1 ^paV '^paNa

[0110] The second score ranges are stored in memory 14 in association with the corresponding pressure ranges.

[0111] With reference to [Fig.4], the preliminary phase further comprises the following steps.

[0112] In a step 302, the signal processing module 10 determines an electromechanical impedance associated with the receiver 8, using a method known to those skilled in the art. The determined impedance is characterized by a module and by an argument.

[0113] In a step 304, the pressure sensor 16 measures a pressure in the tank 2 temporally associated with the determined electromechanical impedance. By “temporally associated”, it is meant that the measured pressure corresponds to the pressure inside the tank 2 during the determination of the impedance. In practice, the pressure measurement step can be carried out during the determination of the impedance, or in a close temporal vicinity of this acquisition (before or after). The pressure measurement is stored in the memory 14.

[0114] Steps 302, 304 form a set of steps 300. This set of steps 300 is repeated over time, so as to obtain electromechanical impedances respectively associated with pressure measurements in the tank 2, these pressure measurements possibly being of different values.

[0115] In a step 310, the signal processing module 10 determines different ranges of electromechanical impedances [A el] associated with different ranges of measured pressures.

[0116] For this, the signal processing module 10 can divide the impedances into different groups according to the pressure measurements associated with them. For this, the module can divide the interval formed by all the measured pressure values ​​into distinct pressure ranges, for example of the same length. Each group is made up of the impedances associated with respective pressure measurements contained in one of the pressure ranges. Then, the signal processing module 10 can calculate, for each group, a range of electromechanical impedances. caniques gl gh] from the mean f, and the standard deviation az of the impedances constituting the group, as follows:

[0117] ^ + 3(7- 4 =

[0118] with

[0119] / I & ^ 2

[0120]

[0121] In these formulas, Nz denotes the number of impedances of a group, and the impedances of the group are denoted £„iV..

[0122] The electromechanical impedance ranges [g' gè] are stored in the memory 14 in association with the corresponding pressure ranges.

[0123] The steps shown in Figures 2, 3 and 4 can be performed in any order. Furthermore, it is possible to alternate the three sets of steps shown in these three figures.

[0124] At the end of the learning phase, the following are present in memory 14: • ranges of first scores [g^ g^] in association with ranges of pressures measured in reservoir 2, • second score ranges [g|7O g^,] in association with pressure ranges measured in reservoir 2, • electromechanical impedance ranges [g^ g^] in association with pressure ranges measured in tank 2.

[0125] Above, it was assumed that groups of first scores, second scores and impedances are created independently during steps 110, 210 and 310. Alternatively, the signal processing module 10 can work on the basis of all the pressure values ​​measured during steps 104, 204, 304 to form groups comprising these three different types of data. Thus, the three steps 110, 210 and 310 can be grouped into one step implemented after the sets of steps 100, 200 and 300.

[0126] As will be seen later, the ranges of first scores and the ranges of second scores are representative of the assumed healthy state of the tank 2, and can serve as reference data to determine whether the tank 2 deteriorates in the monitoring phase after the learning phase. Furthermore, the ranges of electromechanical impedances are representative of a normal operating state of the receiver 8 and the transmitter 12, and can serve as reference data to determine whether the receiver 8 and the transmitter 12 are operating normally or not. in the monitoring phase.

[0127] So far, only one receiver 8 has been considered.

[0128] All the preceding steps can be repeated for several receivers 8 of the monitoring device 6, which has the consequence of multiplying the number of ranges stored in the memory 14 (in association with the corresponding receiver 8). 2.2) Monitoring phase

[0129] The monitoring phase, subsequent to the learning phase, comprises the following steps.

[0130] In a step 401, the signal processing module 10 determines whether the machine 4 is in the activated or deactivated state.

[0131] If the machine 4 is in the activated state, the monitoring device 6 implements a step 402 which corresponds to the set of steps 300 shown in [Fig.4]. In other words, the device obtains at step 402 a new electromechanical impedance associated with a receiver 8, and a new pressure measurement inside the tank 2 associated with this new impedance.

[0132] In a step 404, the signal processing module 10 applies a test to the new electromechanical impedance, so as to evaluate whether the receiver 8 associated with it is functioning correctly or not.

[0133] During step 404, the signal processing module 10 may search in the memory 14 for a range of pressures measured during the learning phase which includes the new pressure measured in step 402, and compare the new electromechanical impedance and the range of electromechanical impedances stored in the memory 14 in association with the range of pressures which includes the new measured pressure.

[0134] If the new electromechanical impedance is included in the range of electromechanical impedances, then the monitoring device 6 implements a step 406 aimed at detecting and / or locating possible defects in the tank 2. Indeed, it is considered in this case that the impedance has a normal value, and consequently that the receiver 8 and / or the transmitter 12 is operating normally.

[0135] If the new electromechanical impedance is not included in the range of electromechanical impedances, then step 406 is not implemented. Indeed, this situation reveals a possible malfunction of the receiver 8. In this case, attempting to detect or locate a fault in the tank 2 using the receiver 8 is not appropriate, because in view of its malfunction, these treatments could lead to incorrect or imprecise results.

[0136] If the machine 4 is in the deactivated state, then the steps 402 and 406 are not implemented for the receiver 8 considered. This saves the expense of energy needed to implement these steps.

[0137] The preceding steps are repeated over time. For example, it may be arranged to periodically trigger step 401.

[0138] Between two successive implementations of step 401, the pressure within the tank may change. In the case where the tank contains liquid hydrogen, a rise in pressure may occur when a portion of the liquid hydrogen transforms into gaseous hydrogen (due to a heat transfer between the inside and the outside of the tank). A drop in pressure may occur when a portion of gaseous hydrogen reduces (for example because it is consumed by the machine or evacuated from the tank for another reason). These pressure fluctuations are taken into account for the detection of defects. In this way, the impact of a pressure variation is not confused with the impact of a defect in the tank 2.

[0139] Furthermore, the preceding steps can be implemented for each receiver 8 of the monitoring device 6.

[0140] 2.2.1) Detection and / or location of fault (active monitoring)

[0141] With reference to [Fig.5], a first embodiment of step 406, which can be called “active tracking”, comprises the following sub-steps.

[0142] In a step 500, the monitoring device 6 once again implements the set of steps 100, using the receiver 8 for which the impedance test was previously carried out. The device operates here in active mode, since guide waves are emitted in step 101 and then received in step 102. Step 500 produces a new first score Ç from a new signal of interest representative of horizontal shear waves of mode SHi among the acoustic waves captured during the new implementation of step 102.

[0143] In a step 502, the device detects whether there is a defect in the tank 2, based on the new first score Ç. This processing produces a result capable of taking a so-called positive value indicating that a defect in the tank 2 has been detected, and a so-called negative value indicating the opposite.

[0144] During step 502, the signal processing module 10 can search in the memory 14 for a range of pressures measured during the learning phase which includes the new pressure measurement obtained during step 500, and compare the new first score and the range of first scores stored in the memory 14 in association with the range of pressures which includes this new measured pressure.

[0145] If the new first score £ is included in the stored first score range, then the signal processing module 10 generates a detection result having the negative value (no fault detected).

[0146] If the new first score f is not included in the range of first scores memorized, then the signal processing module 10 generates a detection result having the positive value (defect detected). This result can be notified to a user via the display screen, if present in the monitoring device 6, or by any other suitable means.

[0147] In the event of a detected fault (and preferably only in this case), the signal processing module 10 implements a step 504 of locating the fault in the tank 2. This location is for example implemented from a Fourier transform of the signal acquired during the new implementation of step 102, and from a group velocity of the horizontal shear waves of mode SHi in this acquired signal (velocity that the person skilled in the art knows how to determine). If the “echo pulse” mode was used for the implementation of steps 100, 102 (receiver 8 = transmitter 12), the module can calculate a position of a fault as follows:

[0148] . +f position(x3) = 4 J

[0149] with • Vg,-: speed of the SHp mode group • m = ( / / / <): dispersion law which links the angular frequency 0) to the wave number k. • Y(w) is the Fourier transform of the signal acquired y(t) during the tracking phase. • x3 corresponds to the direction of propagation.

[0150] In this formula, the division by 2 in the above equation is explained by the Fonde propagation path (outward / return) (outward: incident wave propagating from transmitter 12 / receiver 8 to the fault + return: reflected wave propagating from the fault to transmitter 12 / receiver 8).

[0151] 2.2.2) Detection and / or location of fault (passive monitoring)

[0152] With reference to [Fig.6], a second embodiment of step 406, which is called “passive tracking”, comprises the following sub-steps.

[0153] In a step 600, the monitoring device 6 once again implements the set of steps 200, using the receiver 8 for which the impedance test was previously carried out. The device 6 operates here in passive mode, since no transmitter 12 is used. Step 600 produces a new second score Çpa from a new signal of interest representative of horizontal shear waves of mode SHi among the acoustic waves captured during the new implementation of step 202.

[0154] In a step 602, the device detects whether there is a defect in the tank 2, based on the new second score Çpa. This processing produces a result sus capable of taking a so-called positive value indicating that a fault in tank 2 has been detected, and a so-called negative value indicating the opposite.

[0155] During step 602, the signal processing module 10 can search in the memory 14 for a range of pressures measured during the learning phase which includes the new pressure measurement obtained during step 500, and compare the new second score f pa and the range of second scores stored in the memory 14 in association with the range of pressures which includes this new measured pressure.

[0156] If the new second score Çpa is included in the stored range of first scores, then the signal processing module 10 generates a detection result having the negative value (no fault detected).

[0157] If the new second score Çpa is not included in the stored range of first scores, then the signal processing module 10 generates a detection result having the positive value (defect detected). This result can be notified to a user via the display screen, if present in the monitoring device 6, or by any other suitable means.

[0158] In the event of a detected fault (and preferably only in this case), the signal processing module 10 implements a step 604 of locating the fault in the tank 2.

[0159] The localization 604 may comprise the implementation of a known method called hyperbolic triangulation. This method consists of measuring the differences in arrival times of acoustic waves received by the receiver 8, and by two additional receivers 8 positioned on the tank 2. The three receivers 8 form what is commonly called a localization mesh. Let us denote Cai, Ca2 and Ca3 the three receivers 8 forming this mesh.

[0160] For the pair of sensors Cai and Ca2j, the signal processing module 10 calculates:

[0161] v A t\2~ _ ^2

[0162] A ri2 corresponds to the difference between the arrival times of the acoustic waves on Cai and Ca2. dl and d2 correspond respectively to the distance between the defect and Cai and Ca2. v is the speed of the waves. The equation above is the hyperbola H12 whose two foci are given by the position of the sensors Cai and Ca2.

[0163] For the pair of sensors Ca2 and Ca3, the signal processing module 10 calculates:

[0164] vAt23 = d2-d3

[0165] A t23 is the difference between the arrival times of the acoustic waves on Ca2 and Ca3. d2 and d3 correspond respectively to the distance between the defect and Ca2 and Ca3. The above equation is the hyperbola H23 whose two foci are given by the position of the sensors Ca2 and Ca3.

[0166] For the pair of sensors Ca3 and Cah the signal processing module 10 calculates:

[0167] vAtj^d^-di

[0168] A r23 is the difference between the arrival times of the acoustic waves on Ca3 and Ca i. d3 and dl correspond respectively to the distance between the defect and Ca3 and Cab The equation above is the hyperbola H31 whose two foci are given by the position of the sensors Ca3 and Cab

[0169] To locate a fault by triangulation during step 604, the processing module determines an intersection of these three hyperbolas.

[0170] 2.2.3) Detection and / or location of fault in active tracking and passive tracking

[0171] In a third embodiment, the detection and / or localization step 406 combines at least one implementation of steps 500, 502, 504 (active tracking) and at least one implementation of steps 600, 602, 604 (passive tracking). For example, the device can alternate between these two sequences of steps.

[0172] It should be noted that the results provided in the active mode and in the passive mode are complementary. Indeed, the active mode can lead to the detection of faults not detected in passive mode, and vice versa.

[0173] For example, the monitoring device 10 may condition the implementation of one of the two monitoring methods (active or passive) on the fact that the machine is in its activated state. For example: • when machine 2 is in the deactivated state: passive tracking (steps 600, 602, 604) is implemented in step 406 (if of course the impedance test authorizes it), but active tracking (steps 500, 502, 504), which consumes energy, is not implemented; • when machine 2 is in the activated state: active tracking and passive tracking are implemented in step 406 one after the other.

[0174] When active tracking and passive tracking are combined, it is also possible to distinguish between two operating modes of the monitoring device 6: a start-up mode, and an active / passive tracking mode.

[0175] The start-up mode is a transient mode, triggered when the monitoring device detects that the operating state of the machine 4 changes from the deactivated state to the activated state (in other words, when this machine 4 is started). When the device 6 is in the start-up mode, it can implement once the set of steps 300 (in relation to the electromechanical impedance), or even the active tracking (steps 500, 502, 504) and / or the passive tracking (steps 600, 602, 604).

[0176] The active / passive tracking mode is triggered following the startup mode, and lasts until the monitoring device detects a deactivation of the machine 4. When the monitoring device 6 is in the active / passive tracking mode, it can repeat the active tracking and the passive tracking alternately.

[0177] 2.3) Use of SH0 mode instead of or in addition to SHi mode

[0178] So far, pre-processings 106, 206 producing signals of interest have been described. selectively representative of horizontal shear waves whose mode is SHi (both in the learning phase and in the monitoring phase).

[0179] As a variant, these preprocessings 106, 206 produce signals of interest selectively representative of waves whose mode is SH0 (both in the learning phase and in the monitoring phase). In particular, the modal filtering steps detailed above remain applicable to this variant by replacing all the indices 1 with indices 0 in the notations used in the formulas related to the modal filtering (in particular, by replacing with the wave number k^ associated with the SH0 mode). Thus, in this variant, the parasitic modes Ao, Ai and Si are, where appropriate, eliminated during the mode filtering, as is the SHp mode

[0180] The inventors have found that signals of interest selectively representative of waves whose mode is SH0 have the same advantages as signals of interest selectively representative of waves whose mode is SHi (insensitivity to a change in liquid level in the tank 2 and to a change in orientation of the tank 2 in space).

[0181] In certain embodiments, the two aforementioned types of signals of interest (signal of interest SH0 + signal of interest SHi) can be produced during the preprocessing steps 106, 206 from the same signal y(t) acquired by a receiver during the monitoring phase, and the processing 406 can use these two signals of interest.

[0182] In a particularly advantageous embodiment of step 406, the signal processing module 10 performs the detection of a possible defect (step 502 or 602) on the basis of the signal of interest selectively representative of waves whose mode is SHi, and performs the localization of a defect (step 504 or 604) on the basis of the signal of interest selectively representative of waves whose mode is SH0. This embodiment leads to good detection and localization results, because the SH0 mode is less dispersive than the SHp mode. The SH0 mode is particularly suitable for the localization of large defects in the tank 2 and for the analysis of the dimensions of such defects. 2.4) Other variants

[0183] When the monitoring device 6 comprises several receivers 8, these receivers 8 can be divided into two categories: primary receivers 8 located in critical positions, in particular near a weld bead of the tank 2, and secondary receivers 8 located in less critical positions.

[0184] We have seen previously that the impedance measurement step 402, the test of this impedance 404, and the detection and / or location processing 406, are implemented on condition that the machine 2 which uses the liquid contained in the tank 2 either in operation (activated state).

[0185] In one embodiment of the monitoring phase, this condition only concerns the secondary receivers. Thus, step 402 is carried out for each primary receiver 8 unconditionally, therefore without taking into consideration the state of the machine 4. In other words, when the machine 4 is deactivated, only the primary receivers 8 are used, and when the machine 4 is active, the primary and secondary receivers 8 are used. This solution makes it possible to consume little energy to check the state of the tank 2 when it is not used by the machine 4, by concentrating on critical areas of the tank 2.

[0186] In yet another embodiment, the state of the machine 2 is not checked. Thus, during the monitoring phase, step 402 is triggered repeatedly, for example periodically, regardless of the state of the machine (activated or deactivated).

[0187] All the steps described above in relation to electromechanical impedance are advantageous, because they make it possible to avoid attempts at fault detection or fault location based on a malfunctioning receiver. These steps, however, remain optional.

[0188] Embodiments have also been described in which it is sought not only to detect defects but also to locate them. In other embodiments, the monitoring device is limited to detecting defects, and in other embodiments, the monitoring device is limited to locating defects. Indeed, the advantages provided by the SHi and SH0 modes apply to all embodiments.

[0189] It should also be noted that the tank as shown in [Fig. 1b] is only an example of a tank examined, and in which it is desired to detect or even locate defects. The method described above is applicable to any liquid tank. The tank examined may thus comprise only a single wall and / or not comprise welds. The mounting armor is optional, and may be installed on the annular part of the tank.

Claims

Claims

1. Method comprising steps of: • acquisition (402) by a receiver of a signal representative of acoustic waves having propagated in a reservoir containing a liquid, • preprocessing producing a signal of interest from the acquired signal, the signal of interest being selectively representative of horizontal shear waves of mode SH0 and / or SHi among the acoustic waves, • processing (406) comprising an evaluation of the existence of a reservoir defect and / or a location of a reservoir defect, from the signal of interest.

2. Method according to claim 1, in which the preprocessing comprises steps of • filtering the acquired signal into a filtered signal of the same duration as the acquired signal, the filtered signal consisting of a first temporal half and a second temporal half subsequent to the first temporal half, • extracting the signal of interest in the second temporal half of the filtered signal.

3. Method according to the preceding claim, in which the signal of interest is at the end of the filtered signal.

4. A method according to any preceding claim, wherein the acoustic waves are guided waves emitted by a transmitter positioned on the reservoir based on an excitation signal, and the signal of interest has a duration which is a function of a duration of the excitation signal, for example proportional to the duration of the excitation signal.

5. Method according to any one of the preceding claims, in which: • the preprocessing produces, from the acquired signal, a signal of interest being selectively representative of horizontal shear waves of SHb mode treatment includes an assessment of the existence of a reservoir fault from the signal of interest selectively representative of horizontal shear waves of SHp mode

6. Method according to any one of the preceding claims, in which: • the pre-processing produces, from the acquired signal, a signal of interest being selectively representative of horizontal shear waves of SH0 mode, • the processing comprises a location of a fault in the reservoir from the signal of interest selectively representative of horizontal shear waves of SH0 mode.

7. A method according to any preceding claim, wherein the processing comprises: • determining a score indicating an amount of energy carried by the signal of interest, • comparing the score with a reference score range, • generating a detection result indicating the presence of a defect in the reservoir when the score is not included in a reference score range, and indicating the absence of a defect in the reservoir when the score is included in the reference range.

8. Method according to the preceding claim, in which the signal of interest consists of a sequence of samples having respective amplitudes, and in which: • the acoustic waves are waves emitted by a transmitter positioned on the tank, and the score is a sum of the squares of the amplitudes, or • the acoustic waves are not emitted by a transmitter, and the score is a maximum amplitude among the amplitudes.

9. A method according to any preceding claim, further comprising steps of: • measurement of a pressure inside the tank, the measured pressure being associated with the acquired signal, • selection of the reference score range from a database containing score ranges respectively associated with pressure ranges, the score ranges having been determined from previous signals representative of acoustic waves having propagated in the tank while the tank was in a healthy state, and the pressure ranges having been determined from pressures measured inside the tank while the tank was in the healthy state.

10. Method according to any one of the preceding claims, in which • the acquisition is repeated by the receiver so as to acquire a succession of signals representative of acoustic waves having propagated in the tank, the succession of signals comprising: • at least one signal originating from guided waves emitted by a transmitter positioned on the tank, • at least one signal originating from acoustic waves not being emitted by a transmitter, • the pre-processing and processing steps are applied to each signal forming part of the plurality of signals.

11. Method according to any one of the preceding claims, further comprising steps of: • determining a state of a machine intended to use hydrogen contained in the tank, • implementing the acquisition by the receiver, the pre-processing and / or the processing provided that the state indicates that the system is in operation.

12. A method according to any preceding claim, further comprising steps of: determination of an electromechanical impedance associated with the receiver, implementation of receiver acquisition, preprocessing and / or processing provided that the electromechanical impedance associated with the receiver is included in a range of reference impedances indicative of normal receiver operation.

13. A method according to any preceding claim, wherein: • the acoustic waves are guided waves emitted by a transmitter positioned on the reservoir, and the location of a reservoir defect is implemented from a Fourier transform of the acquired signal and a group velocity of the horizontal shear waves of mode SH0 or SHi among the acoustic waves, or • the acoustic waves are not emitted by a transmitter, and the location of a reservoir fault includes a hyperbolic triangulation carried out between three signals comprising the acquired signal and two additional signals acquired by two additional receivers, the additional signals also being representative of acoustic waves having propagated in the reservoir.

14.

15. A method according to any preceding claim, wherein the reservoir contains liquid hydrogen. Device (6) comprising: • a receiver (8) configured to acquire a signal representative of acoustic waves having propagated in a tank (2) for storing a liquid, • a signal processing module (10) configured to put in operation: • pre-processing producing a signal of interest from the signal acquired by the receiver (8), the signal of interest being selectively representative of horizontal shear waves of SH0 or SHi mode among acoustic waves, and • processing including an assessment of the existence of a reservoir defect and / or a location of a reservoir defect, based on the signal of interest.

16. System (1) comprising a reservoir (2) for storing a liquid, and a device (6) according to the preceding claim.

Images