SYSTEM FOR DETECTING AND LOCATING A MATERIAL LEAK, AND ASSOCIATED METHOD.

Abstract

The invention relates to an instrumentation kit and a system capable of implementing a method for detecting a material leak from an industrial device, comprising the steps of measuring a vibration of a peripheral element arranged to receive a portion of material falling from the industrial device; and, in response to this measurement, and by means of an electronic computing system, generating a calculated signal representative of the fall of the falling portion of material at a drop point located on the peripheral element. Figure to be published with the abstract: Fig. 1

Description

Title of the invention: SYSTEM FOR DETECTING AND LOCATING A MATERIAL LEAK, AND ASSOCIATED METHOD. TECHNICAL FIELD OF THE INVENTION

[0001] The invention relates to the detection and location of a leak of liquid, soft or solid material from an industrial device transporting, transforming or containing this material. TECHNOLOGICAL BACKGROUND

[0002] The manufacturing industry, for example in the fields of chemistry, pharmacy, metallurgy, agriculture, food, semiconductors, household appliances or even petroleum product refinery operations, transforms, handles and transports matter in different forms. This matter can be in solid form such as grains, powders, or masses of food, chemical, mineral or metallic products. This matter can also be in liquid form such as water, chemical solutions, beverages, oil, petroleum, or molten metal.

[0003] In any case, it may happen that material escapes from the circuit that it is intended to follow. For example, solid material may fall from a conveyor belt or liquid material transport pipes may leak, allowing the transported liquid to escape.

[0004] In smaller installations, detecting and locating the place from which the material is escaping remains a simple and rapid operation. However, this operation rapidly increases in complexity and duration with the size of the installation concerned.

[0005] However, a leak from an industrial installation can cause a total or partial shutdown of a production line, depending on the quantity or the danger of the product concerned, for example in the event of environmental contamination when it involves toxic chemical material. There is therefore a strong interest in detecting a leak as early as possible and, ideally, locating it and characterizing its importance.

[0006] Document US 10,692,351 B2 describes the detection of a leak in a water heating system by means of a humidity detection device.

[0007] Document US 8,064,738 B2 describes the detection and localization of a leak in a pipe by means of an optical fiber arranged along the pipe and sensitive to humidity.

[0008] Document US 2016 / 097696 A1 describes a piezoelectric sensor in contact with a piping element (pipe, valve, fitting), which is used to detect a leakage through vibrations caused by a fluid passing through the pipe at the leak level.

[0009] The devices known to the applicant are essentially adapted to liquids and require direct contact between a sensor and the leaking product and / or the piping conveying it.

[0010] A strong need exists for a versatile leak detection method, i.e. one that operates independently of the nature of the leaking product, does not require direct contact with the leaking product or the leaking device, and is capable of assessing the location and extent of the leak and of monitoring it. Description of the invention

[0011] The applicant's objective is to propose a method and the associated system making it possible to meet the needs listed in the preceding paragraph, applying generally to leaks of solid, liquid or soft material.

[0012] The principle on which the system and the method according to the invention are based is that of the indirect detection of a leak on the basis of the vibrations generated by the impact of a portion of material falling onto a peripheral element from an industrial device. This peripheral element preferably takes the form of a plate extending under the industrial device without being limited thereto, so as to receive the portion of material falling from it. The peripheral element may also consist of all or part of another element which can play the role of the plate, such as for example a retention tank, an example chosen in the embodiments detailed below.The peripheral element is instrumented so that the user can measure the vibrations propagating on its surface and generated by the impact of a portion of material falling from the industrial device due to a leak, measurements making it possible to detect the occurrence of the leak, to quantify its importance, and to locate it.

[0013] In order to achieve this aim, one aspect of the invention is a method for detecting a leak of material from an industrial device, comprising the steps of:

[0014] - measuring a vibration of a peripheral element arranged to receive a portion of material falling from the industrial device; and,

[0015] - in response to this measurement, and by means of an electronic calculation system, generate a calculated signal, representative of the fall of the portion of material falling at a drop point located on the peripheral element.

[0016] The method according to the invention makes it possible to detect an impact of a fall of material from an industrial device, and therefore to detect the occurrence of a leak within this industrial device. Proximity or immediate contact of the sensor with the leaking element or the material having fallen is not necessary, the impact being detected indirectly via the vibrations it generates at the level of the peripheral element.

[0017] Thus, it is a method that can be applied indifferently to leaks of solid matter or liquid or soft matter, whatever the chemical nature of this matter. The only restrictive criterion as to the scope of application of this system is that the fall of the material must generate vibrations at the peripheral element capable of being detected by the vibration sensor.

[0018] Furthermore, using the peripheral element, which owes its designation to the fact that it is peripheral to the industrial device being monitored, to detect leaks, allows for monitoring that is completely neutral with respect to this industrial device and independent of any action, modification, maintenance or other that may concern it. It is also possible to install a system for monitoring the industrial device without having to modify it or shut it down, simply by instrumenting a pre-existing element such as a retention tank to detect leaks, or by installing such a tank or more generally a peripheral element that is suitable for the situation in question.

[0019] According to additional non-limiting characteristics of the method according to the invention, considered individually or according to any technically feasible combination:

[0020] - the vibration can be measured at at least two distinct points of the peri-element spherical, and, in response to these measurements, the electronic computing system can generate (540) and transmit a position signal representative of a position of the drop point;

[0021] - the method may further comprise (i) a step of instrumenting the peri element spherical with at least one first sensor when this peripheral element is already arranged vertically above the industrial device, or (ii) a step of instrumenting the peripheral element with the at least one first sensor, followed by a step of arranging the instrumented peripheral element vertically above the industrial device;

[0022] - the calculated signal can be representative of at least one piece of information chosen from an occurrence of the falling portion of the falling material, a kinetic energy of the falling portion of the falling material, a cumulative amount of material having fallen from the industrial device during a given period of time, a distance between a location of the first sensor and the point of fall of the falling portion of the falling material on the peripheral element (110, 110'), a state, either solid, liquid or soft, of the falling portion of the falling material;

[0023] - the calculated signal can be calculated on the basis of a vibration detection signal and a signal representative of a nature of the portion of material, the calculated signal being at least representative of a fall height of the portion of material;

[0024] - the step of detecting the vibration of the peripheral element can be implemented at by means of a first vibration sensor comprising a thin monocrystalline piezoelectric layer extending along an extension plane, of less than 50 pm of thickness and dimensions greater than 100 pm in its plane of extension, this thin layer being rigidly attached to the peripheral element; and

[0025] - the peripheral element can be a retention tank.

[0026] In addition to the advantages already mentioned, the method according to the invention can also make it possible to determine the location, in the horizontal plane defined by the peripheral element, of a drop point of a portion of material falling from the leak, but also the drop height. Each of these pieces of information makes it possible to greatly simplify the search for the origin of the leak and to save valuable time in its location.

[0027] The method can also make it possible to quantify the leak, by evaluating the kinetic energy released by the material during its impact on the peripheral element, energy directly linked to the quantity of material falling, and therefore to the size of the leak. This characterization of the leak makes it possible to schedule an emergency intervention in the event of a significant leak or, on the contrary, to prioritize other actions in the event of a minor leak.

[0028] The invention extends to an instrumentation kit configured to equip a peripheral element intended to be installed directly above an industrial device so as to form a system for detecting a leak of material falling from the industrial device, the kit comprising:

[0029] - at least one vibration sensor provided for generating and emitting a signal of sensor in response to a vibration of the peripheral element resulting from an interaction between a portion of the falling material and the peripheral element; and

[0030] - an electronic computing system capable of being functionally connected to the first sensor and being provided to, in response to the first sensor signal, generate and transmit a calculated signal representative of the fall of the portion of material at a drop point located on the peripheral element.

[0031] According to additional non-limiting characteristics of the instrumentation kit according to the invention, considered individually or according to any technically feasible combination:

[0032] - the first calculated signal can be representative of at least one chosen piece of information among an occurrence of the fall of the portion of the falling material, a kinetic energy of the portion of the falling material, a cumulative quantity of material having fallen from the industrial device during a given period of time, a distance between a location of the first sensor and the point of fall of the portion of the falling material on the peripheral element, a state, either solid, liquid or soft, of the portion of falling material;

[0033] - the first calculated signal can be calculated based on the first sensor signal and of a signal representative of a nature of the portion of material, the first calculated signal being able to be at least representative of a fall height of the portion of matter ;

[0034] - the instrumentation kit may further comprise a plurality of vi sensors each configured to output an associated sensor signal in response to vibration of the peripheral element, and operatively connectable to the electronic computing system, wherein the electronic system may be further configured to generate and output a position signal representative of a position of the drop point, based on the sensor signals; and

[0035] - the at least one sensor may comprise a mono piezoelectric thin layer crystalline extending along a plane of extension, less than 50 pm thick and with dimensions greater than 100 pm in its plane of extension.

[0036] The invention extends to a leak detection system for an industrial device comprising a peripheral element designed to be installed directly above an industrial device and equipped with the instrumentation kit according to the invention.

[0037] According to additional non-limiting characteristics of the system according to the invention, considered individually or according to any technically feasible combination:

[0038] - the system may further comprise a suspension device configured to so as to isolate the peripheral element from mechanical vibrations in its environment; and

[0039] - the peripheral element can be a retention tank. BRIEF DESCRIPTION OF THE FIGURES

[0040] Other characteristics and advantages of the invention will emerge from the detailed description of the invention which follows with reference to the appended figures in which:

[0041] [Fig.l] [Fig.l] represents a general view of a first system according to the invention, in use and seen from the side in a vertical plane;

[0042] [Fig.2] [Fig.2] represents the system of [Fig.l], seen from below;

[0043] [Fig.3] [Fig.3] represents experimental measurement curves;

[0044] [Fig.4] [Fig.4] illustrates a comparison between curves obtained experimentally and curves obtained by simulation;

[0045] [Fig.5] [Fig.5] illustrates curves obtained experimentally for four different materials having fallen at the same point;

[0046] [Fig.6] [Fig.6] illustrates the power spectral densities of signals measured for each of the four materials in [Fig.5];

[0047] [Fig.7] [Fig.7] represents a general view of a second system according to the invention, in use and seen from the side in a vertical plane;

[0048] [Fig.8] [Fig.8] represents the system of [Fig.7], seen from below;

[0049] [Fig.9] [Fig.9] illustrates a variant of the system of [Fig.7];

[0050] [Fig. 10] [Fig. 10] illustrates a diagram of a method for determining the nature, solid, liquid or soft, of a portion of material according to the invention and

[0051] [Fig. 11] [Fig. 11] illustrates a diagram of a method for characterizing a leak according to the invention. DETAILED DESCRIPTION OF THE INVENTION

[0052] First embodiment

[0053] A first embodiment of the present invention is described by means of Figures 1 and 3 and the associated passages below, which describe a system 100 for detecting and characterizing leaks.

[0054] [Fig. 1] illustrates an industrial device 50 seen from the side in a vertical plane, under which a peripheral element has been arranged, which in this example is a retention tank 110 designed to receive portions of material falling from the industrial device due to possible leaks. In this example, liquids falling in the form of drops are considered, for example at imperfectly sealed joints. [Fig. 2] represents a view from below of the retention tank 110. The retention tank is equipped with an instrumentation kit as described below.

[0055] The retention tank 110 is provided with vibration sensors 120a, 120d, 120e and 120f fixed to two side walls of the tank, and vibration sensors 120b and 120c fixed to the bottom of the bottom, under it in the present embodiment. The fixing of the sensors to the walls of the tank is preferably carried out in such a way as to obtain a good mechanical coupling between a sensor and the wall of the tank on which it is fixed, for example by gluing using a thin layer of glue, so that the vibrations transmitted by the walls of the tank are efficiently transmitted to the sensors. More generally, when contact sensors are used (by contact sensor we mean sensors which, by their measuring principle, need to be mechanically secured to the peripheral element), they can be fixed to the peripheral element, the retention tank in this example, by means of an adhesive (glue, wax) or a mechanical fixing (tightening, screwing, etc.) or magnetic.

[0056] The sensors may each be of the piezoelectric type, each based on one or more piezoelectric elements, possibly formed from a monocrystalline piezoelectric thin layer extending along an extension plane, less than 50 μm thick and with dimensions greater than 100 μm in its extension plane, as described in French patent application No. FR2303635. Such piezoelectric thin layers are advantageous in that their high sensitivity makes it possible to detect the impact of a portion of material, such as a drop of water, falling onto the bottom of the retention tank.

[0057] Each of the vibration sensors is operatively connected to a system 140 electronic computing system comprising a computing unit Cale, a memory Mem and a clock Clk connected to each other. The system 140 is connected to a human-machine interface Intf such as a control monitor.

[0058] When a leak A occurs, a certain quantity of material falls in the form of a drop from the point AOr of origin of the leak to a drop point APt located in the retention tank 110, at least approximately in line with the origin of the leak. Upon impact in the retention tank, vibrations AVib are generated at the drop point A Pt and then propagate in the walls of the retention tank. These vibrations will be converted by each of the sensors into detection signals sent to the calculation system 140, which will process these raw signals according to conventional methods to derive detection information therefrom, for example by comparison with the amplitude of the detection signals at a threshold in order to eliminate any possible measurement noise.

[0059] Thus, for each sensor considered independently, the reception of a detection signal by the calculation system can be associated with the occurrence of a leak: in response to the detection signal; the calculation system 140 can calculate a signal associated with the fall, and transmit the corresponding information to an operator by means of the human-machine interface Intf, such as a control monitor or a loudspeaker for emitting an alert signal in the event of detection of a leak.

[0060] At the drop point Apt, the amplitude of the vibrations AVib depends on the kinetic energy of the drop, i.e. its mass and its drop height. For a given situation, the volume of a drop can be defined by the nature of the liquid treated by the industrial device, this volume being conditioned by the density and viscosity of this liquid. Thus, the signal calculated by the calculation system can be representative of the kinetic energy of a drop having fallen from the industrial device.

[0061] The calculation device can also be configured to quantify the leak, by counting the number of detections of occurrence of a fall for a given sensor, or by evaluating the quantity of liquid having fallen on the basis of the amplitude of the vibrations detected. In the second case, it can be a quantity determined to within a factor, that of the height of fall, but which nevertheless provides information on the quantity of product having fallen. The quantity of liquid having fallen can be accumulated (by summation), each fall being associated with the point in time at which it occurred. The quantity of liquid having fallen can be related to a time interval. The two previous options make it possible to provide an indication of the evolution of the importance of the leak over time.In this way, the signal calculated by the calculation system can be representative of the occurrence of the leak, but also of a cumulative quantity of material having fallen from the industrial device, possibly during a given period of time.

[0062] As already stated above, the amplitude of the vibrations due to the impact of a drop depends on the nature of the liquid forming the drop and the drop height. The drop height can therefore be deduced from the amplitude of the vibrations due to the impact and the nature of the liquid, and can therefore provide information on the height of the origin of the leak. Depending on the complexity of the industrial device, this information can prove extremely valuable in locating and treating the leak. In practice, the computer 140 can be configured to calculate the leak height on the basis of a first signal generated by the sensors fitted to the retention tank and a second signal representative of the nature of the liquid treated by the industrial device.

[0063] It is also possible to locate the leak in the horizontal plane, i.e. the position of the drop point in the retention tank, in the most general case by taking advantage of the multiplicity of vibration sensors.

[0064] The applications described above can be applied to a system comprising only a single sensor. The multiplicity of sensors makes it possible to trace the location in the horizontal plane of the drop points in general cases.

[0065] The high sensitivity of the sensors makes them capable of detecting vibrations of low amplitude, and therefore makes it possible to reduce their number to cover a retention tank of given dimensions.

[0066] The retention tank may be provided with an anti-vibration system 130 intended to isolate it from the vibrations of its environment. The reason for this isolation is that vibrations generated by the operation of the industrial device 50 could disturb the measurements carried out by means of the vibration sensors. The anti-vibration system, here in the form of anti-vibration pads, makes it possible to limit the measurement noise and therefore to make leak detection and location more reliable according to the principle detailed below.

[0067] While it is obviously simpler to apply the present invention to a retention tank of simple and regular shape, the retention tank can be of arbitrary shape and dimensions.

[0068] In addition to the retention tank taken as an example in this embodiment, all or part of an element peripheral to the industrial device and capable of receiving the material coming from the point of origin of the leak can be used. This can be, for example, an element taking the form of a flat plate, which can be formed to cover all or part of the footprint of the industrial device.

[0069] The nature of the material used to form the peripheral element and the dimensions of the latter are preferably chosen so that the impact of a portion of material on its surface generates a vibration capable of being detected after propagation on the surface of this element.

[0070] It is thus possible to choose a peripheral element formed for example from steel, Teflon, glass, ceramic or other material adapted to the environment and the nature of the material falling from the industrial device: a material that is chemically and / or mechanically resistant to material falling from the industrial device will preferably be chosen, depending on whether it is a corrosive chemical product or a material in the form of blocks whose impact when dropped may be sufficient to damage the peripheral element if it is not properly chosen.

[0071] When the peripheral element takes the form of a plate (this may be, for example, the bottom of a retention tank), its thickness dimensions are preferably between one and several millimeters thick, up to one centimeter thick, or even three or five centimeters thick; its lateral dimensions depend on the surface to be covered, but may, for example, extend from 10 cm to 1 m, 2 m, 5 m, or 10 m, or even more. There is no maximum lateral dimension to the peripheral element, the limit being the possibility for the sensors to detect a leak, considering that it is possible to use sufficiently sensitive sensors or to multiply the sensors so as to achieve sufficient sensitivity at any point of the peripheral element, or at least at any point considered relevant by the user.However, it is preferable to avoid using a peripheral element whose nature and / or dimensions promote the rapid damping of vibrations generated by the impact of portions of falling material from a leak to be detected or monitored.

[0072] The vibration sensor(s) used may be of different types, such as resistive gauges, piezoelectric gauges, surface microphones, acoustic emission sensors, accelerometers or even laser vibrometers (which do not need to be fixed in contact with the peripheral element), without any particular limitation, except that the sensitivity or dynamic range of the type of sensor chosen must be adapted to the measurements to be carried out. Indeed, the more the portions of material falling due to a leak have a high kinetic energy due to their weight and their drop height, the less critical the sensitivity of the sensors used will be. Also, the dynamic range of the sensors will preferably be wide when portions of material having very different weights are likely to fall, depending on the nature of the industrial device monitored and the type of leak likely to occur.

[0073] Location of a fall in the horizontal plane

[0074] [Fig. 1] illustrates a leak A and a leak B, with distinct drop points APt and Br, and origins AOr and BOr, respectively. The drop point Ar is closer to the sensor 120b than to the sensor 120c, and vice versa for the drop point Ar. The vibrations attenuate as they move away from the impact points Ar and Br where they were generated. It is therefore understood that the amplitude of the vibrations AVib of the leak A as detected by the sensor 120a will be greater than those detected by the sensor 120b. Conversely, it is understood that the amplitude of the vibrations BVib of the leak B such as detected by sensor 120b will be greater than those detected by sensor 120a.

[0075] We see that the information on the amplitudes of the vibrations detected by the different sensors, coupled with the respective locations of these sensors, already gives a first indication on the location of the leaks. The closer the point of fall of a leak is to a sensor, the greater the amplitude of the vibrations detected by this sensor will be, and if, among the different sensors, the amplitude is maximum for this sensor, then this means that it is to this sensor that the point of fall is closest, at least as a first approximation.

[0076] In order to generate more precise location of drop points, or to optimize the number of vibration sensors required for instrumenting a peripheral element, it is possible to go further in analyzing the signals generated by these sensors. Several options are possible. We will briefly detail three methods for locating drop points by exploiting the sensor signals generated by the vibration sensors: triangulation of arrival times, modal analysis, and learning a neural network. Other methods are possible. But above all, we will show the reproducible nature of the vibrations generated by the impacts of portions of material falling onto the peripheral element.

[0077] Reproducibility

[0078] [Fig. 3] illustrates experimental results carried out with a retention tank used as a peripheral element, a controlled drip system and three vibration sensors S1, S2 and S3, the sensitive element of which is a piezoelectric thin film as already described.

[0079] The experiment consisted of dropping six drops: three times at two distinct positions called Pos. 1 and Pos. 2. For each position, each drop is considered as an event, the three events being identified as Ev. 1, Ev. 2 and Ev. 3. [Fig. 3] shows six graphs arranged in two columns and three rows, each graph comprising three curves corresponding respectively to the signals generated by the three vibration sensors S1, S2 and S3. Each curve represents the amplitude of a signal over time: the electric charge in pC generated by a piezoelectric sensor in response to the vibrations caused by the fall of a drop into the retention tank, indicated on the y-axis. The time is indicated in milliseconds on the x-axis. The graphs in the left and right columns correspond respectively to measurements made at position Pos. 1 and to measurements made at position Pos.2, the graphs in the top, middle and bottom row correspond respectively to measurements made during events Ev. 1, Ev. 2 and Ev. 3. .

[0080] We note that, for each location, the profiles and amplitudes are repeated almost identically from one fall to another. Conversely, these profiles are very different from one position to another. Thus, it is clear that signal analysis methods such as those described below will make it possible to deduce the positions of the falls from the signals generated by the vibration sensors.

[0081] Discrimination of the state, whether solid, liquid or soft, of the portion of falling matter

[0082] An impact between a portion of falling material and the peripheral element generates a vibratory wave propagating from the point of fall, which will be detected by one or more sensors as explained above. This vibratory wave is composed of a spectrum of frequencies forming a vibratory signature representative of the interaction between the material having fallen and the surface of the peripheral element. In particular, a first vibratory signature of an interaction between the peripheral element and liquid or soft material and a second vibratory signature of an interaction between the peripheral element and solid material will be distinct, and will make it possible to determine the state, liquid, soft or solid, of the material having generated a detected wave. By material in a soft state, or soft material, is meant material in which the interactions between the elements constituting it are weak.Soft matter is highly susceptible to external mechanical stress and deforms easily. Examples of soft matter include foams, gels, elastomers, liquid crystals, cosmetic creams, mud, pastes, and certain agri-food products.

[0083] [Fig.5] illustrates graphs similar to those of [Fig.3], with curves of measurements obtained experimentally using three sensors SI, S2 and S3, for quantities of matter formed respectively from four different materials and having fallen at the same point. In [Fig.5], each curve is associated with the corresponding sensor. Each curve represents the amplitude of a signal over time: the voltage in volts appearing at the terminals of a piezoelectric sensor in response to the vibrations caused by the fall of a drop into the retention tank, indicated on the y-axis. The time is indicated in milliseconds on the x-axis.

[0084] The curves indicated by (A), (B), (C) and (D) were obtained by dropping a bolt, a screw, a drop of water and a silicone pad, respectively.

[0085] It is noted at first glance that the curves obtained for dropped material made of hard materials (bolts and screws, which are made of metals) are very similar. Similarly, the curves obtained for dropped material made of soft materials (liquid water, silicone pad) are also very similar. This observation is confirmed by [Fig.6].

[0086] [Fig.6] illustrates the power spectral densities (PSD for Power Spectral Density in English terminology) of signals measured for each of the four materials in [Fig.5], the bolt, the screw, the water and the silicone pad, indicated respectively by Boit, Screw, Water and Si Pad. The graph in [Fig.6] expresses the power spectral density in mV / ^Hz on the y-axis according to a logarithmic scale as a function of frequencies expressed in Hz on the x-axis according to a logarithmic scale.

[0087] The overall level of PSDs is representative of the energy dissipated during impacts, increasing with the mass and speed of the quantity of material that has fallen. For soft or liquid objects, such as the water drop and the silicone pad, a rapid drop in PSD is observed below 3.108 mV / ^Hz at 104 Hz, at the high frequencies, from about 102 Hz. Conversely, for hard objects, the bolt and the screw, the PSD level remains high, above 106 mV / ^Hz even at 104 Hz.

[0088] It is understood from these observations that measurements of the impacts of material on the peripheral element by means of vibration sensors make it possible to characterize the state, whether solid, liquid or soft, of the material.

[0089] For this purpose, the method 400 for determining the nature of the falling material, illustrated by [Fig. 10], can be used. This method comprises the following steps 410 to 450.

[0090] In a step 410, the vibration generated by an impact of a portion of material having fallen onto a peripheral element is measured, over a predefined duration around the impact, for example one second, by means of vibration sensors which generate sensor signals.

[0091] In a step 420, the average PSD PSDiow of the vibration measured below a given frequency F and the average PSD PSDhigh above this frequency are calculated, over a range of frequencies considered, from 10 Hz to 104 Hz in the example of [Fig.6]. The given frequency F depends on the characteristics (material, shape, dimensions) of the material having fallen and of the peripheral element. In the example of [Fig.6], F = 500 Hz may be chosen, because above this frequency, the differences in the signals generated by the material having fallen are more marked than below this frequency, which facilitates discrimination of the nature of the material.

[0092] At a step 430, a ratio R between PSDiow and PSDhigh is calculated, such that R=PSDiow / PSDhigh, indicative of the distribution of the vibrational energy for a given impact between the low frequencies and the high frequencies.

[0093] In a step 440, the calculated ratio R is compared to a threshold value Thr, defined as a function of the materials likely to fall onto the peripheral element. A ratio R greater than Thr is indicative of a vibration generated by liquid or soft matter, favoring the generation of vibrations of energy relatively more concentrated in the low frequencies. Conversely, a ratio R less than Thr is indicative of a vibration generated by hard matter, favoring the generation of vibrations relatively more concentrated energy in the high frequencies.

[0094] In a step 450, an SS / L signal representative of the state, either solid, liquid or soft, of the material is generated, in response to the comparison of step 440.

[0095] An advantage of the method described above is that the absolute values ​​of the measurements have no influence on the ratio R, this method will therefore work for more or less strong signals, generated by falls of more or less heavy portions of material or from more or less height.

[0096] All of the above steps can be carried out automatically without human intervention by the measuring and calculating means of the leak detection and characterization system 100. The ratio R, the frequency F and the threshold value Thr may have been defined in advance according to the type of material likely to fall from the industrial device to be monitored, and stored in the memory Mem.

[0097] Such a method can be implemented to detect the nature of a material that has fallen onto the device, in particular when several sources of leakage are possible, which may concern either solid materials or liquid materials. This is for example the case during the transport of silicon wafers, which are likely to break, while being treated with liquid cleaning and / or treatment solutions such as hydrofluoric acid. The method 400 makes it possible to detect whether a detected impact is due to the breaking and falling of a solid silicon wafer or to a liquid leak in the hydrofluoric acid circuit.

[0098] Triangulation of arrival times

[0099] A first method of location of the drop points is that of the triangulation of the arrival times, described for example by Tobias for the case of impacts on a plate (A. Tobia, “Acoustic-emission source location in two dimensions by an array of three sensors, Non-Destructive Testing 9(1):9-12, February 1976).

[0100] The principle is to use three sensors of known relative positions. If we consider the speed of propagation of the vibrations in the plate as known, then we can link the differences in arrival times between the different sensors to distances on the plate, and go back to the position of the point of impact using well-known mathematical calculation methods that are inexpensive in terms of calculations.

[0101] This method, however, has several limitations. A first limitation is that bending waves in plates are dispersive waves, which implies that the high-frequency components will propagate faster than the lower-frequency components. Precisely determining an arrival time is therefore not always easy. The shape of the wave packet generated by the impact changes with the propagation distance; the attenuation of the waves in the structure due to losses in the material will also modify the shape of the wave packet. However, improvements have been proposed to compensate for the dispersion of the waves during propagation and improve accuracy (PD Wilcox, "A rapid signal processing technique to remove the effect of dispersion from guided wave signals", IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 50, no. 4, pp. 419-427, April 2003). A second limitation is that this method is based on an assumption of homogeneity of the structure: it is assumed that the waves propagate in the same way over the entire structure, which is assimilated to an infinite plate. In practice, the structures studied are often far from this ideal, and the method can lose a lot of accuracy. It will be advisable for the practitioner to check that the geometry of the peripheral element he uses and the positioning of the sensors are compatible with the triangulation method and the degree of accuracy it requires.

[0102] Modal analysis

[0103] Modal analysis can provide an alternative to the triangulation method when the latter is not appropriate for a given situation, for example in the case of a structure of the peripheral element that is too complex or poorly adapted, for example due to its shape. In addition, the accuracy of calculating arrival times is often better when using the high-frequency components of the vibration signal. Unfortunately, the attenuation of the waves during propagation in the structure is also higher at high frequency, which limits the distance between the impact and the sensors for which the method works correctly.

[0104] To overcome these shortcomings, an approach using the low-frequency components of the signal was developed by (D. Goutaudier et al., "Impact localization on a composite fuselage with a sparse network of accelerometers", Volume 348, issue 3 (2020), p. 191-209), which uses the projection of the impact force onto the modal base of the structure, truncated to keep only the first eigenmodes. The general idea is that the different modes of the structure are excited in different proportions depending on the location of the impact.

[0105] Learning - Neural Network

[0106] The principle of learning methods is to make a neural network learn the forms of signals associated with vibrations generated by impacts at different positions. Following this learning, the neural network is capable of matching a signal or a set of signals with the location of the impact that generated this signal or this set of signals.

[0107] The main prerequisite for this type of method is to have a learning database, which contains the signals measured by the sensors during impact at many known locations. The more exhaustive this database is, the greater the learning and robustness of the models developed will be. There are essentially two approaches to building this learning database.

[0108] The first approach is experimental: the sensors are installed and the structure is then impacted at many known locations. This approach can be time-consuming, and the type of impact possible during training is limited, as they must not damage the structure. An example of this type of approach is explained by Worden and Staszewski (K. Worden and W.J. Staszewski, “Impact Location and Quantification on a Composite Panel using Neural Networks and a Genetic Algorithm,” in Strain, Vol. 36 Issue 2, pages 61-68, May 2000).

[0109] The second approach is numerical: a numerical model of the structure is developed, for example by finite elements, and used to simulate various impacts at many locations. This base of simulated events is then used for learning. If the numerical model is sufficiently faithful, then the performance of the model that has learned on the numerical base is good on the real structure. An example of this type of approach is explained by Sahra Sharif Khodaei (Z.Khodaei et al., “Determination of impact location on composite stiffened panels”, in Smart Materials and Structures, IOP Publishing Ltd, Vol. 21, No. 10).

[0110] [Fig.4] compares the results of the signals from three sensors SI, S2 and S3 of vibrations generated in response to an impact at a given location for an experimental situation in (A) and for a computer-simulated situation in (B). These figures show a very good match between the signals measured by the sensors and the simulated signals. This shows that a digital database can be built by simulation for the purpose of learning the neural network, and that it will not be necessary to build this learning base experimentally.

[0111] Second embodiment

[0112] A second embodiment of the invention is described by means of Figures 7 and 8 and the associated passages below, which describe a variant of the first embodiment. The industrial device and the retention tank are different from those of the first embodiment. The other elements constituting the leak detection and characterization system 100 are, on the other hand, identical, except that the number of vibration sensors has been reduced for this second embodiment.

[0113] Here, as illustrated in side view along a vertical plane by [Fig.7], the industrial device consists of a set of pipes 50', under which has been placed a retention tank 110' designed to receive portions of material falling from the industrial device due to possible leaks.

[0114] The specificity of the present embodiment is that the industrial device is quasi-one-dimensional, in the sense that the piping extends essentially along one dimension, defined by the longitudinal extension of the piping, along an axis Absc. [Fig.8] thus illustrates the retention tank seen from below, significantly longer than it is wide, with for example a length L at least five times greater than its width 1.

[0115] In such a case, using three sensors is not mandatory to locate the drop point Cr of a leak C with sufficient precision for an operator to be able to easily determine the origin COr of the leak. Indeed, a simple longitudinal location along the pipe is sufficient, and the two sensors 120b and 120c together allow this information to be obtained: the arrival times associated with the distances dl20b and dl20c separating the drop point CR from the sensors 120b and 120c, respectively, are sufficient to determine an abscissa x(Cr) of the drop point CR along the axis Absc.

[0116] Third embodiment

[0117] [Fig.9] illustrates a third embodiment, herein described as based on the second embodiment and its geometry a simplified version of the second embodiment.

[0118] In this embodiment, the quasi-one-dimensional geometry of the retention tank 110' is coupled with the location of a sensor 120ext located at one of the ends of the retention tank. If we take for example the hypothesis of leaks causing the fall of drops of liquid of similar volumes and of a pipe of constant height, a simple calibration of the sensor makes it possible to associate a detected vibration amplitude with a distance d^oext of the point Dr of fall of material due to a leak D with respect to the sensor di20ext, and therefore to locate the point of fall by an abscissa x(Dft) equal to the distance di20ext, the origin 0 of the abscissa axis being set to the position of the sensor 120ext.

[0119] Fourth embodiment

[0120] [Fig. 11] is a diagram illustrating a method 500 according to the invention, implemented by means of the system 100 described in the preceding embodiments, having the function of detecting a leak of a portion of material falling from an industrial device.

[0121] In a step 510 of the method, the retention tank 110 or 110' is provided as a peripheral element, directly above the industrial device 50 or 50', the possible leaks of which are to be detected, the retention tank being equipped with at least a first vibration sensor (120a, 120b, 120c, 120d, or 120ext), and possibly a second, a third and / or other vibration sensors fixed to the retention tank, the vibration sensors being functionally connected to the electronic calculation system 140.

[0122] Step 510 may comprise a step 5140A of instrumenting the peripheral element 110 or 110' with at least one first vibration sensor when this peripheral element is already arranged directly above the industrial device. Alternatively, step 510 may comprise a step 510B of instrumenting the peripheral element with the at least one first vibration sensor, followed by a step 510C of arranging the element instrumented device directly above the industrial device.

[0123] A fall of material from the industrial device, a fall caused by a leak, causes an impact lmp at the bottom of the retention tank, this impact in turn causing a vibration propagating in the walls of the tank.

[0124] At a step 520 of the method, in response to the fall of the portion of material at a drop point (Ar, Br, Cr, Dr) located in the retention tank, the first vibration sensor and the possible second and third sensors respectively generate a first, a second and a third sensor signal Scapti, Scapt2, Scapt3 and transmit these sensor signals to the electronic calculation system 140, thus carrying out a measurement of the vibration generated by the impact.

[0125] At a step 530 of the method, in response to the transmission of the sensor signal(s), the electronic calculation system generates and transmits calculated signals Scaici, S caic2 and Scaic3 representative of the fall of the portion of material at the drop point located in the retention tank as perceived by the first, and, where appropriate, the second and third sensors, respectively. The calculations may represent the implementation of known methods for locating impact points as mentioned in the first embodiment.These calculations may also have the function of generating signals each representative of one or more of the following information: an occurrence of the fall of the portion of material, a kinetic energy of the portion of material, a cumulative quantity of material having fallen from the industrial device during a given period of time, a distance between a location of the first vibration sensor and the point of fall of the portion of material in the retention tank.

[0126] When one or other of the calculated signals Scaici, Scaic2 and Scaic3 is calculated on the basis of the corresponding sensor signal and a signal SMat representative of a nature of the portion of material, the calculated signal can be at least representative of a drop height of the portion of material, as explained in the section concerning the first embodiment. The signal SMat can be entered manually by an operator, or be generated by the computer itself on the basis of information stored in the computer memory Mem and representative of the type of material which is processed by the industrial device and therefore likely to fall into the retention tank.

[0127] In a step 540, on the basis of one or more of the calculated signals Scaici, Scaic2 and S caic3, that is to say from one or more of the corresponding sensor signals S capti, Scapt2 and Scapt3, the electronic calculation system 140 calculates and emits a position signal Spos representative of a position of the drop point, by means, for example, of one or other of the three methods (triangulation, modal analysis, learning) set out above.

[0128] In parallel to step 530, based on one or more of the sensor signals, the method 400 can be implemented, with step 520 replacing step 410.

[0129] Any combination of embodiments one through four is acceptable and falls within the scope of the invention described herein.

[0130] In this description, we have taken the example of drops of material in liquid form, but the system and the method described by means of these examples also apply to drops of material in solid form, possibly in calibrated form when it is a question of calculating the drop height by analogy with the calculation applied to a drop already described.

[0131] The invention is not limited to the embodiments described above and variations may be made thereto without departing from the scope of the invention as defined by the claims.

Claims

Claims

1. Method (400) for detecting a leak (A, B, C, D) of material from an industrial device (50, 50'), comprising the steps of: - measuring (520; 410) a vibration (Avib, Bvib, Cvib) of a peripheral element (110, 110') arranged to receive a portion of material falling from the industrial device (50, 50'); and, - in response to this measurement, and by means of an electronic calculation system (140), generating (530) a calculated signal (Scaici), representative of the fall of the portion of material falling at a drop point (Ar, Br, Cr, Dr) located on the peripheral element (110, 110').

2. Detection method according to claim 1, in which the vibration is measured at at least two distinct points of the peripheral element (110, 110'), and, in response to these measurements, the electronic calculation system (140) generates (540) and emits a position signal (Spos) representative of a position of the drop point.

3. Method according to claim 1 or 2, further comprising: - a step (510A) of instrumenting the peripheral element (110, 110') with at least one first sensor when this peripheral element is already arranged vertically above the industrial device (50, 50'), or - a step (510B) of instrumenting the peripheral element with the at least one first sensor, followed by a step (510C) of arranging the instrumented peripheral element vertically above the industrial device.

4. The method according to any one of claims 1 to 3, wherein the calculated signal is representative of at least one piece of information selected from an occurrence of the fall of the portion of the falling material, a kinetic energy of the portion of the falling material, a cumulative quantity of material having fallen from the industrial device during a given period of time, a distance between a location of the first sensor and the point of fall of the portion of the falling material on the peripheral element (110, 110'), a state, either solid, liquid or soft, of the portion of falling material.

5. The method according to any one of claims 1 to 4, in which the calculated signal is calculated on the basis of a detection signal (Scapti) of the vibration and a signal (SMat) representative of a nature of the portion of material, the calculated signal being at least representative of a fall height of the portion of material.

6. The method of any one of claims 1 to 5, wherein the step of detecting the vibration (Avib, Bvib, Cvib) of the peripheral element (110, 110') is implemented by means of a first vibration sensor comprising a monocrystalline piezoelectric thin layer extending along an extension plane, less than 50 pm thick and with dimensions greater than 100 pm in its extension plane, this thin layer being rigidly secured to the peripheral element.

7. The method of any one of claims 1 to 6, wherein the peripheral element is a retention tank.

8. Instrumentation kit configured to equip a peripheral element (110, 110') intended to be installed directly above an industrial device so as to form a system (100) for detecting a leak (A, B, C, D) of material falling from the industrial device (50, 50'), the kit comprising: - at least one vibration sensor (120a, 120b, 120c, 120d, 120ext) provided to generate and emit a sensor signal (Scapti) in response to a vibration (AVib, BVib, CVib) of the peripheral element resulting from an interaction between a portion of the falling material and the peripheral element (130); and - an electronic calculation system (140) which can be functionally connected to the first sensor and is provided to, in response to the first sensor signal (Scapti), generate and emit a calculated signal (Scaici) representative of the fall of the portion of material at a drop point (Apt, Bpt, Cpt, DpJ located on the peripheral element.

9. The instrumentation kit according to claim 8, wherein the first calculated signal is representative of at least one piece of information selected from an occurrence of the fall of the portion of the falling material, a kinetic energy of the portion of the falling material, a cumulative quantity of material having fallen from the industrial device during a given period of time, a distance between a location of the first sensor and the point of fall of the portion of the falling material on the peripheral element (110, 110'), a state, either solid, liquid or soft, of the portion of falling material.

10. The instrumentation kit according to claim 8 or 9, wherein the first calculated signal is calculated on the basis of the first sensor signal and a signal (SMat) representative of a nature of the portion of material, the first calculated signal being at least representative of a fall height of the portion of material.

11. The instrumentation kit according to any one of claims 8 to 10, comprising a plurality of sensors (120a, 120b, 120c, 120d) of vibrations each provided to emit an associated sensor signal (Scapti, Scapt2, Scapts) in response to the vibration of the peripheral element, and operatively connectable to the electronic computing system, wherein the electronic system is further configured to generate and emit a position signal (Spos) representative of a position of the drop point (Apt, Bpt, CPt, D^), based on the sensor signals (S capt b Scapt2, Scapt2).

12. The instrumentation kit according to any one of claims 8 to 11, wherein the at least one sensor comprises a monocrystalline piezoelectric thin layer extending along an extension plane, less than 50 pm thick and with dimensions greater than 100 pm in its extension plane.

13. System (100) for detecting a leak in an industrial device (50, 50') comprising a peripheral element (110, 110') intended to be installed directly above an industrial device and equipped with the instrumentation kit according to any one of claims 8 to 12.

14. The detection system of claim 13, further comprising a suspension device (130) configured to isolate the peripheral element from mechanical vibrations of its environment.

15. The detection system of claim 13 or 14, wherein the peripheral element is a retention tank.

Images