DE2937709A1 - Material crack-testing system by sound - compares computed distance between source and fourth detector with measured value - Google Patents

Abstract

System tests materials for cracks etc. , by sound emission, partic. pressure vessels of nuclear reactors. The emission signals and/or the arrival times at the converters are measured, the signals being processed in a computer having a sound emitter. Four or more detectors are used, the distance between the source calculated by the first three and the fourth detector being computed, and the result compared with the value measured by the fourth detector. The detectors can be arranged also to measure signal parameters such as length, energy, max. amplitude and ascent time, etc.

Description

Method and device for

Acoustic emission test on materials The invention relates to a Method for acoustic emission testing of materials According to the preamble of the claim 1 and a device for carrying out the method.

1. Problem and objective of the noise emission test (SEP) should be used in the monitoring of pressure samples from reactor pressure vessels and eventual. to the Operational monitoring can be used.

The phenomena of sheet emissions when cracks occur and Analyzes crack propagation and correlations between acoustic emission parameters and Development of cracks or cracks as a basis for the interpretation.

2. Requirements for an acoustic emission test system for triangulation and evaluation 2.1 Event rate To estimate the maximum event rate at a Pressure test that has to be processed by an SE test system is assumed that there is a crack in a reactor pressure vessel, which is critical at the test pressure becomes (stress intensity factor = Kc). The maximum expected Event rate during such a print attempt is based on the observed Relationship between event dose and the volume of the plastically deformed zone before the crack tip / report / estimated.

For a critical flaw in the RPV it was added the size of the plastically deformed zone in front of the crack tip and its change over time during a pressure test estimated. This "volume rate" then results from the above-mentioned relationship the event rate to be determined between the event dose and the plastically deformed volume.

The estimation was carried out for 2 different types of errors: For a circular planar crack in the center of the container wall with the diameter 2a (defect type 1) as well as for one running along the axis of the container Crack (infinite length, depth = a) on the inside of the container wall (defect type 2).

To get from the formulas for the size of the plastically deformed volume Being able to determine the change per unit of time is a function of the stress intensity factor K from the time. To this end, two borderline cases are considered: 2.1.1 Type of load: load controlled The error is relatively large, then the pressure test corresponds to a load-controlled tensile test, i.e. as long as no crack growth occurs, K increases linearly with time, the law of time resulting from the conditions of the printing attempt.

It is here with a pressure increase rate of 1 bar / min.

expected.

2.1.2 Type of load: path-controlled If the error is relatively small to Container wall, the stress on the crack is due to the expansion of the surrounding material specified, i.e. the load corresponds to a displacement-controlled tensile test. The increase the stress intensity factor over time is then slower than with one load-controlled attempt.

The resulting time laws are only used for estimation of extreme cases. In detail, they depend on the deformation behavior of the material. The deformation processes in the event of a critical crack in the container are corresponding run a law between these two extreme cases.

If one sets the time laws for the stress intensity factor, the calculated critical crack sizes and for the yield strength the value for 22 NiMoCr 37 a, the volumes of the plastic zone result as a function of the test time the pressure test. This also gives the increase in volume per unit of time.

For the laboratory experiments carried out, the result is 50 mm CT samples from 22 NiMoCr 37 have a mean "volume rate Vp = 0.29 mm3 / s. The event rate achieves values of up to 500 events / s. This results in the one you are looking for Relationship between the event rate and the volume rate (with a threshold of 3 bar): n = 1730. Vp mm3 This allows the expected event rate for a Calculate pressure test; the values are entered in table 1.

This consideration only applies to small signals, i.e. if the Source is close to the probe because the signals still exceed the specified threshold have to. As the worst case scenario, however, the consideration is more decisive Meaning. It can be seen that in the worst case, an event rate of 1500 events per second must be processed when you have a circular flaw inside the container wall assumes. It has not yet been taken into account that the SE is a stochastic Process, in fact a measuring system would have to have higher event rates for a short time can process.

TABLE 1 Biblis A HDR 17.5 11 test pressure p (N mm-2) 2,500 1,535 Container radius R (mm) 242 112 Wall thickness b (mm) 4,000 4,000 KIC (N mm -3/2) 82 58 Crack size (mm) Defect type 1 65 46.5 Crack size (mm) Defect type 2 520 370 n (s-1) Error type 1 path-controlled 140 100 n (s-1) Error type 2 path-controlled 1,320 1,540 n (s-1) Error type 1 load-controlled 350 410 n (s-1) Error type 2 load-controlled 2.2 Signal superimposition Since in an array with an edge length of 6 m, runtimes of 2 ms without time Superimpositions, i.e. only signal repetition frequencies of a maximum of 500 Hz if they are present several Signal sources can occur and there continue to be event rates of more than 1000 Hz are expected, processing of temporally superimposed Signals be possible.

2.3 Interference level The system must be as sensitive as possible can be driven to the signals even after long walks with appropriate To be able to detect the weakening. The resulting, relatively high level of interference must not significantly affect the measurements.

2.4 Detection of modes Especially when measuring thin-walled Objects, e.g.

Pipelines, different wave modes are excited.

The location should be adjustable with only one mode with a fixed one Speed can be carried out. For thick-walled objects the possibility may have to be used exist, when calculating the location, a speed of propagation that depends on the route apply.

2.5 Signal analysis The system must carry out an evaluation with the aid of the signal analysis and by recording the frequency distributions of sound emission parameters allow. From today's perspective, the SE parameters are the energy and the amplitude to capture an event. The dynamic at the input must be at least 50 dB.

2.6 Correction of the path of the SE parameters It must be for the Noise emission parameters a path correction can be carried out in order to adjust the parameter e.g. To be able to determine energy at the source.

3. Overview of location systems for sale today There are currently eight device manufacturers that offer larger location systems. The systems offered however, can be traced back to only three fundamentally different modes of operation: 3.1.

Measurement of the transit time differences from an event (manufacturer: AET Corporation, Intercontrole, Compagnie General de Radiologie, Trodyne Corporation, Dunegan / Enderco Acoustic Emission Consultants Ltd.) The systems work normally with 3 or 4 sensors per array. A SE signal is detected at the moment where the amplitude of a signal has a predetermined fixed threshold or one of the interference level dependent variable threshold exceeds. For the purpose of a clear signal definition it is necessary in this case that a certain Rest has been observed.

The moment the first signal arrives, two become simultaneously or three clocks (with three or four probes) started showing the time differences up to determine the arrival of an acoustic signal at the other transducers. In one Computers are used to calculate the location coordinates of the source from the transit time differences (This problem is different for different manufacturers solved). Further signal analysis, e.g.

the measurement of the energy is possible with some systems.

3.2 Measurement of the transit time differences between pairs of sensors (Exxon Nuclear, Southwest Research Institute) The transit time differences between the arrival of sound signals measured at two sensors and for 2 or 3 pairs, the frequency distribution of this delay time difference is determined on line. For the transit time differences that occur relatively frequently, a Computer's possible source locations. The so-called energy release rate determined. The outputs of 5-10 sensors each are used for this purpose summarized on an RMS voltmeter.

Further signal analysis is not possible.

3.3 Coincidence detection location method (Southwest Research Institute) The location is done with the help of 3 sensors. In contrast to the two before The methods mentioned are used in the CDL system (Coincidence detection location system) the signals are read in without dead time. That is to say, it is possible that that too is temporal Overlapping signals are detected.

Any combination of arrival times at the 3 transducers, within an SE source location is assigned to the maximum running time for an acoustic signal. One further signal analysis is not possible. With this system you will already be reached significantly higher speeds than with the two previously mentioned. However, in the event of a high sequence of events, mislocation is possible, i.e. it will every correct source location is displayed, but non-existent Sources detected.

4. The object of the invention is to reduce the number of events with a higher sequence of events To enable mislocation. According to the invention, this is achieved by the method according to Claim 1 and with a device for carrying out the method according to claim 4 solved. Advantageous method steps are characterized in the subclaims.

4.1 Conception of the ESEP system The from the above estimate of the maximum Event rate resulting values are not mi serially operating Systems to dissipate; because for an ereiynisserte (Ies system result for a probe arrangement with a maximum distance of 6 m, depending on the time of the elaboration and the event definition, maximum permissible values of 300 Hz (for details see section 4.5). Under real Conditions, i.e. for signals that do not occur periodically, should be the maximum Speed are only around 100 Hz.

Since the processing of temporally superimposed events continues should be possible, only the method of an event-parallel remains for development System left.

4.1. 1 How the system works The locating part of the test system works according to the real-time method. That means that all detectable, so Signals lying above a certain threshold in the location logic without dead time can be processed. The system does not switch as with all previously implemented and the systems described in Section 3 (except CDL) for the duration of the term of a signal to neighboring sensors and beyond for a certain waiting time away. With ESEP, the localization happens without dead time, at the moment when a sentence of there are three time differences for the location. A location is considered more correct at the moment Place accepted by all 4 arrival times with a prior in a bok-up-table stored patterns match. Each place corresponds to a certain pattern.

(In the following, the location in the plane with 4 probes is always considered, of course, there are correspondingly simpler systems for locating on a Straights possible). The analog part of the system consists of four transducers and four Preamplifiers as well as four main amplifiers per array. After amplification and filtering the signal passes through a signal conditioner to define the event (decisive for all positioning systems). The trigger signals will wind on arrival time detector passed on. This arrival time detector works as a buffer with the help of shift registers and is used to determine the source location according to predetermined Sample requested (see below for more detailed information). After this query, the Result in the form of a data word delayed for the purpose of synchronization with the signal parameter (e.g. energy). After the subsequent recalculation the energy at the source is followed by a storage of the calculated value. This Memory can be used, for example, for display on a screen via digital-to-analog converters can be read out.

4.1.2 Functional description of the location logic The location logic is further developed from that of the CDL system (see Fig. 1). There is a shift register for each transducer assigned in which the signals supplied by the signal conditioner are registered Events can be saved and pushed forward at a fixed rate. The maximal Storage duration is equal to the number of shift register locations, multiplied by the cycle time and corresponds to the maximum runtime of a signal between the am farthest apart sensors in a field. The shift registers are filled with ones and zeros, with a one indicating that within this a certain time cycle when an event has started. A zero says ous, within There was no signal start for this time cycle. Every possible source location within of the probe array corresponds to a specific assignment of shift register positions (pattern), which results from the shape of the arrays and the number of shift register locations.

These patterns and the associated location are stored in a look-up table. The location logic now queries three shift register locations for their content. (Not all register space combinations correspond to a meaningful place; at this So the first eliminations take place of interference signals). If the system is used as it has been described so far, this corresponds to the essentially the aforementioned CDL system. The weakness of such a system is the large number of mislocations. It is possible that at the time of arrival of a signal in the other shift registers contain more than one trigger signal is.

The logic would then usually determine more than one place of origin. The influence of such mislocations is reduced by statistical averaging, however, it still contributes to the deterioration of the signal-to-noise ratio. It It is therefore important to reduce such mislocations as much as possible. Such Incorrect location can be reduced by adding additional sensors reach.

The more additional transducers are introduced into the system, the more It is more certain that the display does not show any mislocation. The more additional However, the more complex the system becomes. For this Basically, we will use four sensors for the real-time noise emission test system limit, i.e. three sensors for the actual location and one transducer for Control. The functioning of the system is thus as follows: The one through the first arrival / arrival times given by three sensors clearly determine the location or a location field with a certain fuzziness, which is defined by the cycle time. Of this Position field, if the coordinates of the fourth oscillator are known, the distance Calculate between the source and the fourth transducer and therefore with a known speed of sound propagation the transit time and thus also the arrival time at the fourth transducer are also included a certain temporal fuzziness, which is, however, greater than that in the other shift registers. Used in the interval between maximum and minimum Runtime if an event is found in the fourth shift register, the previously with The location detected by three arrival times is accepted as correct, if not, it is discarded and thus the number of mislocations is limited.

The limits of the system are based on the fact that Uncertainty for each location occupies several register positions in the fourth shift register could be. So if the event rate is relatively high, there are a lot of register locations occupied in the fourth shift register. The fourth shift register can no longer contribute to a restriction of the mislocation, since the allowed assignments cover. With this system it is theoretically possible to frequency signal fol to process up to the clock frequency.

All emitting sources are detected. The higher the repetition rate is, however, the greater the probability that additional mislocations appear. In contrast to this, an event-serial system becomes too high if the event rate is too high do not carry out any incorrect location, but also no "correct" location.

4.2 Working out the mathematics for localization only on any Kuielfläche and in level 4. 2. 1 ttlunderFeldform The shapes of the location fields result as sections of two hyperbolic areas (see Fig. 2). The hyperbolic areas result from the cycle time of the shift register. The width of the hyperbolic areas on the connecting line between two sensors is therefore the cycle time multiplied with the speed of sound. Let us now construct the field shape on a concrete one Example based on shift registers with 2 places. All three channels are on the same place is filled with a one, i.e. comes at all three transducers the event at the same time. Fig. 3 shows the result of the cut from two time difference intervals.

The result is a square, with each side of the square being the area is a hyperbola. Aiis three arrival times can, however win not just two time differences but three. With the help of this third time difference this square can now be reduced to a hexagon.

The field shape is slightly dependent on the arrival times, they however, it is independent of the length of the shift register.

The result is hexagons, which are about 2/6 of the area cover with every neighboring hexagon (see Fig.

4). The reason for the overlapping of the hexagons is the width (temporal Blurring) of the shift register. For example, if a storage space corresponds to a Microsecond, then follows as the possible distance between two signals in neighboring register locations a time interval from t that is greater than zero and less than 2 microseconds (see Fig. 5). Field overlaps are possible because the trigger time the signals and the shift clock are not correlated.

4.2.2 Calculation of the locations from two arrival time differences The derivation of the locating equations follows in the main the considerations of Tobias (1) and Asty (2).

The longest distance between sensors SO, S1 and S2 (Fig. 6) corresponds to the length of the shift registers, which are otherwise all the same length to have. The cycle time T is defined according to the relationship V. T (N-1) = L where N the number of shift register positions, V the speed of sound and L the length between the sensors So, S1 and S2.

When calculating the source coordinates from the time of arrival differences ambiguities arise in the vicinity of the calibration probes.

From these ambiguities must now with the help of plausibility considerations a solution to be eliminated. This is done by placing the vibrators on the most distant solution declared wrong, given signals from such a source be located in a neighboring array.

All ambiguities lie in the seams of the transducer probes. One Illustration of the ambiguity problem is shown in Fig. 7. As a rule, the one to be excluded Solution much further away from the transducers than the second.

4.2.3 Discussion of the necessary lengths of the various shift registers If one assumes an equilateral pressure and at the same time assumes that the fourth Transducer is not further away from at least two other transducers than the three transducers in this array one below the other, it results as required The length of the fourth shift register is a maximum of three times the length of the normal 3 others Shift register. The extreme cases are discussed using Fig. 8.

Case 1: The four transducers form a hexagonal arrangement, the source is at transducer 1. The path to transducer 2 and transducer 3 is therefore an edge length. The way to transducer 4 is longer than that to transducer 3 and transducer 2, namely by less than twice the distance to oscillator 2. One clocks now the contents of the shift register, the event is first sent to channel 1 at the output at. Then after exactly one normal shift register length, i.e. edge length of the array, the event reaches the output of channel 2 and channel 3. After about twice the length of the shift register, the signal reaches the output of channel 4. Since all signals for location are at the same time in the shift registers need to be stored, we need a corresponding tracking register for channel 4 one edge length (easy option).

Case 2: There is again a hexagonal arrangement of the probes, The source is now in the middle of the connection line between probe 2 and probe 3. The event is thus first detected in probe 2 and probe 3 and then comes to probe 1 after about twice the running time up to probe 2 and probe 4 at the same time. In this case, the length calculated above is the shift register sufficient for all 4 probes.

Case 3: Three probes form an equilateral triangle, the fourth The probe lies outside the triangle or on the line connecting the second and the third probe.

The source is directly on the fourth probe. In this case that comes Event first arrives at transducer 4, then reaches channels 2 and 3 and arrives last on channel 1. The result shows that channel 4 is already ahead of the others 3 channels must record events, so must have a certain lead time. This can be achieved by connecting a shift register of length A upstream. The same result is obtained for a hexagonal arrangement of the four probes, if the source is outside the area formed by the three main probes, but closest to probe 4.

4.2.4 Encouraging the allowed places in the fourth shift register As in As shown in Fig. 3, the locating fields result as hexagons, their boundary lines are given by six hyperbolas. As Fig. 4 shows, however, every hexagon of covers six other hexagons so that each corner of a hexagon is the center another hexagon is. Because the edges of the hexagons are only slightly can be curved outwards and one expediently on a sensor arrangement where the fourth transducer is not positioned too far from the array is, the distance circles (see Fig. 9) therefore have a greater curvature than the six edges, the query of the most extreme points of a field can be made on the query reduce to the most distant (or closest) corner point. These two values are rounded up (down) and then indicate the limits of the fourth register allowed places.

4.3.5 Computer simulation of the ESEP system The ESEP system becomes the Purpose of preliminary examinations, according to the functional description given under point 3 simulated on the computer.

Both the CDL system and the event-based system leave program in a similar way to the ESEP program on the computer. The CDL system can be derived directly from the ESEP system, simply by: that the check for coincidence is not carried out with the help of the fourth channel. There is then only a simple location with the help of three sensors. To the simulation the conventional event-series system (let's call it KSEP system from here on), it is ensured that the present detected events really are belong to a source by accepting only those events before them Detection has ruled a certain "transmission pause". So a waiting period must be introduced will.

5. Comparison of the three positioning systems ESEP, CDL and KSEP 5.1 Comparison of the systems with the help of computer-simulated To make a first comparison To enable the three systems, a program was written that makes it possible to periodic sound emission signals (without reflection and sound propagation phenomena) to simulate. The program generates needle pulses at any selectable location, so that only one shift register position per channel is occupied by such a signal will. All three systems react to different arrival times, i.e. different ones Locations, different. So there were three for the investigations, the whole variety the possible location coordinates as well as possible overlapping location coordinates as source coordinates chosen.

With fixed source coordinates in each case, the time interval of the Signals, i.e. the signal repetition frequency varies and the respective reaction of the three Systems recorded.

The results are shown in Figs. 10, 11 and 12. Fig.

10 shows the best case for the event-based system. There the results all relate to an equilateral triangle with a central probe, is to achieve this result in the case where the event is exactly in the middle of the array, i.e. the central probe. You can use for this borderline case the event-serial system with an array with an edge length of 5 m a maximum frequency of about 540 Hz. However, this result is purely hypothetical The result, firstly, because the dead time in this case is as low as possible, as low as possible as in reality it was never set to be selectable, because, secondly, the data processing times have been neglected and, thirdly, since periodic signals have been used, so that no disturbances could influence the result. So it works up to about 540 Hz the KSEP system is correct, then it is no longer able to carry out localizations.

Both the CDL system and the ESEP system lead up to the highest possible Frequencies, all locations through.

The real place is displayed correctly every time. However, surrendered for higher frequencies for both the CDL system and the ESEP system Ads in places where no sources are available. For As already mentioned, the KSEP system works up to 540 Hz at a location in the middle of the array, Both CDL and ESEP systems bring all locations correctly, and no mislocations up to 600 Hz. So there is a slight improvement. Then be Although all true locations are correctly displayed by the two event-parallel systems, however, there are additional mislocations. Set an arbitrary limit so that at least 20 X of the displayed locations must be correct, so one can say the CDL system works correctly up to around 750 Hz, whereas the ESEP system works up to 1200 Hz, for a given location in the center of the array.

Fig. 11 shows the other extreme case, the source is now very much close to one of the sensors. The calculations apply again to the equilateral Array with an edge length of 5 m. The conventional system in this case is able to locate up to about 290 Hz, then it does not bring any more locations. The CDL system brings all localizations without errors up to about 380 Hz, the ESEP system up to 600 Hz.

If the frequencies are given again at 20% correct locations, so This results in a maximum frequency of 820 Hz for the CDL system and for the ESEP system of 1240 Hz.

Fig. 15 is representative of the average of the locations.

So here those locations are recorded that are between the center and Corner sensor lie. The event-based system works up to about 340 Hz, then it does not bring any more localizations. The CDL system works perfectly without mislocation up to about 380 Hz, the ESEP system on the other hand up to 810 Hz returns the frequencies at 20 X correct locations, this results for the CDL system a cut-off frequency of 800 Hz for the ESEP system a cut-off frequency of approx. 1050 Hz.

This periodic data shows a clear one improvement the locating efficiency compared to the conventional event-series system. To With event-parallel systems, all of them become towards the highest frequencies emitting sources recorded. Additionally displayed, nonexistent sources always only occur above the frequency limit of the event-series system. Around to name a comparative figure, which of course only refers to a mean estimate one can say that the CDL system is minimal in terms of correct positioning better than the conventional event-serial system. For the ESEP system, however results in the mean, i.e. averaged over all possible locations, roughly double permitted frequency. If one now includes in the consideration that the event-parallel Systems up to the highest frequencies correctly capture and accept all true sources you can find the location at the moment when 20% correct locations are displayed, the frequency for the CDL system is roughly doubled compared to the event-series system and for the ESEP system roughly tripling the frequency. In any case, ESEP can be used to detect an acceptable number of mislocations Reach frequencies of over 1000 Hz. The system therefore fulfills those mentioned in point 2 required specifications of a new system to be created.

KSEP CDL ESEP array center 540 750 1200 Surrounding the corner probes 290 820 1240 range between probes 340 800 1050 Table 1: processed frequency (Hz) in a 6m array with a system quality of 20 Z For those that occur in reality Signals, including sound propagation, statistical character and additional Interference signals, So irregular event intervals will be the result for the parallel events Systems can still be improved, as the mislocations are always in different places occur and can thus be averaged out. The ads will be the true sources add up and thus stand out more clearly. In contrast, will worsen the result for the conventional event-series system, since the required dead time due to the influence of interfering signals got to.

Claims (4)

Claims 0 method for acoustic emission testing on materials, with the acoustic emission signals and / or the arrival times at the transducers several transducers can be measured without a crime time and the signals in a computer a sound emission source location is assigned, d a d u r c h indicates that a) at least four sensors are provided, b) with known coordinates of the fourth Transducer is the distance between the source location calculated by the first three transducers and the fourth transducer is determined by calculation, c) this is determined by calculation is compared with the value measured by the four-transducer. 2. The method according to claim 1 d a d u r c h characterized in that in addition the signal parameters, such as signal length, energy, maximum amplitude, rise time, measured by the transducers. 3. The method according to claim 1 and claim 2 d a d u r c h characterized, that the energy of the measured signal is calculated back to the source location. 4. Device for carrying out the method according to claim 1 - 3, d a d u r c h that four transducers and associated amplifiers and converters are provided, as well as three shift registers of equal length for the first three transducers and for the fourth pick-up a shift register with three times the length for intermediate storage the measured signals are provided and a computer for further processing.