RU2124242C1 - Method for vibration-noise diagnostics of pressurized-water reactors - Google Patents

Abstract

FIELD: in-service diagnostics of pressurized-water reactors such as those of VVER type. SUBSTANCE: method involves joined analysis of signal fluctuations in pressure pulsation sensors, relative and absolute displacement transducers, core external ionizing chambers, core internal forward charge sensors, and thermocouples. Desired signals of low- and high-frequency ranges are chosen at program level for setting signal frequency range, and multichannel recording is carried out. Recorded sequence of computing procedures is executed on the basis of pre-selected diagnostic characteristics for each reactor effect. EFFECT: provision for automatic diagnostics. 3 cl, 2 dwg

Description

 The proposed method relates to methods for noise diagnostics of reactors with water under pressure, for example, VVER type, is implemented by the system of vibration noise diagnostics (SVShD) and can be used to determine the technical condition of the equipment elements of the reactor installation (RU) during its operation.

 Known methods of vibro-noise diagnostics implemented by "SUS" systems (manufactured by Siemens, see Germany standards DIN 25475, part 2 and KTA 3204) and KARD (manufactured by Hungary, see "KAZMER" A COMPLEX NOISE DIAGNOSTIC SYSTEM FOR 1000 MWE PWR WWER TYPE NUCLEAR POWER UNITS G.Por, LA Sokolov "Paper presented at IAEA Technical Committee Meeting on" Utilization of Condition Monitoring and Degradation Diagnostic System to Improve Nuclear Safety "Vienna, 7-9 October 1991).

The standard set of sensors for the vibration diagnostics system developed by Siemens, supplied to Russian WWER-440 reactors, consists of
absolute displacement sensors (DAP) - 4 pcs.,
relative displacement sensors (DOP) - 24 pcs.,
pressure pulsation sensors (DPD) - 4 pcs.

 The same company supplies a similar system to VVER-1000 reactors, and another type of sensors is added to the sensors mentioned above: direct charge (DP) sensors.

The disadvantages of the method implemented by the system "SUS" are:
1) high requirements for the qualifications of the operating personnel in terms of setting the initial data for spectral estimation,
2) the need to attract experienced experts for the physical interpretation of the auto and mutual spectral characteristics of the recorded noise,
3) a standard set of spectral estimates, not communicated to specific diagnoses, unmodifiable when delivered to a particular switchgear,
4) the need to fix in advance, manually, a certain set of resonances of spectral functions (that is, the user of the system requires knowledge of the physical interpretation of the diagnostic value of each of the resonances selected by him),
5) a predetermined number of mutual characteristics (not more than 20 pcs.) To be evaluated and archived in a database (DB) despite the fact that the total number of possible pairs of signals for producing mutual spectral characteristics is more than a thousand,
6) The Fourier transform of noise implementations immediately after digitization already at the entrance to the vibrodiagnostic system, which excludes the possibility of using other analysis methods, for example, multivariate autoregressive analysis (MAP analysis),
7) archiving spectral estimates without archiving the source processes of sufficient length,
8) the impossibility of introducing other noise, non-vibrodiagnostic algorithms in SUS without changing its software.

The standard set of sensors of the KARD system, supplied to the Russian VVER-1000 type reactor, includes DPZ, IK, DPD and thermocouples (TP). The disadvantages described in paragraphs 1) - 4) apply to the KARD system and, in addition, there are the following own disadvantages:
the absence of vibration sensors, which drastically narrows the possibilities of vibration diagnostics,
non-adaptation of diagnostic algorithms to VVER-type reactors.

The features of the proposed method are
automatic diagnosis by formalizing expert knowledge of each diagnosed situation in the form of a scenario - a fixed sequence of computational procedures adapted to a particular switchgear that works with a pre-selected set of signals, frequency ranges and parameters for evaluating spectral characteristics, as a result of which the script is not required highly qualified operating personnel,
the possibility of implementing at the program level any algorithm of vibration and noise diagnostics due to the developed software tools in the form of a set of elementary operations, which, in particular, makes it possible to solve in one system both the problems of vibration diagnostics and the problems of neutron-thermal hydraulic noise monitoring of the state of the active zone,
supply to RU of a start script library, which is expanded by new scenarios during the operation of the system without reprogramming it,
identification of anomalies at an early stage, when the manual method for isolating diagnostic signs is ineffective due to masking by a large number of other spectral effects.

 In FIG. 1 shows a General implementation diagram of the proposed method, which includes sensors, analog and digital signal processing devices, computers and software. The system uses six types of sensors: DPZ (1), IR (2), DPD (3), TP (4), DAP (5), DOP (6). In addition to the vibration monitoring functions, the system, using the first four types of sensors, also performs neutron-thermal control of the core.

In order to minimize the number of analog paths based on signals from the DPS and TP, the following switching is provided:
the DPZ switch (7) out of 64 fuel assemblies (FAs), each of which has 7 DPZs, selects any 14 DPZs or, as an option, 7 DPZs of one FA and 7 DPZs of another FA. That is, for analogue protection, 14 analog paths are needed;
TP switch (8) out of 64 TPs installed at the outputs of each fuel assembly selects any 4 TPs.

 Using a switch (9), immediately before recording the signals, all electronic paths are calibrated by a standard signal generator (10), the output of which is either “white” noise in the studied frequency bands of known intensity, or a sinusoid of a given amplitude and frequency. In calibration mode, the SVDS software evaluates the transfer functions for all measurement channels at once. Thus, their metrological verification and periodic routine maintenance are carried out.

 In the analog paths of the DPZ, IR and TP, there are galvanically isolated pre-amplification blocks (11), on which the signal is divided into constant (=) and fluctuating components (≈). The extracted constant components are used further for absolute measurements of the above signals and the subsequent normalization of the fluctuating components to a dimensionless form.

 The generation of signals from vibration sensors (DAP and DOP) is performed in block 11a in a known manner (see "DP-100 relative displacement sensor. Technical conditions" and "DAP-08 absolute vertical displacement sensors. Technical conditions." DAP08.00.000.TU).

 The frequency range in which vibro-noise diagnostics of reactors with water under pressure is performed is covered by the frequency interval (0.01 - 100) Hz. Low-frequency (12a) and high-frequency (12b) band-pass filters provide the operation of the HFSM in two frequency ranges, three decades each: either in the low-frequency (LF) range (0.01 - 1 + e) Hz, or in the high-frequency (HF) range [(1 -e) - 100] Hz. That is, instead of one process in a wide frequency range, two processes in two frequency sub-ranges are analyzed. In order to further “stitch” the spectral characteristics along the frequency axis into a single range [0.01 - 100] Hz, overlapping of subbands with a value of "2e" was introduced. The value of ε is easy to choose, given that the fast Fourier transform (FFT) is performed by the number of points equal to a certain power of two (in practice, this is usually the number 1024, 2048, 4096, 8192). In order for the resolution of the low-frequency spectral analysis (frequency step) to be a "round" number, it suffices to choose ε = 0.024 Hz.

 When digitizing processes from the two frequency sub-bands indicated above, an analog-to-digital converter (ADC) with a resolution of no more than 14 is sufficient. The low-frequency range will be digitized with a frequency of 2 (1 + ε) Hz, and the implementation length should be at least two hours. The high-frequency range will be digitized with a frequency of 200 Hz, and the implementation length will be about 5 minutes. Thus, the decrease in the amount of information to be analyzed and stored without loss of diagnostic value is several orders of magnitude compared with the direct solution to the problem of digitizing the range [0.01 - 100] Hz with a 16-bit ADC.

 Switching "LF-HF" (13) is carried out immediately for all channels VHF, that is, VHF cannot work part of its channels in the LF range, and the other part of the channels in the HF range.

 After filtering, amplification (14) of the fluctuating part of the signal is amplified to the value of the standard normalized signal. Under the control of the computer1 (25), through the digital-to-analog converter (15) and the feedback line (24) in block 14, the gain is automatically selected.

 The last operation with analog signals is their switching (16) according to a pre-compiled list of vibro-noise diagnostic scenarios. As a result of this switching, no more than 32 signals are selected from the entire series of analog signals. Thus, the next digital part of the SVShD is 32-channel.

The above commutation:
switching of signals DPZ, TP at the lower level,
"HF-LF" - switching,
switching normalized analog signals in accordance with the scenario
switching the standard signal generator for calibration
implemented at the program level.

 Since the SHDS operates only with stationary random processes and should not operate in transient reactor modes, the system provides for blocking (17) of the DPZ, IR, TP, and DPD channels when the state of the core controls changes.

 The thirty-two-channel digital part (18) of the SVShD is constructed according to the traditional scheme: multiplexer, 14-bit ADC, computer data input device1.

The software for the SVShD computer consists of four parts:
the first performs service functions (setting paths in the mode of automatic selection of gain factors, production of software switching (system configuration), diagnostics of individual nodes of the internal hard disk drive, calibration),
the second performs registration in the database (19) of multi-channel implementations in accordance with the selected sequence of scenarios,
the third - the most high-tech - script library software system (20), which operates in deferred mode with multi-channel implementations of signals from the database,
the fourth is real-time software (21), for the implementation of which ADCs (22), a fast Fourier transform processor (23), and computers (26) are provided.

 When performing spectral analysis, the main computing resources are spent on performing FFT. Therefore, until recently, this operation was transferred to the hardware level (hardware level) and performed on the so-called Fourier processors (FFT processors). Thus, the program part began with the level of Fourier coefficients, from which various auto and mutual spectral characteristics were obtained. But a significant drawback of such a construction of the SHDS is the inability to implement MAP analysis and various decomposition procedures, for which the processes themselves, and not their Fourier images, are necessary.

 If we take into account that the bulk of the diagnostic information in the SHDS is retrieved by delayed analysis, and modern computers are able to perform FFT at the software level quickly enough, then for the delayed analysis it is advisable to abandon the FFT at the hardware level, and leave it only for real-time functions. In real time, the monitoring of vibrations of the most important from the point of view of ensuring safe operation of switchgear assemblies, such as the core mine (A3) and the main elements of the main circulation circuit (HCC) (housing, main circulation pumps (MCP), steam generators (GH)), should be carried out in real time. .

 Scenario - a macro operation compiled by a group of experts to identify diagnostic signs for a specific reactor effect. Scenario - macrooperation uncontrolled from the side of NPP operation. The number of scenarios is not limited and can be continuously replenished. A significant difference of this method is that it is not the user who, according to his understanding, fixes the resonances, but the system itself (SHSH) selects them. Moreover, not one set of resonances is produced, but several, each of which belongs to a specific reactor effect. This approach is implemented by the following sequence of actions.

 1. Experts in reactor noise, having studied the implementation of random processes on this SSHD sensor configuration of a given reactor, draw up algorithms for distinguishing various reactor effects.

 2. These algorithms are implemented by experts on spectral estimates in the form of scenarios - a sequence of actions on a set of recorded processes from SFS sensors.

 3. Scenarios, being part of the SHDS, are applied every time to each new multi-channel recording of processes obtained as a result of periodic vibration monitoring of the switchgear.

 Thus, the operating personnel of the SHDS do not need to think about the parameters for evaluating the computational procedures, or about fixing pairs of signals, or about highlighting some set of resonances. All this intellectual work has already been done once by experts on the evaluation of spectral characteristics. Run a script - pressing a single computer key. All estimation parameters, no matter how complex computational procedures are, are fixed for a given switchgear, for a given SHDS. As a result of such periodic inspections, many resonances will accumulate for each reactor effect. The a priori classification of sets of resonances by reactor effects is the intellectual work that has already been done by reactor noise experts. The advantage of automatic resonance isolation is that anomalies are detected at an early stage, when they are not yet visible "by eye".

An example implementation of the method for a reactor of type VVER-1000
The number of sensors and the number of SHDS channels coincide only in the ideal case. With such a construction, it is only necessary to provide a multi-channel recording of all processes and any existing diagnostic scenario, as well as any scenario that appears in the future, will be easily implemented. That is, hardware (hardware) will not limit the development of software (software). However, this is a very wasteful decision. For example, to monitor the vibrations of a fuel assembly, any DPZ signal may be required, and they are 64 • 7 = 448 in the VVER-1000 core (A3). At the same time, specific, practically implemented diagnostic scenarios operate with no more than 30 signals. Thus, the expert must be given a choice of several hundred signals so that he captures no more than thirty signals when constructing a specific scenario. In the proposed method, this is achieved by developed means of switching signals, which are mentioned above.

Modern SHDS should form diagnostic signs by the following effects:
vibrations of the shaft of the core and the reactor vessel,
oscillations of the main circulation circuit and oscillations of the main equipment (SG, MCP),
MCP vibrations at the reverse frequency, its harmonics and subharmonics,
collective vibrations of fuel assemblies,
vibrations of internal enclosures (ICU) arising due to acoustic standing waves of heat carrier pressure,
collisions of VKU,
coolant boiling,
and also make assessments:
temperature and barometric reactivity coefficients,
local and global axial non-uniformity of the core energy release field,
values of per-channel coolant flow rates,
time constants of sensors, efficiencies of remote sensing and infrared.

The totality of the listed computational procedures (script library) is implemented by the following set of sensors:
at least 2 additional (radial and tangential) for each MCP and each steam generator, i.e. 16 additional for VVER-1000,
not less than 4 DPD,
all DPZ,
eight IR
four DAPs on the cover of the reactor vessel,
four thermocouples at the outputs of the fuel assembly.

 In order to carry out a full cycle of diagnostics over the entire list of the library of scenarios, it is necessary to make several consecutive multichannel records of reactor noise in time, each time switching signals. In one such record, more than one scenario may be implemented.

 As an example in FIG. Figure 2 presents a diagram of a monitoring scenario for collective vibrations of fuel assemblies. It consists of two parts. In the first, the frequencies of the collective vibrations of the fuel assemblies are detected, in the second, the autospectral power density (ASF) is estimated for seven RPZs of the same assembly and the frequencies found in the first part are substituted.

 Among the large number of signs by which this type of oscillation is distinguished (see Bulavin V.V., Gutsev D.F., Pavelko V.I. "Studies on VVER-1000 Vibrodiagnostics under Operating Conditions", Atomic energy, issue 5, November, 1995), the most reliable, having a general reactor character, were selected. Among the many signals in which this type of oscillation is displayed (all IR, all DPS, all DAP, all DPS), only four of the most sensitive ones were selected.

 Collective vibrations of fuel assemblies represent a powerful phenomenon of internal-body vibrations that “swing” not only the shaft A3, but also the reactor vessel itself. Therefore, they will necessarily appear in a pair of signals DPZ-DAP. The phase relationships here are unpredictable; therefore, no phase selection is performed. The following pairs of signals are based on phase characteristics. For two distant from each other (diametrically spaced along A3) DPS (DPS1 - DPS3), located in different neutron field gradients, antiphase should be observed. Exactly the same symptom works for a pair of DPZ-remote IR. Obviously, for two SCRs located in one channel (DPZ1 - DPZ2) there should be phase matching in the first mode of collective vibrations of the fuel assembly.

At the first step, traditional spectral characteristics are evaluated: all kinds of AFM (A), coherence (G), and phase (φ _xy (f)). Further, the frequencies of all resonances (R) are found for all coherence functions. The frequencies found for all these pairs of signals intersect, as a result of which only the frequencies of the collective vibrations of the fuel assembly - Ω are extracted. The amplitude distributions corresponding to the vibrations of the fuel assemblies can be distorted as the efforts of the spring block or deformations of the fuel assemblies weaken as fuel burns out. For this purpose, seven AFPM signals of the DPZ of one assembly are further evaluated. The number of such assemblies should be at least three: one, two, and three years of operation, and for full monitoring of the vibrations of the fuel assemblies, all fuel assemblies in which the DPS are installed should be investigated.

Claims (3)

 1. The method of vibro-noise diagnostics of reactors with water under pressure, which consists in a joint analysis of fluctuations in the signals of sensors of pressure pulsations, relative and absolute displacement, extra-ionization chambers, characterized in that they further analyze the signals of intraband sensors of direct charge and thermocouples, at the software level, make the selection to make a specific diagnosis of signals of the low-frequency or high-frequency range, produce multi-channel recording and record a given sequence of computational procedures based on previously identified diagnostic features for each reactor effect, which ensures automatic diagnosis.  2. The method according to claim 1, characterized in that a start library of fixed sequences of computational procedures is supplied to the reactor installation, which is expanded by new scenarios during operation without reprogramming it.  3. The method according to claim 1, characterized in that the starting library of sequences of computational procedures generates diagnostic signs for the following effects: vibration of the shaft of the active zone and reactor vessel, vibration of the main circulation circuit and oscillations of the main equipment, vibration of the main circulation pumps at the reverse frequency, its harmonics and subharmonics, collective vibrations of fuel assemblies, vibrations of internals arising from acoustic standing waves of coolant pressure, SOU arenes internals, coolant boiling, and produces estimates of the temperature and barometric reactivity coefficients, the local and global non-uniformity of the axial field energy of the core, variables per channel coolant flow rates, time constants and efficiencies of sensors.

Images