This page has been accessed
900 times since 23-Apr-96 CronCount
The main field of activity of the Department has always been related to the development of data acquisition, monitoring and operator support systems to be applied at various nuclear facilities.
The most important results are briefly overviewed below.
Parallel to the commissioning of the four VVER-440/V213 type units at the Hungarian NPP at Paks the Applied Reactor Physics Department delivered a well furnished core monitoring system to support the operators in controlling the reactors. The first system-unit entered into operation in 1985 at the reactor unit No.1, the last in 1987 at unit No. 4. The system obtained the standard measured core data and certain other measured quantities via the Hindukus data acquisition system.
Services of the system satisfied virtually all operational core monitoring needs arising at the time. Display monitors were situated in the main control room and in the main computer room of all units. No connection among the units, nor with any other site (e.g. management) was available. Display screens with numerical, coloured and graphical information served the operator with all data relevant to the status of the core.
The system has been operational for more than eight years and
was gradually replaced by an upgraded system VERONA-u to be described below.
The MR materials testing reactor has been operational in the Kurchatov Institute, Moscow from 1965 to 1992. A very complex data acquisition and information system of the reactor has been designed, developed and implemented at the reactor by the Applied Reactor Physics Department. The core resident database handling and the graphical user interface was developed by the Computer and Automation Research Institute, Budapest. The system included five primary data acquisition computers (IDACS's) of TPA 11/70 type, three VAX 780 type mainframes and 12 IBM PC compatible graphics workstations.
1000 analogue inputs were received and processed in every second, 1056 discrete inputs were registered with a resolution of 0.1 sec. About 300 technological schemes were available on the screens of the graphics workstations. Dedicated workstations served the operators of the various experimental loops. An on-line, fault-tree based disturbance analysis module provided early detection of any failure in the system.
The highly redundant system automatically reconfigured itself in case of any failure in any of its main hardware components.
The system was removed from operation before its full scope
utilisation, parallel to the shutdown of the test reactor.
In the framework of a large scale project a noise diagnostics system has been developed for the Kalinin 1000 MW VVER NPP (Russia). The system consists of three parts: the KARD neutron noise system (developed by the AEKI Applied Reactor Physics Department), the ARGUS vibration diagnostics system (Electric Power Research Institute, Budapest) and the ALMOS acoustic emission leakage monitoring system (AEKI High Reliability Devices Lab.). The project was lead and co-ordinated by the Applied Reactor Physics Department.
The KARD system obtains measured data from 30 incore self powered neutron detectors, 6 excore ionisation chambers, 6 pressure transducers and 6 accelerometers.
The system is meant for the detection of core barrel motion, monitoring of control rod vibration, estimation of various parameters such as coolant velocity, reactivity coefficients, or thermocouple temperature response time.
The KARD system runs on an IBM compatible personal computer, its functioning is menu-driven and the results are displayed via colour monitor screens.
The KARD system is also operational at the units No. 3 and 4 of
the Paks NPP. An advanced version of the system, supplemented
with signal validation capabilities is under development (c.f.
JEDI system).
Because of the physical and moral ageing of the core monitoring hardware and software tools, an upgrading process has been initiated in 1991. In the framework of this process the primary data acquisition system Hindukus has been replaced by a set of VME-based redundant systems called PDA (provided by Comproject Ltd, Budapest). Five PDA crates, each with a high performance Motorola processor, collect the measured data, the most important ones reaching two crates in parallel. All standard incore measurements along with those from the primary and secondary loops that are important from core monitoring purposes are available for the system. More than 700 analogue and 360 discrete signals are processed once in every 2 seconds. A two way data connection with the plant process computer ensures the transfer of certain data not available in the other system.
Two MicroVAX 3100 computers host the VERONA-u core monitoring software, both in network connection with the five PDAs. (The communication software was developed by Akribia Ltd, Budapest.) The two hosts are full substitute to each other, should the active host break down for any reason the other (hot standby) computer takes over the entire functioning of the system. As much as eight graphical workstations can be served by a system at a reactor unit (two in the control room, two in the computer centre and four remote stations anywhere in the plant's network). The four local stations are dedicated to the given unit, the remote stations can reach any of the unitwise systems.
Two remote VERONA systems (VERONA-r and VERONA-t) serve independent developing, experimental and planning purposes at the Reactor Physics and the Technological Computer Departments, resp. of the NPP. These remote systems can obtain on-line data from any unit, or can run over recorded data in the standard "archives replay" operational mode of VERONA-u. The plantwise VERONA configuration thus includes 10 MicroVAX computers, 20 workstations and 20 primary data acquisition PDAs, all connected via a local area network. An independent full capacity system, the VERONA-s is running in co-operation with the full scope training simulator of the plant.
Advanced programming, database handling and user interface tools have been exploited during the development of the system. A great part of the core analysis calculations has been developed in the NPP Reactor Physics Department.
The uniquely versatile services of the system include the display of the assemblywise distribution of every important reactor parameter (e.g. outlet temperature, heatup, 2D and 3D flux and power, peaking factors, burnup, etc.) in the form of (numerical as well as various colour coded) core maps , axial distribution of the flux,
power or linear heat rate along selected assemblies, values of the most important parameters, control rod positions , SPND readings, event lists trends of selected parameters, various logs, hardcopies etc. A transient planning and strategy verification procedure is also available in the system.
Various recorded data sets (archives) make a posteriory data processing and event recovery easy and also offer the possibility of archives replay parallel to the on-line functioning of the system.
The system is functional uninterrupted at all four units of the Paks NPP so
far having
served about seven reactor years with virtually 100%
availability. Besides its classical
operator assisting function it proved
to be a very effective tool of the NPP safety and
control management.
The G2 expert system shell (product of GENSYM Corp. ) has been used to build up a prototype intelligent process monitoring and alarm generation system (GPCS - G2-based Plant Computer Subsystem).
The GPCS includes an object oriented description of the main subsystems of the plant and concentrates on the fast evaluation and display of the measurements (e.g. Feedwater Subsystem or Turbine Subsystem) and the alarms . The high-level information, reflecting the actual plant safety status is synthesised from primary measured data by forming global alarms and by evaluating logical diagrams.
For the functional testing of the prototype the plant
measurements, processed by the system are provided by a compact
simulator of the Hungarian Paks NPP, also developed in the
Atomic Energy Research Institute.
Based on the long term experience in developing noise diagnostics systems and starting from the successful KARD system a new system, integrating noise diagnostics and signal validation capabilities has been developed with the financial support of the National Committee for Technological Development (OMFB ha van hozz link).
The system is able to receive technological data from the standard core instrumentation, performs discretisation, stores the data on disk and performs processing and analysis. Up to 32 data channels can be processed simultaneously (with an option of increasing the channel number to 64). Processing is menu driven and includes Fast Fourier Transform, univariate as well as multivariate AutoRegressive and Moving Average AutoRegressive analyses, Kalman Filtering, Sequential Probability Ratio Testing, Parity Space representation analysis.
The noise diagnostics analysis conforms with the options in the KARD system, i.e. it includes monitoring of reactor core barrel motion, coolant velocity estimation, control rod vibration detection and analysis of effects on the basis of characteristic maxima in the spectra of the measured quantities (temperature, pressure, neutron flux, mechanical vibration).
The system runs on an IBM compatible high performance PC under MS Windows, with modules written in Borland C++. The usual Windows menu system directs the user in a self-explanatory manner. The displayed schemes represent parts of the technological system in a hierarchically ordered and encapsulated manner. Selection of the sensors to be processed as well as of the methods to be applied is highly flexible and comfortable. The results of the analyses appear automatically in both graphical and textual forms. Long term storage and trending of the results are also options.
Noise diagnostics analysis has lead to the identification of
abnormal processes in the core several times in the past. The
huge potential offered by the fact that all incore
measurements (about 550 signals at every unit) are available
through the diagnostics output channels of the incore data
acquisition system PDA shall in all probability enhance the
role of the noise diagnostics and signal validation methods in
reactor core surveillance.
Several methods applied in the noise diagnostics and signal validation systems have first been founded and validated theoretically. In the sequel a brief overview of the most important developments is given.
A new, theoretically sound formulation of the ARMA equations and their solutions has been derived in [1, 2]. The Kalman Filter technique has been refined and applied to specific estimation problems [3, 4]. A new, generalised form and a single-sided application of the Sequential Probability Ratio Test have been developed and applied to diagnostics problems [5, 7]. The Parity Space method has been generalised to provide more information than in its original form, and has been coupled with the SPRT [8].
A signal validation and early failure detection method - to be applied to the large amount of core measurements - has been developed based on the statistical coherence and correlation among the simultaneously measured quantities [9].
Thorough theoretical and numerical investigations have been
conducted to establish a stable method for the estimation of
the neutron flux from the prompt component of the self powered
neutron flux detector current [10].
| [1] | Tran Dinh Tri, A New Algorithm for Recursive Estimation of ARMA Parameters in Reactor Noise Analysis, Ann. Nucl. Energy, 19, 287 - 301 (1992). |
| [2] | Tran Dinh Tri, Generalisation of the Signal Transmission Path Method for ARMA Model in Reactor Noise Analysis, Ann. Nucl. Energy, 19, 341 - 345 (1992). |
| [3] | A. Rácz, On the Estimation of a Small Reactivity Change in Critical Reactors by Kalman Filtering Technique, Ann. Nucl. Energy, 19, 527 - 538 (1992). |
| [4] | A. Rácz, On the "Simultaneous Estimation of Neutron Density and Reactivity in a Nuclear Reactor Using a bank of Kalman Filters", Nucl. Sci. Eng., 113, 93 - 95, (1993). |
| [5] | A. Rácz, Detection of Small Leakage by a Combination of Dedicated Kalman Filters and an Extended Version of the Binary Sequential Probability Ratio Test, Nucl. Technology, 104, 128 - 146 (1993). |
| [6] | A. Rácz and I. Lux, A One-Sided Sequential Test, Ann. Nucl. Energy, 23, 997 - 1010 (1996). |
| [7] | A. Rácz, Comments on the Sequential Probability Ratio Testing Methods, Ann. Nucl. Energy, 23, 919 - 934 (1996). |
| [8] | A. Rácz, An Improved Single Sensor Parity Space Algorithm for Sequential Probability Ratio Test, Ann. Nucl. Energy, 22, 747 - 761 (1995). |
| [9] | F. Adorján, T. Morita, Correlation-Based Signal Validation Method for Fixed In-Core Detectors, submitted to Nucl. Technology. |
| [10] | K. Kulacsy and I. Lux, A Method for Prompt Calculation of Neutron Flux from Measured SPND Currents, Ann. Nucl. Energy, to appear. |