CORE MONITORING EXPERIENCE WITH GARDEL

Transcription

1 CORE MONITORING EXPERIENCE WITH GARDEL Axel Becker, Alejandro Noël Studsvik Scandpower GmbH Studsvik Scandpower Suisse GmbH Abstract The GARDEL core surveillance and analysis system is a standard, modular system developed for use at the most common commercial reactor types and it is completely independent from the computer platform. This paper presents an overview of the functionality and the operating experience of GARDEL at several PWR and BWR plants in Sweden, in the States, in Switzerland and in Germany. The experience so far has demonstrated The accuracy of GARDEL s underlaying methods. The power of GARDEL s analysis tools. The flexibility and modularity of the system providing the customer with complete freedom to select the operating system as well as the network configuration. GARDEL's flexibility to configure the system according to the customers hardware and network environment predestines the system for replacements of existing core monitoring systems. Introduction GARDEL (ref. 1, 3) is an on-line core monitoring and analysis system for PWR and BWR. Its development started a few years ago with the goal to provide a standard system for the industry independent from reactor and fuel vendor, independent from the computer platform and fully compatible with Studsvik Scandpower's core design methods CASMO/SIMULATE (ref. 4, 5) and HELIOS/PRESTO (ref. 6, 7). Up to now GARDEL is in operation in 6 reactors from the vendors Westinghouse, Siemens-KWU and ABB 1

2 Combustion Engineering. A project for an additional installation in a General Electric BWR in the States is on the way and will be implemented in spring this year. An excellent overview of the status of today's core monitoring systems and the requirements in respect to methods and functions is given in ref. 2. Compared to these requirements GARDEL is an up-to-date core monitoring system providing all functions for core surveillance and operational support. The gap between off-line core design and online core surveillance is eliminated improving the accuracy calculating 3D core power distribution and core thermal limits and improving the whole core management process. GARDEL System Functionality GARDEL provides a common software platform for both BWRs and PWRs, but with the functionality tailored to each individual reactor type. The main functions are automatic surveillance of fuel thermal margins, PCI limits, etc. with displays for the Reactor Operators in the control room on demand computing capability to aid the Reactor Engineer in the support and planning of a safe and efficient reactor operation automatic documentation of reactor operation by means of periodic core supervision and isotopic reports automatic archiving of plant computer data and calculations into one encapsulated data base Core Monitoring The Reactor Operator in the control room has continuous access to summary of current reactor operating conditions current core thermal margins, PCI limits, etc. trend plots and displays of key operating parameters In case of BWR heat balance calculations are repeatedly performed with a high frequency for determination of reactor operating conditions. 3D core simulations are performed to evaluate the local power distribution in the core. The calculations are executed periodically (at least once per hour) or triggered on a predefined change in power, flow or control rod position. The operator may also trigger the 3D calculations manually. Based on the updated 3D power distribution parameters like thermal margins, Xenon/Iodine concentrations, fuel exposure, and PCI limits are calculated. Feedback from in-core and ex-core instrumentation is taken into account for validation of the calculated results. Adaptation to the instrument readings may optionally be applied in the post-processing of the thermal limits. Different adaptation methods are available. The authorised Reactor Engineer defines whether thermal margins shown to the operators should be based on pure predictions or adapted to the instrumentation readings. Independent of the process data acquisition system GARDEL has its own signal handling function indicating whether e.g. a measured signal is failed, automatically defaulted or manually deactivated for core surveillance purposes by an authorised user. 2

3 In GARDEL the same physics methods are used in the plant for core surveillance and operational support as in the core design work by the in-core fuel management. Significant benefits are gained both with regard to the accuracy of the core surveillance and the ease of updating the system for a new core loading. Operational Support The reactor engineer makes extensive use of GARDEL for detailed analysis of 3D results, such as plotting core maps, recording data trends, applying data filters, data operations on 3D distributions and data export functions. When the operation of the Travelling In-core Probes (TIP) has been activated, the data are transferred from the process computer into the GARDEL data base. The reactor engineer will then perform a verification of the data by comparison with the calculated 3D results and evaluate the data for acceptance/rejection. Furthermore, new adaptation factors may be calculated for application in the core monitoring. Other examples of GARDEL used by the reactor engineer are: re-calculation/analysis of past core conditions or events manoeuvre and long term predictions from current or past core conditions cold critical calculations reactivity coefficient calculations correlating predicted k-effective levels to core conditions Predictive calculations are facilitated by means of GUI tools for specification of operating scenarios, etc.. The 3D-simulators have been extended with GARDEL-specific options to ensure efficient and accurate predictive calculations. System Configuration and Software Design GARDEL is operating in very different computer environments and network configurations. Because of its modular design it is easy to install and to maintain. In many cases GARDEL is implemented as a redundant computer system providing a high system availability and reliability. GARDEL is using the HERMES data base which is an in-house development specially designed to handle large data arrays. The HERMES data base concept has been implemented in a wide range of nuclear engineering applications and ensures data compatibility all the way from lattice calculations through core design and on-line core monitoring. HERMES ensures a very high data access speed and data security while at the same time providing user friendly retrieval functions of an ordinary data base. The GARDEL graphical user interface is configurable to adapt it to the needs of different users like shift personnel, reactor engineers and system administrator. The experience of GARDEL installations from several plants is summarised in the next sections. Examples for comparisons of measured and calculated detector signals, implementation on different hardware platforms, and redundancy configurations are presented. 3

4 GARDEL at BEZNAU NPP Beznau NPP consists of two 2-loop Westinghouse PWRs. GARDEL is currently performing on-line core monitoring for units 1 and 2 since the summer of The operation of GARDEL at Beznau started in January 2000 with a demo installation. In April 2002 the system has been installed on the production hardware and since summer 2002 GARDEL is running on-line operation in unit 1 (cycle 31) and unit 2 (cycle 29) for final validation. Both Beznau units are identical, the main core parameters relevant for core monitoring with GARDEL are summarised in table 1. Rated core thermal power 1130 MW No. of fuel assemblies 121 Cycle length 1 year Current cycle 31 (unit 1), 29 (unit 2) In-core instrumentation Movable detectors (fission 30 strings chambers) Thermocouples at the top of 35 active core Ex-core detectors (power range) 4x2 Table 1: Core parameters of Beznau unit 1 and 2 About 20% of the fuel assemblies in both units are MOX fuel assemblies in the current cycle. GARDEL implementation GARDEL was implemented at Beznau on a double computer configuration as shown by fig. 1. SUN WS01 Gardel- 01 Process Computer System SUN WS02 Gardel- 02 Interactive Gardel users Fig. 1: GARDEL implementation at Beznau NPP Gardel-01 and Gardel-02 are SUN Blade 1000 computers running the operating system Solaris 5.8. SUN WS01 and SUN WS02 are the computers of the process computer system in charge of the data exchange with GARDEL. Gardel-01 and Gardel-02 are continuously receiving process data and sending back core monitoring results to the process computer system. Only one of the process computer servers can be active, being the other one in hot stand-by. The active process computer server will collect the results provided by 4

5 the GARDEL server directly attached to it and send these results further for presentation in the control room panels. The GARDEL servers continuously exchange information to verify each other s status (handshake). The redundancy logic is described more in detail for NPP Brunsbüttel below which is quite similar. Interactive users always logon to the active GARDEL server (the one attached to the active process computer) to run GARDEL-GUI. Off-line calculations can only be started through GARDEL-GUI and are executed on both servers simultaneously, i.e. no matter on which server the calculation was started, the user will find the results on both. Figure 2 shows an example of the main GARDEL-GUI window providing a summary of the most important core parameters and arrays. Main results Beznau has performed an extensive benchmarking of GARDEL s methods covering more than a total of 30 cycles of operation for both plants. The results of that benchmark fall outside the scope of this publication. This section below summarises the results for flux map comparisons for both Figure 2: GARDEL main window plants since GARDEL s start up in the production environment in the summer of 2002 and it shows an example for boron letdown and FdH margin. Table 2 gives a summary for the flux map comparison of KKB1 cycle 31 and KKB2 cycle 29. The statistic covers the power range between zero power and full power for both plants (c.f. figure 3 for KKB1 cycle 31). Unit, Cycle Average Average Average Nodal rms Axial rms Radial rms KKB1 C KKB2 C Table 2: Flux map comparison Figure 3 shows the comparison of measured versus calculated boron for KKB1 cycle 31. 5

6 Figure 3: Measured (BORON) vs. calculated (BORPRD) boron in KKB1 cycle 31 Comparison of predictive (FDHM0) vs. adaptive FdH margins is shown in fig. 4. FDHM1 is the adaptive margin calculated using information from latest flux map measurement only. FDHM2 is the adaptive margin calculated combining information from the latest flux map with the current thermocouple readings. Figure 4: Comparison of predictive (FDHM0) vs. adaptive FdH margins (FDHM1) Conclusions of GARDEL at Beznau The operation so far has been consistent with Beznau s benchmark: Very good agreement between calculated and measured reaction rates for all flux map measurements, with nodal rms systematically below 3% at rated power. The excellent radial flux map comparisons with rms values systematically below 2% confirms that GARDEL is properly modeling the MOX fuel assemblies. The MOX fuel assemblies do not show any particular trend as compared to ordinary UO2 assemblies. The level of agreement achieved between adaptive and predictive thermal margins indicates the reliability of GARDEL to function as predictive tool supporting cycle operation. 6

7 GARDEL at Brunsbüttel NPP Brunsbüttel NPP is a Siemens-KWU BWR. A GARDEL demo system is running in parallel on-line operation since April 2001 until the plant was shut down in February In August 2002 GARDEL has been implemented on the production hardware to replace the old core monitoring system after restart of the plant within the next cycle. The main core parameters relevant for core monitoring with GARDEL are summarised in table 3: Rated core thermal power 2292 MW No. of fuel assemblies 532 Cycle length 1 year Current cycle 17 In-core instrumentation Local power range monitors (LPRM) 120 Movable TIP detectors 30 strings Table 3: Core parameters of Brunsbüttel NPP GARDEL implementation GARDEL is implemented at Brunsbüttel on a double computer configuration as shown by figure 5. Gardel server S1 and S2 are continuously receiving process data from the process computer units PCU1 and PCU2. In normal mode GARDEL server S1 PC-Control room Teleperm Process Computer is active, S2 is in hot stand-by mode. The active GARDEL server sends back the core monitoring results to both process computer units. PC-Physics Gardel-Server S1 PCU 1 PCU 2 Gardel-Server S2 Fig. 5: GARDEL implementation at Brunsbüttel The GARDEL servers continuously exchange information to verify each other s status (handshake) using a fast direct network connection. In case the GARDEL server S1 stops responding, the S2 continues normal operation by taking over the active role and sending back results to the PCU's. Once the communication is re-established, the recovering GARDEL server will request the missing process data from the other server, and perform a fast replay to complete its database with the missing data. Once the replay is finished, the two servers are synchronised again and continue with ordinary operation. The handshake logic is a site-specific extension of GARDEL. The fast replay function is, however, a standard GARDEL feature. 7

8 The process computer units are SUN workstations of the Siemens Teleperm system. The GARDEL computers are Intel Pentium-III Server with Linux 7.3 as operating system. The redundant configuration of S1 and S2 components like double processors redundant power supply hot-swap mirror system disks (Raid level 1) hot-swap disks for data storage (Raid level 5) plus 1 spare disk guarantees a high availability of the GARDEL system. Conclusions of GARDEL at Brunsbüttel The GARDEL system is installed and prepared for the validation period after Brunsbüttel's restart for next cycle. GARDEL is supposed to take over the responsible core monitoring at KKB within cycle 17. Comparisons between the old core monitoring system and GARDEL during the demonstration period in cycle 16 have proven already to give the same results because both systems are using the same methods. The core simulator PRESTO-2 is already in operation as part of the old system. ABB CE PWR Application GARDEL is performing parallel on-line core supervision at an ABB CE-PWR since August The core instrumentation relevant for core monitoring is summarised in table 4. In-core instrumentation Movable detectors (fission chambers) Fixed detectors (Rh) Thermocouples at the top of active core Ex-core detectors none 28 strings with 4 axial positions per string 28 6x2 Table 4: Core instrumentation of ABB-CE PWR Current core supervision methods This reactor employs fixed in-core self-powered Rhodium detectors. The Rhodium in-core detector depletes approximately 1% per full-power month operation. Therefore corrections are necessary to the detector signal to account for the effects of the Rhodium depletion. These corrections are results from the change in Rhodium concentration itself, and the change in detector self shielding due to Rhodium burnup. CECOR is the currently licensed tool for continuous evaluation of thermal margins from the fixed in-core detector measurements in ABB-CE reactors. This package makes use of pre-defined coefficients to estimate the power distribution from the in-core detector readings applying several approximations: It works with quarter core coefficients. Any asymmetry is smeared off by the method. 8

9 All signal-to-power and geometric expansion coefficients are pre-calculated at core conditions that may differ from the actual conditions. In addition, the current system does not employ any detailed neutronic core model and therefore does not support calculations for operational support. GARDEL implementation For the ABB-CE implementation, GARDEL-PWR was extended to: Access raw Rh detector signals, integrate their depletions and calibrate them using the resulting sensitivities. Perform on-line calculation of measured detector signal to local power factors. Evaluate adaptive power distribution and derived thermal margins adjusting the power shape calculated by Simulate-3 with the local power measurement provided by the Rh detectors. Present relevant Rh detector and other reactor type-specific data in GARDEL s GUI. Support off-set core geometry. The hardware configuration at this ABB-CE plant consists of: HP-UX B server: this is GARDEL s computing and data server, all on-line processes run on this computer. Interactive users also execute GARDEL-GUI on the server. Several desktop Windows PCs. Each user uses his desktop computer as an X- terminal to run GARDEL s GUI on the server. Conclusions of GARDEL operation Although the system has not completed a whole cycle yet, the comparisons between predictive and adaptive power shapes from BOC until December 15 th show very good agreement. E.g. the average radial rms is 1.0%, the nodal rms is 2.7 %. This degree of accuracy would make it possible to reduce uncertainty factors used in the definition of core design and core surveillance thermal limits, as well as to use GARDEL for reliable prediction calculations. Experience at RINGHALS NPP Ringhals NPP consists of 4 units, a Westinghouse-Atom built BWR unit and 3 Westinghouse 3 loop PWRs. GARDEL is currently performing parallel on-line core supervision for unit 2 (since June 2002) and unit 4 (since September 2001). Results are presented for Ringhals unit 2 Ringhals Unit 2 is a Westinghouse-built 3-loop PWR. The main core parameters relevant for core monitoring with GARDEL are summarised in table 5. The main difference between Ringhals 2 and other 3-loop Westinghouse PWRs is that 12 movable detectors channels were replaced by fixed gamma thermometer strings with nine axial positions each. 9

10 Rated core thermal power 2652 MW No. of fuel assemblies 157 Fuel assembly design 15x15 pins w/o enrichment Cycle length One year Current cycle 27 In-core instrumentation Movable detectors (fission chambers) 38 strings Fixed detectors (gamma thermometers) 12 strings 9 axial positions per string Thermocouples at the top of active core 16 Ex-core detectors 4x2 Table 5: Core parameters of Ringhals 2 Current core supervision methods The objective with the replacement of some movable detector strings with fixed in-core detectors was to provide continuous monitoring of the power shape inside the core. The gamma thermometers were selected as fixed in-core instrumentation based on features like the possibility for absolute calibration, robustness and an expected long lifetime. Since the introduction of the gamma thermometers, Ringhals 2 has been using a combination of different software packages for on-line core monitoring: CECOR: calculates thermal margins (F Q and F H) continuously using gamma thermometers and process signals as input. TEL-HC: calculates DNBR margin continuously based on process signals. SCORPIO: strategy generator with a core simulator and process signals as input. INCORE: licensed tool for evaluation of thermal margins from movable detector measurements every 30 EFPD. RADCAL: converts and adjusts raw gamma-sensor signal to useful signal units for use in CECOR. This combination of packages suffers limitations with regard to combining movable detectors and gamma detectors and use of simplified methods that limits its applicability to modern fuel and core designs. GARDEL implementation The limitations mentioned above were the motivation to introduce GARDEL to evaluate whether it would be capable of replacing the software packages in use. The standard GARDEL system was therefore extended to: Access raw gamma-thermometer signals and calibrate them using calibration factors from the latest heater cable auto-calibration. Perform on-line calculation of gamma-thermometer signal to linear power factors. Evaluate adaptive power distribution and derived thermal margins adjusting the power shape calculated by Simulate-3 with the local power measurement provided by the gamma-thermometers. Present gamma-thermometer and adaptive results in the GUI. The hardware configuration in Ringhals 2 consists of: 10

11 SUN blade 100 work station: this is GARDEL s computing and data server, all online processes run on this computer. Several desktop Windows PCs: each user runs GARDEL s GUI on his own desktop computer, accessing the on-line data through the network. Cycle 27 operation, initial observations GARDEL has not yet completed a whole cycle of operation at Ringhals 2. However, this short operational experience has been sufficient to show some unexpected results. Those results meant a reformulation of the adaptive methods. The gamma-thermometers are expected to provide an absolute measure of the linear power in the vicinity of the detector. In reality, a detailed tracking of the gamma-detectors and derived power shapes with GARDEL has shown that the reliability and accuracy of the gamma-thermometers is limited and their use for online adaptation must be carefully evaluated. Conclusions of Ringhals 2 C27 operation Ringhals 2 s operation since BOC27 until December 15, 2002, has shown: Good agreement between purely predictive and measured power distributions based on the movable detectors. After many years of operation, the gamma thermometers have a limited capability to provide a reliable absolute measure of the local power. Combining information from movable and fixed in-core detectors to calculate adaptive power distribution and margins seem to be the most appropriate method. The conclusion of cycle 27 s operation is that GARDEL is capable of replacing the current software methods while improving the accuracy in the determination of thermal margins. References A.Noël, A. Becker, " State-of-the-Art Core Monitoring in BWR and PWR", Proceedings of Jahrestagung Kerntechnik, Fachsitzung, Dresden, Mai T.Lefvert, On-Line Core Monitoring for BWR, PWR. Overview and Comments on Operation Experience and Further Development, Annex 1, Proceedings of the Workshop on Core Monitoring for Commercial Reactors: Improvements in Systems and Methods, Stockholm, 4-5 October A.Noël, L.J.Covington, A.Nilsson, D.Greiner, Core Monitoring based on Advanced Nodal Methods: Experience and Plans for Further Improvements and Development, p.163, Proceedings of the Workshop on Core Monitoring for Commercial Reactors: Improvements in Systems and Methods, Stockholm, 4-5 October D. Knott and M. Edenius, The Two-Dimensional Transport Solution within CASMO-4, Trans. Am. Nucl. Soc., Vol. 68, p 457, June 1993 Kord Smith, et.al., "SIMULATE-3 Advanced Three-Dimensional Two-Group Reactor Analysis Code", SOA-95/15, October J. Casal, R. Stamm'ler, E. Villarino and A. Ferri, "HELIOS: Geometric capabilities of a new fuel-assembly program," Intl Topical Meeting on Advances in Mathematics, Computations, and Reactor Physics, Pittsburgh, Pennsylvania, April 28-May 2, S.Børresen, N.E.Patiño, Methods of the Advanced Nodal Simulator PRESTO-2, Proc. ANS Topl. Meeting, Vol.1, Myrtle Beach, South Carolina, March