Contact Author:
Prof. Vladimir Valentinovich Likhanskii
likhansk@mail.ru
P:+7 495 3346182
F:+7 495 3345158
V.V. Likhanskii,
SRC RF TRINITI
Troitsk, Moscow region 142190
Russia
WWER EXPERT SYSTEM FOR FUEL FAILURE ANALYSIS USING DATA ON PRIMARY COOLANT ACTIVITY
I.A. EVDOKIMOV (SRC RF TRINITI, Russia), A.A. SOROKIN (SRC RF TRINITI, Russia), A.G. KHROMOV (SRC RF TRINITI, Russia), V.D. KANUKOVA (SRC RF TRINITI, Russia), O.V. APOLLONOVA (SRC RF TRINITI, Russia), A.V. UGRYUMOV (JSC TVEL, Russia)
WWER EXPERT SYSTEM FOR FUEL FAILURE ANALYSIS USING DATA ON PRIMARY COOLANT ACTIVITY
V.V. LIKHANSKII, I.A. EVDOKIMOV, A.A. SOROKIN,
A.G. KHROMOV, V.D. KANUKOVA, O.V. APOLLONOVA
SRC RF TRINITI, 142190, Troitsk, Moscow Reg., Russia
A.V. UGRYUMOV
JSC TVEL, 119017, 24/26 Bolshaya Ordynka st., Moscow, Russia
ABSTRACT
Fuel failures at operating LWRs lead to release of radioactive fission products into primary coolant. Local massive hydriding of cladding following a failure may result in significant cladding degradation and washout of fuel fragments from failed fuel rod. In this case severe contamination of primary circuit is possible.
To provide radiation safety at nuclear power plants (NPPs) failed fuel is identified in leakage tests during reactor refueling outages. These tests are time consuming and involve high financial costs. To reduce expenses and a risk to miss a failure most NPPs use a preliminary evaluation of failure characteristics before the end of fuel campaign. The major failure parameters to be determined are the number and burnup of failed fuel rods as well as leak size.
The state-of-the-art methods of failure evaluation under operation conditions are based on monitoring of primary coolant activity and application of computer codes for data interpretation. These methods can be integrated into a computer expert system for the on-line failure diagnosis at operating units.
At present time the expert system for failure diagnosis at operating WWERs is under development [1-3]. The system is based on the mechanistic RTOP-CA code and application of neural network methods. The RTOP-CA code is a tool for solving the direct problem – prediction of coolant activity when parameters of failure are known. The RTOP-CA serial calculations with variation of failure characteristics make it possible to configure and train neural networks for solving the inverse problem – evaluation of failure parameters by measured data on primary coolant activity.
Analysis of coolant activity data by the expert system includes:
· evaluation of tramp uranium mass in reactor core;
· detection of failures by iodine and caesium spiking;
· burnup estimation for failed fuel by ratio of 134Cs to 137Cs activity in spikes;
· a new technique for estimation of failed fuel burnup by relationship between activities of gaseous fission products under conditions of steady state reactor operation;
· assessment of fuel washout for severe failures;
· application of neural-network analysis for getting failure characteristics in more detail.
The current version of the expert system has been tested against data for several fuel campaigns at different WWER units in Russia. The data included results of coolant activity monitoring during a campaign and results of leakage tests after reactor shutdown for refueling. In all the cases predictions of the expert system were in good agreement with findings of leakage tests [3].
Since publication [3] the neural network program incorporated into the expert system has been optimized for higher reliability of predictions. The paper comprises some new results concerning verification of the modified version of the expert system against WWER data.
The technique for estimation of failed fuel burnup by relationship between activities of gaseous fission products in primary coolant is presented. The main idea lies in dependence of fission yields of radionuclides on fuel isotope composition. In course of in-reactor operation plutonium is accumulated in UO2-fuel. Amount of plutonium is a function of burnup. With Pu being accumulated in fuel krypton nuclides (83m,85mKr, 87-90Kr) demonstrate more significant decrease of cumulative yields than xenon (133,135Xe). So, the measured ratios of krypton to xenon activities in coolant should indicate the burnup of failed fuel. New results of application of this technique for estimations of failed fuel burnup at operating WWER units are presented.
REFERENCES
1. V. Likhanskii, I. Evdokimov, A. Sorokin, V. Kanukova, A. Khromov, E. Afanasieva, “Failed fuel diagnosis during WWER reactor operation using the RTOP-CA code,” Proc. 6th Int. Conf. on WWER Fuel Performance “Modelling and Experimental Support", Albena, Bulgaria, 19-23 September 2005, paper 1.13.
2. V. Likhanskii, E. Afanasieva, N. Efremov, I. Evdokimov, V. Kanukova, D. Kirilenko, A. Khromov, A. Sorokin, P. Svotin, V. Molchanov, A. Sharikov, “Integrated approach to detection of defective WWER fuel assemblies” Proc. Water Reactor Fuel Performance Meet., Kyoto, Japan, 2-6 October 2005, pp.382-395.
3. V.V. Likhanskii, I.A. Evdokimov, A.A. Sorokin, A.G. Khromov, V.D. Kanukova, O.V. Apollonova, V.B. Ionov, A.V. Ugryumov, “Development of Expert System for Failed Fuel Diagnosis under WWER Operation Conditions,” Proc. Int. Meet. LWR Fuel Performance, TopFuel 2006, Salamanca, Spain, 22-26 October 2006, pp.466-470.