Search

This paper presents our practical experience of expert opinion deduced from concrete field applications, concerning mainly safety and reliability modeling. First, expertise is defined and the approach used for eliciting expertise is shortly described. Quality and value of information from experts are important challenges for decision making from risk analysis. Expertise is very useful for supplying data when not enough field observations are available for estimating parameter uncertainties and quantifying risk analysis models, which are inputs to a decision-making process. Applications presented herein concern mainly the role of expert judgment in reliability.

When too many failures occur on a given piece of equipment, the dependability engineer needs to decide whether these failures are attributable to poor initial design or if they are due to a phenomenon of aging. If aging is confirmed, the problem is then to determine the moment at which the process began and what corrective measure (generally, a modification in the design or in the preventive maintenance program) is the best suited to delay the occurrence of the failure. This measure will thus make it possible to extend the lifetime of the equipment. The method is based on the simple hypothesis of a model of step aging and on Bayesian techniques. The principal benefit of this method is the determination of the time at which aging begins (and the related uncertainties), the evolution in the failure rate of the component in its initial state and once modified, and the probability of success of the corrective measure. The IBTV software was developed to implement this methodology.

Reliability-centered maintenance (RCM) is a rational approach that can be used to identify the equipment of facilities that may turn out to be critical with respect to safety, to availability, or to maintenance costs. Is is dor these critical pieces of equipment alone that a corrective (one waits for a failure) or preventive (the type and frequency are specified) maintenance policy is established. But this approach has limitations: • •when there is little operating feedback and it concerns rare events affecting a piece of equipment judged critical on a priori grounds (how is it possible, in this case, to decide whether or not it is critical, since there is conflict between the gravity of the potential failure and its frequency?); • •when the aim is propose an optimal maintenance frequency for a critical piece of equipment - changing the maintenance frequency hitherto applied may cause a significant drift in the observed reliability of the equipment, an aspect not generally taken into account in the RCM approach. In these two situations, expert judgments can be combined with the available operating feedback (Bayesian approach) and the combination of risk of failure and economic consequences taken into account (statistical decision theory) to achieve a true optimization of maintenance policy choices. This paper presents an application on the maintenace of diesel generator component.

This paper describes the procedure Electricite de France uses to exploit the information on pressurized water reactor operation it receives back from the field (operation feedback). The first requirement in analyzing such data is a knowledge of past records. The first step, therefore, is to record the data, particularly events occurring on the plant and failures occurring on equipment, in large reliability data banks. However, the ‘raw’ information stored is rarely usable directly. The first step in the second stage—analysis—is to review and qualify the data before using it for any purpose. This difficult, but essential, review provides valuable information on the improvement of equipment reliability. The greater knowledge of plant and equipment behaviour, and the damage mechanisms involved, allows: • ⊎ safety to be kept at a high level: operation feedback is also essential for probabilistic safety studies, • ⊎ improvement of availability and preventive maintenance practices, • ⊎ correction of the initial design (design changes) and help for designing future plants. Finally, operation feedback is a source of progress. Although it requires heavy initial investment, it is also a source of profit. It is a source of learning. The analysis results make it possible to define more suitable procedures and better preventive maintenance practices and thus improve the operation and safety of existing and future plants.

Cet article vise a fournir une methodologie generale d'etude du vieillissement industriel d'un composant ou d'une installation. Ce vieillissement se traduit par une degradation continue, plus ou moins rapide des performances des composants, systemes et structures, ou par leur perte de fonction. Un certain nombre de parades (programme de maintenance, de renovation...) peuvent ralentir ou limiter ce phenomene, permettant ainsi d'optimiser leurs performances, tout en maitrisant les risques. Encore faut-il anticiper les defaillances potentielles, et prevoir les consequences a long terme des remedes et des options mises en oeuvre preventivement. Pour cela, le retour d'experience d'exploitation et son analyse, l'utilisation de l'expertise industrielle sont des elements strategiques. L'article insiste sur l'anticipation puis sur l'evaluation de la fiabilite previsionnelle, qui a pour but d'optimiser le choix de l'option preventive et de maitriser les risques industriels. Il conclut sur les caracteristiques essentielles d'une etude de vieillissement et sur quelques pistes de recherche et developpement pour le futur.

A sum up of the EDF feedback organisation is presented. The three main files are described: the event data bank, the incident data bank and the component event data bank.

At the beginning of the 1970s, when Electricite de France (EDF) opted for a policy of rapid development of its PWR nuclear power plant programme, the demonstration of the anticipated benefit with respect to the required safety was the major objective of engineers. At that time only foreign experience was available, and the collection and recording of this experience in a computerized data bank was decided upon for use in prospective studies, while waiting for the creation of specific French data banks as from 1978. All the collected data are recorded in the EDF Research and Development Directorate central computer (IBM) located at Clamart (near Paris), either directly for the foreign data bank or by the plant local computer (CII), with subsequent transfer onto the central IBM, for the French data banks. These data banks are particularly concerned with reliability and availability of plants, or of their components. Only the component reliability data bank will be described here in detail.

Steady state thermal measurements, water and shell temperature, are made in the downcomer of the steam generator of the PWR BUGEY-4. The measurements show that the flow is of a swirling type with the degree of swirl being a function of the load (power). The angular temperature distribution of water and shell exhibits a minimum that rotates from the hot leg to the cold leg with increasing load. The results also admit the conclusion that the feedwater and recirculation water are well mixed in the cylindrical section of the downcomer. Measurements and computations of fluid and shell temperatures are in good agreement for the steam generator in a state of natural circulation. For the state of forced circulation, the agreement is poor when it is assumed that the flow in the downcomer is turbulent and the steam generator outer shell is well insulated. The agreement is excellent for turbulent flow but with air leakage, or infiltration, between the steam generator outer shell and its insulation. If the hypothesis of air infiltration is ruled out, agreement between measurements and computations is much improved when the flow in the downcomer is taken as a turbulent core flow with an attendant laminar boundary layer. The existence of a laminar boundary layer would require the flat plate transition Reynolds number be enhanced by a factor of 100. This enhancement could result from the combined effects of swirl and density gradient in the downcomer.

Two transients, an open grid and a scram at 50% load, were conducted on unit 4 of the PWR power plant Bugey. The thermal hydraulic response of the steam generator was recorded. For the open grid test, the following observations are noted: No alarming phenomena are observed in the steam generator during the transient. Primary pressure oscillations were very mild, and did not exceed about 4.8 bar/min with a maximum amplitude of ±8 bar. This condition should not result in significant stress levels. Steam generator outer shell metal temperature gradients remained within very acceptable limits; a maximum amplitude of about +13°C and a rate not exceeding 0.8°C/min are obtained. This slow rate is explained by a fall in primary water temperature that allows for a temperature decrease inside the U-tube bundle. Similarly, the temperature rise on the tube sheet does not exceed an amplitude of 20°C with a rate of about 2°C/min. Again these conditions do not lead to any significant thermal shock on the tube sheet. The steam generator feed controls maintain the level within the normal operation range and the small addition of colder feedwater does not lead to great temperature changes because of the large mass of the recirculation water in the steam generator. For the scram at 50% load, the following observations are noted: no severe thermal or pressure transients are observed in this test. Fluid temperature fluctuations occur with rates not exceeding 1°C/s and a maximum amplitude of about 20°C in the downcomer and 10°C on the tube sheet. Steam generator outer shell temperature varies at a rate of about ±0.8°C/min with a maximum amplitude of about 16°C. These thermal transients should lead to thermally induced stresses of acceptable levels.

Systeme de collecte et de traitement de donnees concernant la fiabilite des centrales nucleaires en France

Measured downcomer water and metal shell temperatures in the steam generator No. 20 of the PWR Tricastin 1 show that the downcomer flow is of the swirling type, just as found previously in Bugey 4. A comparison of results for Tricastin 1 and Bugey 4 shows that the addition in Tricastin 1 of a flow distribution baffle plate, between the tube sheet and the first cross plate, while reducing the height of the opening between the tube sheet and the shell surrounding the bundle, may have resulted in the observed reduction (by a factor one half) of sludge deposit upon the tube sheet in Tricastin 1, and in fixing, with extended period of operation, the boiling zone in the cold leg (a desired event) and near the tube-free tube lane.

Two Electricite de France data banks have been used to analyse reliability characteristics of two critical components of French nuclear power plants. These components are: primary and secondary valves and feedwater pumps. The main failure modes and causes are presented with their perspective failure rates. The impact of components failures on plant availability is shown. Practical conclusions are drawn concerning pump maintenance intervals.

Downcomer fluid velocity, shell, fluid, and tube sheet temperature, and feedwater distributions across the cold and hot legs of Paluel 1 steam generator No. 81 are measured as a function of load. The results indicate that swirling motion in the downcomer is negligible compared with results reported earlier for the PWRs Bugey 4 and Tricastin 1. Overall circulation ratio and saturation pressure decrease almost linearly with increasing load. Carry- under and carryover are found negligible at all loads. A nonuniformly drilled feedwater ring distributes 80% of the feedwater onto the hot leg and 20% onto the cold leg. The combination of hydrocyclonic steam-water separator dryer design and nonuniformly drilled feedwater ring is successful in retaining a high percentage (∼ 70%) of the rather pure (compared with recirculation water) feedwater at tube sheet level across the hot leg, even at full load. Consequently, chemical quality, and turbidity level of secondary water above the tube sheet in the hot leg is found superior to that in the cold leg.

Thermohydraulic measurements made on steam generator No. 14 (Westinghouse type 51A) of unit 4 of Bugey (a French PWR) allowed the determination of the circulation ratio and the percentage of carry-under as a function of the percentage of full load. At full load the circulation ratio, obtained from a mass balance, was found to be about 4.6, rising to about 14.6 at 30% of full load. Both exceed the estimated design value of 4.17 at full load. The percentage of carry-under is found to be negligible and relatively independent of load up to about 50% of full load, but increases thereafter to about 1.23% at full load. This level of carry-under does not appear to present any cavitation problems in case of sudden depressurization.

This paper presents the EDF reliability data collection system (SRDF), operating since 1978 on 6 PWR nuclear units and collecting at the end of 1982 24 units × years of experience. About 1000 components are monitored for a pair of units, among them the most important for the safety of safeguard systems or for the availability of the plants: pumps, valves, motors, turbine, generators, etc. During 1982 a balance of the positive and negative aspects of the SRDF has been made before extending the system to all the EDF PWR units and to the first large fast breeder reactor: Super Phenix. During may 1983, 12 new units have been connected to the data collection system. The revision of the SRDF has allowed us to develop a failure logic analysis method, a combination of the chronological description of the failure and the event and fault tree analysis of this failure, giving a better homogeneity between information coming from different plants and a large simplification of the interrogation system. One example is given with the statistical analysis of the operation of the valves.

This paper discusses the application of a thermodynamic method of measurement to determine the efficiency of boiler feed pumps in large and modern generating sets, and reasons for the development of the method are presented. The principle of the thermodynamic method is defined, and the difference between this method and the classical one is specified. A description is given of the instrumentation used when the thermodynamic method is applied. The results of thermodynamic measurements are analysed and discussed, and the calculated efficiency is compared with the efficiency determined when the classic method is used. The paper finally presents the programme of trials that were undertaken on the boiler feed pumps of modern units of Electricité de France.