Prooress in Nuclear Energy, Vol. 12, No. 3, pp. 227 233, 1983. 0149 1970/83 $0.00+.50 Printed in Great Britain.

PWR NUCLEAR POWER PLANT OPERATION AND INCIDENTS

H. PROCACCIA

Division Fiabilit6 Fonctionnement, Electricit6 de France Etudes et Recherches, 25, Allee Privee, 93206 Saint-Denis, France

(Received 22 August 1983)

1. INTRODUCTION

In 1974 the French Government opted for a policy of rapid development of its PWR nuclear power plant programme. The Westinghouse PWR system was selected at this time. Using documents published by the United States Nuclear Regulatory Commission (Grey Books), the Research and Development Department of Electricit6 de France (EdF) decided to establish an initial data bank based on the American experience with the operation of these power plants.

This data bank is concerned with general inform- ation such as availability and power production figures for American PWR nuclear power plants in addition to data about equipment incidents which affect electrical output or which involve violation of technical specifications. The purpose of this data bank is to support prospective analyses of behaviour of future power plants and to assist in the evaluation of proposed improvements. Its name is FI.

As a substantial body of operating experience accumulated with plants in France, over a number of years a second data base was established specifically for events involving the operation of French nuclear power plants. This file is called FE. It must fulfil three objectives, namely:

- -Describe events involving nuclear power plants. - -Faci l i ta te technical data exchanges by reducing the

amount of physical documents required. - - Improve the accessibility of the data to the numer-

ous users.

It is necessary that this file be of national scope since it concerns nuclear power plant operators, as well as safety authorities, and the technical teams administra- tively in charge of nuclear power plant operation. This file was provisionally inaugurated by the Research and Development Department (Reliability and Operations Division) in January 1981. The Thermal Production Department (General Studies Division) officially put the file into service two years later, in January 1983.

In 1978 a third data file, called SRDF, was set up independent from the other two and concerned more with the general operating problems of the power production units. This file is more detailed so as to provide data on the basic reliability of major equip- ment components for considerations of safety, or installation on units under construction or power unit availability.

Table 1 shows the general relationships and func- tions of the three main data systems which are in use. The foreign data is compiled from available sources including the International Atomic Energy Agency (Vienna), the 'Grey Books' (US-NRC), Significant Events (US-INPO) and commercial sources, such as Westinghouse, Nuclear Plant Experience and Atomic Energy Clearing House.

The balance of this paper describes the other two systems, namely: SRDF, which covers Components (Syst6me de Recueil de Donn6es de Fiabilit~, or Reliability Data System), and RDE, which covers events (Recueil Donn6es D'Exploitation, or Operating Data Collection), and comes from FE.

2. RELIABILITY DATA GATHERING SYSTEM (SRDF)

In 1974 Electricit~ de France decided to implement the Reliability Data Gathering System (SRDF). The first application was for the six 900 MWe PWR units of Fessenheim and Bugey. The purpose of this system is to enable one to conduct a thorough monitoring of the behaviour of certain operating equipment selected according to its importance, in relation to either the safety or the availability impacts of the equipment monitored.

2.1. Equipment coverage

The information gathering started in 1977 in Fessenheim, initially on the basis of a list of 4000 pieces of equipment to be monitored. The numerous dif-

227

228 H. PROCACCIA

Table 1. Nuclear EDF power plant data banks

Incidents Collection (FI)

-Foreign Events '74

-Incident Data Collection

U.S.

Japan Germany J Flhien Incidents Other

.Reports Compiled Yearly Comparisons Annual Outputs Cumulative System or Component Behaviour

Events Collection (FE)

f! RANCE '78

aily Telex reports

vents File (Fichien Evenements)

eports Co(m~i~)ed Monthly Comparisons Cross Check SRDF Operation Data Collection synthesis (RDE)

-Component Data (SRDF)

lO00 Components SRDF File

Coverage - Since (Date)

6 units "78 18 units "83 28 units "84

-Reports Compiled Comparisons with

Events File Foreign Data

.Provisional Reliability Date File

(RPDF)

ficulties encountered led to a revision of the project, reducing the number of pieces ofequipment to 800 per pair of units when the system was put into operation in April 1978. Presently, 1000 pieces of equipment are followed:

400 electrical components, including 130 motors and 140 electrical devices such as circuit breakers, contactors, etc.; 600 pieces of mechanical equipment, including 150 pumps and 350 valves.

Apart from their importance in respect to safety or availability, these pieces of equipment generally have the characteristic of being equipped with surge meters and timers in order to monitor the numbers of efforts and the operating time.

2.2. Information gathering

The information gathering for the pieces of equip- ment listed above is done with five different categories of data input cards, the contents of which are described in the following sections.

(a) The description cards: A single identity card is permanently established for each piece of equipment. It gives the technical characteristics of the piece of equipment concerned, specifying the operating con- ditions and environmental information. These cards enable one to automatically establish identical groups of equipment which will be subjected to statistical processing.

(b) The operating card: Each year a card is completed which summarizes the main reactor operat- ing data and data for each monitored piece of

equipment. The number of operating hours is given as well as the number of efforts during the past year.

(c) The failure cards: For any failure of a monitored piece of equipment, a card is issued which contains the following available information:

the time of failure the modes and causes of such failure its consequences the circumstances in which it occurred the time of unavailability the time of repair the cost of repair

Furthermore, a clear description of the incident completes the above mentioned information.

Since the definition of failure modes and causes can be interpreted in different ways, however, in many cases it was difficult to determine if an incident characteristic had to be classified among the modes or the causes. Thus, even though it was first decided to combine these two features by establishing at the failure card level an item entitled 'Modes and Causes of Failure', various exchanges of views with the rep- resentatives of other companies (which also have data gathering systems) later led to a thorough re- evaluation of the subject. It was finally decided to create three new items, instead of the 'Modes and Causes' item, which are:

(I) a list of modes -I . . _ ) connccteo ny

(2) a list of causes ? . . . . . j m u l t tree anatys,s

(3) a list ol spare parts.

(d) The unavailability cards: Cards were initially created for all the unavailabilities affecting the moni-

PWR plant operation and incidents 229

tored equipment---except, of course, the ones which are the consequence of a failure and which are the subject of a failure card. Now they are only completed in specific cases because of the work load it involves.

(e ) The systematic maintenance card: At the time of the design of the SDRF, it was planned that any systematic maintenance operation performed on a monitored piece of equipment would be the subject of a specific card. Here again, because of the work load it involves, this could not be done. However, when the system next gets modified, the information regarding the frequency and the nature of the systematic maintenance operations will be incorporated in the description cards.

2.3. Information processing at the local level

For each pair of reactor units, an agent who is at the executive level is responsible for the system. He is usually assisted by a technician. He compiles computer files in which all the information contained in the cards are stored. The data input is done through a display keyboard connected to the site computer. Before any input the card is checked and the name of the person who completed it as well as the name of the person who checked it both appear on the card. This procedure is part of a wider structure, which was defined at the same time as the system in order to insure a certain standard of quality to the gathered information. As soon as the card is introduced, it is subjected to a certain number of tests, performed automatically, which allow one to check the accuracy of the information which is judged to be absolutely necessary, etc. From the information contained in the cards, the site computer makes up groups of identical equipment which will later be the subject of statistical processing. Periodically, a sum- mary of the failures is done for each group of identical equipment. The input information can be checked, modified and completed at any time.

Software, called the Reliability Data Interrogating system, allows one to consult the files through questions put in the form of Boolean equations. It is therefore possible to list the failures which occurred during a given period on a certain type of equipment which belongs to a given system. Such a system, because of its flexibility, quickly gives exact answers to the questions asked by the operating personnel.

2.4. Data processing at the national level

The data furnished by the nuclear plants are all gathered in Clamart, France, at the Calculation Center of the Study & Research Management. The files made up from this data can be reviewed at any time

according to various types of classification criteria. Various types of processing on request, presently being defined, will soon enable the users to ask questions of a various nature: reliability parameter calculation of a population of equipment of given characteristics, research of the law, research of the statistic which will be best adapted to represent the life time of a given set of equipment, etc.

For the time being, the main task consists of the yearly systematic processing. This processing, per- formed each year at the same time, aims to update completely all the files from the information gathered during the preceding year. Then the retrieval program applied to the updated files gives four compilations, which we will refer to as Tables, as follows:

Table I supplies the specification of the monitored pieces of equipment and specifies for each equip- ment some statistical data such as: operating time, number of efforts, number of shutdown failures, operating failures, and number of failures under effort. This information is supplied for the preceding year of observation, as well as for the whole period starting on the date of the start of the collection. Table 2 specifies the constitution of identical equip- ment groups. The software of the centralized processing reconstitutes the groups from the des- cription cards supplied by all the nuclear units. Two pieces of equipment are considered identical when they are of the same type (motor, circuit breaker, tank, etc.) whenever they are produced by the same manufacturer who gave them the same references. For each national group created in that way we obtain the numbers of the local groups which are part of it, as well as the exact identification of all the equipment it contains. Table 3 gives the statistical results with respect to the identical equipment groups. The table shows the results of the mathematical processing concerning the information gathered in Tables 1 and 2, which themselves constitute only part of the retrieval in a specific form of the information collected on the site. Results are proposed at different levels which correspond to populations less numerous but more homogeneous:

For all the equipment of a given type (ex: pump, alternator, etc.).

- - F o r all the equipment which constitutes a nat- ional group of identical pieces of equipment. For each 'local' group of identical pieces of equipment.

The retrieval supplied at each of these levels includes the following data:

230 H. PROCACCIA

The size of the population concerned. The cumulative operating time. The cumulative number of efforts. The number of shutdown failures, operating failures

and failures under effort. The number of failures which caused the replace-

ment of the equipment. The rate of shutdown failures, failures under effort

and operating failures. The average repair time on the spot. The average repair time in the shop. The average time of equipment replacement. The average duration of unavailability following the

failure.

As in Table 1, the information is supplied for the preceding year of observation and for the entire period starting on the date the data started to be collected. All the reliability parameters are shown in the form of their average value and of the value of the limits of the confidence interval at the level of 90~o. For the time being, since the quantity of available data is relatively small, only the model of the exponential law has been selected for all the parameters, but the software allows also for Weibull and Erlang-type models, etc., and includes some tests which allow one to check the validity of the used function selected to fit with various parameters to describe failure rates.

2.5. Qualitative and quantitative use of the data

The analysis of the equipment reliability conducted at the Electricit6 de France provides both qualitative and quantitative aspects of equipment reliability.

A qualitative analysis of the behaviour of the operating equipment is performed in all the nuclear plants which are just starting operation. It helps to know the influence of each component on the availabi- lity of the nuclear plant and to anticipate the possible need for design modifications.

A quantitative analysis of the equipment behaviour is also necessary in order to respond to any demand requiring a survey evaluation of the nuclear unit system safety. The corresponding surveys are con- ducted with the use of probabilistic methods which require the knowledge of the basic reliability data of the equipment concerned. The first data which have been used come from the analysis of the operating behaviour or the analysis of the reliability of such equipment availability (data from thermal plants). The uncertainty of such data was not an obstacle for the launching of the studies. This difficulty is by-passed:

- - b y taking into account the amplitude of the variations which are likely to happen:

- - b y using the confidence interval and not absolute values;

- - b y applying methods of relative judgement.

To make up for these above-mentioned limitations, Electricit6 de France established a powerful data gathering system. It is still too soon to judge the full impact of SRDF as a tool for improvement in plant reliability. However we are of the opinion, in the light of the experience acquired, that this sytem as a general rule meets the objectives which were set for it. Some minor modifications have been incorporated and there are likely to be others, but the structure of the system appears to be satisfactory.

In order to obtain satisfactory reliability data for the electromechanical equipment and the mechanical equipment, it is desirable to have an extended period of observation to have an adequate size of the homogen- eous statistical population. This is why the develop- ment of the SRDF is expected to have increasing impact with time. First, of course, the access to and use of reliability data will be easier. A deeper impact expected is from the vocabulary, the concepts, and the reasoning specific to the subject of 'reliability'. These should contribute some elements new to daily experi- ence of the operating personnel and to supplement or supplant intuitive or localized concepts of reliability.

It appears important to carry on further efforts, increase the coverage of reliability data to increase the communications between the users of data for similar equipment, various installations, of important steps toward efficient communication of operating experi- ence internationally have been taken, and this useful exchange is worthy of further efforts to broaden its coverage.

3. THE RDE SYSTEM FOR COLLECTING OPERATING DATA AT FRENCH NUCLEAR

POWER PLANTS

An Event Data Bank (FE) was established in all of EDF's nuclear power plants in 1981. At first, the basic data consisted of all periodicals and occasional documents issued by the Department of Thermal Production. These documents include:

The Monthly Progress Report of starting units (Document 412) The internal or external incident cards (Document 221) The Equipment Damage Card (Document 222)

As of March 1981, additional information is col- lected from the Event Files and included in the French nuclear power plants operating data collection: RDE.

PWR plant operation and incidents

3.1. Implementation of the RDE data bank

Three cards are used to feed the bank. These are: (1) An identification card (2) An availability card (3) An incident card

3.1.1. Identification card. This card includes every- thing which is required to identify the unit. At present, 40 American units and 22 French units have been entered as they were put into commercial operation.

3.1.2. Availability card. The availability card is completed from monthly operating data. There is one availability card per unit per month. This applies to all units from their commercial operation.

The data collected at this level allows one to obtain some results. It is necessary, however, to give some definitions:

- -The power of the unit is the net design power. - -The service factor is the ratio of the number of hours

during which the generator is synchronized to the network, and the number of hours of the reference period.

- -The availability factor is the ratio of the number of hours during which the plant is available and the number of hours of the reference period.

- -The output factor is the ratio of net electrical power generated and the net design power during the reference period.

3.1.3. Incident card. This card is completed each time there is a plant shutdown regardless of cause (i.e. incident, damage, test, maintenance, etc.). In addition, a card is filled out for any equipment event regardless of load change. This type of card only gives inform- ation on the behaviour of operating equipment. This card is used only for a specific incident or a specified shutdown at a given date.

The reason given for the shutdown allows one to determine if it was due to equipment failure, mainten- ance, or if the causes were not related to the equipment itself. Reasons for shutdowns are as follows:

failure operating error fuel reloading, maintenance, inspection equipment maintenance test restriction by administrative decision extension of reloading due to equipment failure or

maintenance extension of reloading not due to equipment failure

or maintenance external cause

231

The equipment affected by shutdowns or incidents is filed either in plant sub-groups themselves divided into functional circuits or into components. Using the mentioned system, a table can be established in which the columns represent the sub-groups and the lines represent the components. They are as follows:

l reactor (fuel not included) 2--primary pumps 3--steam generators 4--reactor coolant system 5--auxiliary and safeguard system of the reactor 6 instrumentation and reactor control 7--water plant 8 ~ o w e r generator and internal feeder 9 fuel handling

10 waste treatment I 1--containment 12~-other systems 13--non-specified systems

An item of equipment can only have one assignment in a functional system which is part of a given sub- group. Some equipment, because of its importance, is considered as a separate sub-group (a steam generator, for example). The components are parts which have the same function in different systems. The designation of the components is the one which was given by the U.S. Atomic Energy Commission when the bank was first established in 1975. Thirty-seven codes are used. Consequently, using the above mentioned method- ology for matrix filling, any damage, shutdown, or incident reported in the basic documents has an assignment.

3.2. Use of the RDE data bank

Presently it is possible to retrieve the following information as listings or tables:

listing of plant characteristics tables of different operating factors incident listings shutdown time period shutdown and incident tables

3.2.1. Operatin 9 parameters of a nuclear unit. The data is fed into the computer in a table format. These tables may cover a group of given power plants, a given power range, and a given time range equal to 12 months or less. Three tables with the same presen- tation are obtained systematically and provide the results of the following factors: (1) Service Factor (2) Availability Factor (3) Output Factor

232 H. PROCACCIA

3.3.2. Classification of shut-down duration. A table automatically classifies the shut-down duration which is divided into five ranges:

duration: 0 hr 0-2 hr 2-12 hr 12-72 hr duration > 72 hr

A print-out comprised of any given time period can be obtained.

3.2.3. Incident print-out. A program can retrieve all the incidents stored in the bank. Such a program is selective and allows a certain number of choices. These choices are:

A group of power units defined by the user (Vendor, Type of Reactor, Power) A sub-group of the plant (one of the 15 sub-groups mentioned previously) In one of the sub-groups a functional system can be chosen (i.e. RCP) One of the 37 components can be chosen

3.2.4. Table of incidents and shutdowns. Three sum- mary tables are obtained systematically for a given group of units (Vendor, Type of Reactor, Power) and for any period since the establishment of the bank: (1) table with total number of incidents regardless of

load change (2) table with total number of incidents with shut-

down of reactor (3) table with number of hours of shutdown

The tables comprise 15 columns and 37 lines (15 sub-groups and 37 components). There are too many systems to be able to have one table for each of them. A question concerning a functional system will have its answer in the above mentioned print-out.

3.3. Outputs available

Two kinds of results are available from this bank:

(1) Specific reports. (2) Periodical reports among which there is the RDE

concerning the French nuclear power plants.

3.3.1. Specific reports. Presently some specific studies were made on the primary pumps, the valves and their control, the main turbine, the steam gen- erators, the operating transients. All these studies were made from an equipment availability point of view, that is to say: the number of incidents per equipment per year, the average shutdown period, the unavail-

ability of the power plant due to an equipment failure.

3.3.2. Periodic reports. These reports are mainly:

The annual summary reports The reports based on different experiences since the establishment of the bank. The presentation of these reports is similar.

For example, some of the main results which have been obtained from five years of American power plant operations are listed below:

50% of the units have an availability factor higher than 74% 50% of the power units ( 7 0 0 < P < 9 0 0 M W e ) have an availability factor higher than 70% Excluding fuel reloading, 50~ of the power units have a production factor higher than 82% Excluding fuel reloading, 50~ of the power units ( 7 0 0 < P < 9 0 0 M W e ) have a production factor higher than 77 ~o There is no significant change in the power plant availability factor in terms of age. The average is 77~ for all power plants, and 70~o for plants with a rating between 700 and 900 MWe.

Including fuel reloading the results are as follows:

50~o of the power units have a production factor higher than 64 % 50~o of the power units (700-900 MWe) have a production factor higher than 58%.

Excluding fuel reloading, 50% of the power units have a production factor higher than 75~o Excluding fuel reloading, 50~ of the power units (700-900 MWe) have a production factor higher than 70~o.

There is no significant change in the production factor in terms of age for the power plants between 700-900 MWE. The average is 62%. For all power units we have:

Years All units U.S. 700-900 MWe 1 63% 62Vo 2 68% 67~o 3 72% 70~ 4 75~o 74~ 5 78% 77%

We can notice that there is no significant difference between all the power units and the power units which are in the 700-900 MWe range.

The heat exchangers induce most of the number of shutdowns (20.7%), then come the valves and their control systems (19.1%).

PWR plant operation and incidents 233

The heat exchangers induce the most important part of the shutdown periods (14.8%) and this is mainly due to the steam generator (11.1 Yo) followed by the pumps (8.5 ~o). The fuel reloading, the maintenance, the inspection all combined are responsible for 45 y,,, of the power unit unavailability.

3.4. Uses of RDE to date

The advantage of such a bank with its limitations is generally evident. The French power plants (PWRs) have now reached their commercial operation phase and all the information related to their operating conditions is now gathered and compiled. While waiting to get some significant statistical results regarding these nuclear plants, the results concerning

the American nuclear plants have been analysed under different aspects by several EDF teams. Thus, for example: (1) The General Economic Surveys Department did

some analysis in order to obtain concrete inform- ation which will enable us to estimate the operat- ing costs and to evaluate as accurately as possible the expected performance of the French nuclear plants.

(2) The Equipment Department was in a position to specify the design of some critical equipment by estimating the number of transients to which the equipment might be subjected during its lifetime.

(3) The Thermal Production Department is in a position to optimize the operation and the main- tenance of the equipment (spare equipment and recurrence frequency of maintenance).