HARMONISATION OF FAULT AND INTERRUPTION DATA IN THE NORDIC COUNTRIES Jørn Heggset, Olve Mogstad SINTEF Energy Research jorn.heggset@sintef.no olve.mogstad@sintef.no Morten Møller Jensen DEFU mmj@defu.dk Matz Tapper Svensk Energi matz.tapper@svenskenergi.se Jenny Edfast SwedPower jenny.edfast@swedpower.com Morten Lossius Agder Energi Nett morten.lossius@ae.no Rune Kristian Mork Statnett rune.mork@statnett.no Abstract The paper presents a work that is being carried out in a common Nordic research project called OPAL. The aim of the work is to establish common guidelines in Scandinavia for fault reporting, and to prepare the foundation for a Nordic fault database for voltage levels above 1 kv. The Nordel guidelines for reporting of disturbances above 100 kv are the most important basis for this work. The paper also discusses how users of a future database can be given access to the fault reports through the internet. The long-term goal of utilising component information in addition to the corresponding fault data through an internet database is also pointed out. Introduction Collection and application of fault and interruption data in the Nordic power system is done with different guidelines for each country. For voltage levels above 100 kv there are some harmonisation through NORDEL, but for the MV level there is no common standards. In order to give the network companies (and others) a better decision basis for investments, operation, maintenance and renewal of the networks a common Nordic project called OPAL (Optimisation of reliability in power networks) was launched in 2002. The objectives of the project are: Development of common Nordic guidelines for collecting and reporting fault data Contribute to a better utilisation of fault and interruption data Development of methods for calculation of reliability of supply in power networks One of the activities in OPAL is the specification of a common database for fault and interruptions. Considerable challenges for this work are the differences in detailing level between the countries and to a certain degree variations in the collected information. Similarities and differences between the countries were reported in [1], and together with the Nordel guidelines for classification of disturbances [3] this reference is the most relevant basis for the further work. 1
The aim is that a Nordic database in the future will be available for the data contributors (and possibly others) through the Internet. In the paper we point out the differences and similarities in today's guidelines and systems, describe the first version of common Nordic guidelines, show plans for the development of a database that will be accessible through the Internet, and discuss various problems, challenges and possibilities in this context. Organisation of national statistics in Scandinavia There is different focus for the statistics in the Scandinavian countries. Norway has a focus on energy not supplied and corresponding costs, due to the CENS arrangement 1. Denmark has the best fault statistics on component level. Common for Norway, Denmark and Finland is that interruptions are reported at the MV/LV transformation level or at a HV connected customer, while they in Sweden are reported on customer level. In Finland all interruptions are reported, in Denmark interruptions >1 minute are reported while in Sweden and Norway interruptions >3 minutes are reported to the authorities. In Finland there are little focus on a national statistics for faults and causes for component failures. In Denmark all reporting is voluntary and in Norway only fault statistics for MV distribution is voluntary. In Sweden and Finland a limited reporting of interruptions is compulsory and fault registration is voluntary. Different tools and systems are used for the registration of faults and interruptions: Denmark A special programme developed by DEFU (EL-FAS) is used by all network companies. Finland The registration is manual for the small network companies. The bigger companies use tools within different NIS or SCADA-systems. Norway A special programme (FASIT) is used by all network companies. A common specification for FASIT is developed by SINTEF and owned by EBL Kompetanse, but with different suppliers delivering the programme as a stand-alone system or integrated in NIS or SCADA-systems. Sweden The registration is manual for the small network companies. The bigger companies use tools within different NIS or SCADA-systems. The format for the registration and interchange of data is specified by DARWIN which is the system used by Svensk Energi. The registration details vary a lot between the countries. In Table 1 some of the registrations are compared. 1 The Norwegian regulator (NVE) has from 2001 introduced quality dependent revenue caps (the CENS arrangement), a model for incentive based regulation of supply quality. The network companies revenue caps depend on the development of the quality of supply in such a way that the companies will be better off by balancing their internal costs and quality of supply towards a socio-economic optimum. 2
Table 1: Comparison of the fault data collection in the Nordic countries Registration Denmark Finland Norway Sweden Primary cause 31 choices 16 choices 8 main levels with 5 23 choices 13 secondary choices Secondary cause 28 choices Not used 2 x 8 main levels, Not used with 5 13 secondary choices Fault type 24 choices 4 main level 26 choices Not used Component with fault Choice in list with type, place and size Repair Registration of 7 repair types Choice in list with type, place and size etc. Nothing, only planned outages Detailed about which component, little about type and dimension No details, but calculation of repair time Choice in list with localization and voltage Nothing The Nordel guidelines The Grid Disturbance Group, organised under Nordel s Operations Committee, has developed guidelines for classification and calculation of disturbances, faults and energy not supplied (ENS) for the statistics published by Nordel [3]. These guidelines are first of all accommodated for transmission units with a system voltage of at least 100 kv, including units for reactive compensation. The purpose of the grid disturbance statistics is to compile data that can form the basis of: A correct assessment of the quality and function of the different components A calculation or assessment of the reliability of the transmission system An assessment of the availability of delivery points Studies of trends and comparisons of different parts of the Nordel grid The members of Nordel (Denmark, Finland, Iceland, Norway and Sweden) report grid disturbance data annually for the purposes mentioned above. For reliable statistics and possibility for comparison between the countries, it is important that definitions related to disturbances and methods for calculations are coordinated. If so, the quality of the data basis and the reliability of the statistics will increase. In this context the Nordel guidelines ensure that the countries are using the same basis for their reliability data collection, and this has been an obvious starting point for the work done in OPAL. Development of common guidelines for collecting and reporting of fault While Nordel gives common guidelines for reporting disturbances in transmission grids with a voltage of at least 100 kv, research has proven [1] that there are many differences in the way faults and interruptions in the distribution (MV) grids are reported. This leads to problems when trying to use statistical data from more than one country in various analyses. 3
The existing fault information about MV networks is in many cases not precise enough for e.g. reliability analyses. To increase the number of components in the statistical material it seems like a good idea to develop a set of common guidelines for collecting and reporting of fault data for the Nordic countries. The countries have different, but reliable, systems for the reporting of customer interruptions. An accurate fault statistic is published in Norway and Denmark only, but the collected data may prove insufficient there as well. The Danish and Norwegian guidelines are to some extend based on the Nordel guidelines, which consequently leads to a lot of similarities in the basic principles for registration. However, there are many differences at the detailed level and in the interpretation, which makes it difficult or even impossible to compare the data sets. It is a problem that the registration forms do not ask similar questions or give the same answering alternatives. Therefore, a new set of guidelines should be developed. These guidelines will be based on the basic principles of the Nordel guidelines, but there will be some changes and additions to reflect e.g. the different network structures used at the MV level in the countries. An important aim of the project is to get a far better basis for decisions concerning maintenance and investments in the grid. This makes it important to collect detailed information about e.g.: Manufacturer and type of the faulty component. Age of the faulty component The environment in which the faulty component was located The cause of the fault Not all of this information is available in today's statistics. Thus, it is important to develop a set of guidelines that facilitates this type of registration. Furthermore, it is important to ensure that all faults are registered. Due to the requirements in the present national and Nordel guidelines it is in practice only faults that have resulted in disturbances that are reported. This makes it difficult to attain reliable information about the fault rates of different components. It is therefore important to include component faults even if they do not cause disturbances, i.e. faults detected by inspections, condition monitoring, etc. To get a comparable set of data it is important that interpretation of the guidelines differ as little as possible between countries and between companies. This can be very difficult to achieve, but work has to be done defining clear interpretations, especially in the borderline cases. In order to motivate the companies to follow the guidelines something should be given in return. The opportunity to perform self-defined searches in a common database to get reliability data about components present in their own network should be a strong motivation for the companies to comply with the guidelines. It is therefore desirable to establish a database that facilitates different user queries. The guidelines that are being developed in the OPAL project are mainly focused on the MV network (1-100 kv), since there is no Nordic coordination on these voltage levels today. However, the reporting format will also be valid for the voltage levels above 100 kv. This part is entirely based on the Nordel guidelines and the corresponding fault statistics that are being 4
published annually. In addition it offers an opportunity to collect more information about the fault than the current Nordel guidelines. Explanatory variables Investigation of the explanatory variables for the occurrence of faults in power grids is also a part of the work in OPAL. To be able to make decisions on investments, renewal, operation and maintenance strategies it is important to have knowledge about the variables affecting the network and in what way the network is affected. First a broader separation of variables affecting the network can be made. Four (4) different variables have been identified: Climate. The weather conditions in regions are often correlated with the faults. Voltage level. The type of equipment differs to a certain degree between the voltage levels. Network structure. The consequences of faults depend on the network structure (e.g. radial network). Method of earthing. The different types of earthing affect the behaviour of the network. A table with a categorisation of each network company that constitute the data base, in accordance with the systematic described above, will be linked to the fault reports in order to make it possible to choose comparable companies and network areas for comparative studies or simply to enlarge the sample. In addition, we see a very large benefit in utilising some descriptive information about each component in the population, in order to get more specific and detailed fault statistics in the future. This aspect is described in a later section. Web access to the database Today, companies report disturbance data for distribution grid to the national statistics, and they put a lot of effort into this work. The statistics that are being published are predefined in accordance with the most common data needs. In addition, the statistics are based on the entire population, and are usually presented on an aggregate level. This means that the statistics are not necessarily customised and adjusted for each analysis. In the worst case, results of the analyses may lead to incorrect decisions because of maladjusted data basis. Since it is not possible to present all requested statistics in the publications, another solution is to make the data available for the users to define statistics after their own requirements and needs. It is important to exploit the potential and possibilities that is in the database. The internet makes it possible to make data available for every potential user. Through a web site with functionality for searching into the database, a skilled user will be able to generate the statistics and parameters required for the analyses. This web site should for instance have functions for generating statistics based on whatever data that is representative for the requested reliability parameter. There will also be functionality for comparison between grid companies, 5
between networks with similar external conditions, between components of same age and type, etc. When data is available in this way it is expected to have many positive effects. This can be illustrated as follows: Better data availability More frequent use Increased understanding and benefit Developing better methods and routines Optimal investment, maintenance and operational strategy Figure 1 Effects of better data availability and more flexible search facilities The harmonizing of disturbance data from the Nordic countries will make it easier to collect a bigger data basis. As a consequence, the component statistics will be more reliable. When this is up and running, hopefully in not a too far future, the Nordic grid companies will have a very good foundation for making reliable component statistics as basis for operation, maintenance and investment analyses. We are convinced that there is a large potential for increased profit by using more reliable input parameters into the economic analyses. Utilisation of other data types The utilisation of fault statistics is very often limited due to lack of information about the component population. In particular, this is evident in analyses where we need estimates of future fault frequencies for one component or a group of comparable components, e.g. in LCC analyses and maintenance and renewal evaluations. Available fault statistics give limited possibilities for estimation of life distributions for components. Some of the main challenges are: Insufficient information about the age of the components. No information about the components that have not failed. The statistics are presented on an aggregated level (thus, only valid for repairable systems). Insufficient classification of component type, manufacturer, etc. There is an underlying assumption that all failures are independent. For life distribution analyses we need more detailed fault statistics than what is available today. One way of dealing with this problem is to link the fault reports with other data types, such as inventory data, maintenance data and information about technical condition (state). With such links the analysts will be able to get more specific statistics sorted on various explanatory 6
variables about the component, such as manufacturer, type, age, operational conditions, etc. If this information in addition could be standardised and collected into the Nordic database a large and valuable data basis could be established for very many planning and analysing purposes. The principles are illustrated in Figure 2. Maintenance... Maintenance 3 Failure.. Failure 2 Maintenance 2 Maintenance 1 Failure 1 Inventory.. Inventory 2 Inventory 1 State information Figure 2 Principle drawing of links between different data types as a basis for "producing" reliability data However, it is not realistic to establish such a reporting system on a short term, and the Nordic reporting format that is being specified in the OPAL project will not contain detailed inventory and maintenance data. Instead, some key information about the failed component will be included in the failure reports. But if we look beyond today's (or tomorrow's) solutions, a standardisation and linking between different data types as illustrated in Figure 2 will bring the potential for utilising reliability data a big step forward. Consequently, as a part of the research work carried out in OPAL we are trying to establish a foundation for a future standardisation of data and links between the different types. The need for a better data basis In this section we present an example on an analysis where fault data are used as a basis for investigating the quality of XLPE cables from 2 suppliers. The purpose with the example is to illustrate how a detailed fault statistics, with fault data linked with component data, can be used for investigating the lifetime of XLPE cables from different manufacturers. 7
It is a fictitious example about a network company that has first generation XLPE cables from two manufacturers, Producer A and Producer B. Experience data indicate that cables from Producer A have a higher failure rate than cables from Producer B. The company wants to replace the cables from the poorest age groups first, but finds it important to postpone the reinvestments as long as it is economical feasible. As basis for these decisions fault statistics for the two cable types are applied. The following information is needed: Number of faults for each cable type for each year within the data collection period Cable length distributed on manufacturer and age group, for each year in the data collection period (it is important to keep track of the length of each age group for each year, because some cables may have been replaced within the data collection period) Failure causes The year each failure has occurred Information about the failure cause is necessary in order to select only the relevant faults, in this case material related faults. The company realises that their own data basis is too small for this kind of analysis, hence they supplements the analysis with data from the (not yet realised...) Nordic fault database. The failure frequencies for the two cable types are shown in Figure 3 and 4. (Notice the difference in scales in the two figures.) Fejl på PEX-kabler af fabrikat A af forskellig årgang 10 23,5 9 8 Antal fejl pr. 100 km pr. å 7 6 5 4 3 2 1 0 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 Årgang Figure 3 Failure frequency for XLPE cables from Producer A. It should be noted that the numbers from 1977 and 1987 in Figure 3 are based on very limited cable length, hence the corresponding fault frequencies should be considered with some reservation. 8
Fejl på grafiterede PEX-kabler af fabrikat B af forskellig årgang 5 4,5 4 Antal fejl pr. 100 km pr. å 3,5 3 2,5 2 1,5 1 0,5 0 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 Årgang Figure 4 Failure frequency for XLPE cables from Producer B. It can be seen from Figure 3 and 4 that there is a quite noticeable difference in failure frequency between the two cable types. Especially the first three age groups from Producer A have considerable higher values than the rest of the population. It is well worth noticing that Producer B changed their production method in 1977-78, something that resulted in considerable fewer faults caused by water trees. In Figure 5 the failure frequency as a function of operational time is illustrated. Fejl på PEX-kabler som funktion af driftstiden Producent B Producent A 7 6 5 Antal fejl pr. 100 km 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Antal år i drift Figure 5 Failure frequency for XLPE cables vs. operational time (years) 9
From Fig ure 5 it can be seen that there is a considerable increase in failure frequency for cables from Producer A after 10 years of operation, and these age groups are obvious candidates for replacements. However, a cost-benefit calculation, with the failure probability being one of the elements, should be performed in order to decide the optimal replacement time. This example illustrates quite clearly the usefulness of a good failure statistics, where one is able to combine the fault reports with component information. In this case the information in question is the manufacturer, type and age of XLPE cables. Conclusions The paper has presented the work being done in the OPAL project where the aim is to establish common guidelines for fault reporting in the Nordic MV and HV network, based on the Nordel guidelines for voltage levels above 100 kv. As the work in the project is still in progress these conclusions should be considered as preliminary. However, some main points may be pointed out. We believe that there is a need for a larger and better data basis for decisions within the grid companies, as both the owners and the authorities are focusing on network efficiency and cost reductions. To cope with these challenges the companies have to justify the money spent on investments, reinvestments, maintenance and operation. In this context a solid fault data basis with user access through the internet will be essential. However, due to several circumstances this is not easy. The main challenge is probably the differences in today's registration schemes and the traditions between the countries. Changes in (to a certain degree) well-functioning systems and routines should not be done without very good reasons. We believe that some of the differences may be eliminated through a national data adaptation before the data are entered into the Nordic database. However, some changes in the national systems and guidelines must be reckoned on. It is very important to utilise information about the components and their environment as a supplement to the fault reports. We suggest a step-wise approach with the inclusion of some key information in the first phase, but with the long-term goal of having flexible links between several data sources, such as fault data, component information and maintenance data. This is without doubt a challenging, but indeed an important process. References [1] Mogstad O, Jensen M M, Eklund T, Rehngren L, Tapper M: "Collection and reporting of fault and interruptions in the Nordic countries. Requirements, organisation and content." (in Scandinavian) Trondheim: SINTEF Energy Research, April 2003 (EBL-K 115-2003) [2] Various memos from the OPAL project (in Scandinavian). See web page: http://www.energy.sintef.no/prosjekt/opal [3] Nordels riktlinjer för klassificering driftstörningar (in Swedish) 10