Draft Guide for the Interpretation of Gases in Oil Immersed Transformers

Transcription

1 Draft Guide for the Interpretation of Gases in Oil Immersed Transformers Sponsored by the Transformers Committee of the IEEE Power Engineering Society Copyright 2004 by the Institute of Electrical and Electronics Engineers, Inc. Three Park Avenue New York, New York , USA All rights reserved. This document is an unapproved draft of a proposed IEEE-SA Standard USE AT YOUR OWN RISK. As such, this document is subject to change. Permission is hereby granted for IEEE Standards Committee participants to reproduce this document for purposes of IEEE standardization activities only. Prior to submitting this document to another standard development organization for standardization activities, permission must first be obtained from the Manager, Standards Licensing and Contracts, IEEE Standards Activities Department. Other entities seeking permission to reproduce portions of this document must obtain the appropriate license from the Manager, Standards Licensing and Contracts, IEEE Standards Activities Department. The IEEE is the sole entity that may authorize the use of IEEE owned trademarks, certification marks, or other designations that may indicate compliance with the materials contained herein. IEEE Standards Activities Department Standards Licensing and Contracts 445 Hoes Lane, P.O. Box 1331 Piscataway, NJ , USA This is an unapproved IEEE Standards Draft, subject to change. 1

2 Introduction (This introduction is not part of PC57.104, Draft Guide for the Interpretation of Gases in Oil Immersed Transformers. ) This revised IEEE guide for the interpretation of gases generated in operating oil-immersed transformers presents to the operators and manufacturers of oil filled transformers an improved routine surveillance oil sampling program and procedures for detecting the presence of combustible gases in the equipment during service that indicate faults which if not detected in the very early or incipient stages may eventually lead to failure of the transformer. The detection methods presented here frequently provide the first available indications of a malfunction. They employ consensus guidelines that allow operators of transformers of any age which are absent any significant gas data or operators of transformers with established gas databases to ascertain by surveillance sampling whether the transformer is generating normal quantities of gas, increasingly abnormal quantities in the caution range, or dangerous quantities in the warning range. A procedure is given for the statistical development of surveillance norms from a gas database and a detection procedure is given for transformers that employ norms from units with identical design and service conditions. Sampling intervals and operating guidelines are suggested for each range and generating rate. A Key gas screening procedure is given to permit an estimate of a gas generating mechanism, called a fault, in the caution range between normal and warning levels. A ratio procedure is also provided for refined diagnoses in the warning range which provides empirical but reliable diagnoses of internal problems and the use of graphical trends is suggested to further refine the diagnoses. The revised guide suggests preference for the statistical development of detection norms jointly acceptable to user and manufacturer for units under warranty and between user and insurer for units beyond manufacturing warranty. These revisions provide the transformer user with an earlier and clearer picture of his operating options and shorten the time required to evaluate the current status and make more timely and appropriate operating decisions. This Guide was prepared by the Insulating Fluids Subcommittee of the IEEE Power Engineering Society. At the time this Guide was completed, the Working Group had the following membership: F.W Heinrichs, Chair J. Corkran A. Darwin J. Goudie T. Haupert F. Jakob J. Kelly D. Kim R. Ladroga J. Lackey S. Lindgren T. Lundquist S. McNelly P. McShane T.V. Oommen T. Prevost T. Rouse J. Smith The following persons were members of the balloting group that approved this document for submission to the IEEE Standards board as a recommended practice. This is an unapproved IEEE Standards Draft, subject to change. ii

3 Contents Introduction... ii 1. Overview Scope Limitations References Definitions General decomposition theory Cellulose decomposition Mineral oil decomposition Combustible gas generation in equipment Thermal fault Thermal faults involving cellulose Electrical faults Sampling and laboratory procedures Variability and precision of analysis Detection and operating guidelines Detection procedure for transformers in Type 1 category Detection procedure for Type 2 transformers Diagnosis of transformer faults Analysis in the caution surveillance range Diagnosis in the warning surveillance range Diagnosis of cellulose decomposition Trend and graphical analysis Construction of a 3 D graph Detection and analysis of gases in gas space, relay or other fixed or portable devices...18 Annex A (Normative) TCG detection and operating procedure...19 This is an unapproved IEEE Standards Draft, subject to change. iii

4 Annex B (Normative) TCG diagnosis of fault type...20 This is an unapproved IEEE Standards Draft, subject to change. iv

5 Draft Guide for the Interpretation of Gases Generated in Oil Immersed Transformers 1. Overview The detection of certain gases generated in an operating mineral oil-filled transformer in service is frequently the first available indication of abnormalities called faults that may eventually lead to failure if not corrected. Electrical faults such as arcing, low energy sparking and partial discharge (previously called corona) are described. Thermal faults such as severe overloading, cooling pump failure, and overheating in the insulation system are described. These conditions occurring singly, or as several events can result in decomposition of the insulating materials and the formation of various combustible gases dissolved in the insulating oil, in the inert gas space above the oil, or in gas collecting devices. The procedure for detection of faults for transformers of any age or design but with no significant gas history, defined as Type 1 transformers, is given in clause 6.1. Detection procedures for transformers that have been operating long enough to have a gas history or database, defined as Type 2 transformers, are given in 6.2. Detection in Type 1 or Type 2 transformers begins with regular surveillance sampling and analysis to determine whether the gas concentrations have reached normal, caution, or warning surveillance ranges. Guidelines for typical gas concentrations in the normal, caution and warning surveillance ranges for type 1 transformers are offered in Table 1. For Type 2 transformers where sufficient data exists, statistically derived norms are preferred. They should ultimately be agreed upon by the manufacturer and user in transformers under warranty or by manufacturer, insurer, or transformer operators beyond warranty as appropriate. The increasing order of the surveillance ranges reflect fault severity. Surveillance involves regular oil sampling or other testing and monitoring procedures, which are subject to errors. The norms found in Table 1 for Type 1 transformers are current reported values based on industry experience. Special applications and service requirements may be considered in establishing the initial sampling interval. For instance, operators of small transformers in very critical applications may choose shorter sampling intervals than they would for larger units in a less demanding application. However, the guide suggests a preference for the statistical development of norms as soon as sufficient data is available. In any case, detection and the resulting adjustment of sampling interval and operating procedure requires an evaluation of the quantities of generated gases present and determination of their rates of generation. A Key gas diagnostic screening procedure is given for use in the caution range. A Ratio procedure for diagnosis in the warning range is given to indicate the possible fault mechanism and it s source. These procedures may reveal the source of the disturbance based on an empirical relationship between generated combustible gas species and fault types. Trends and graphical analysis may further refine diagnoses. It should be noted that normal operation of the transformer might also result in the formation of small quantities of combustible gases. In fact, it is possible for some transformers to operate throughout their useful life with even larger quantities of combustible gases present although it is not a normal occurrence. Proper application of the step-by-step procedures described in this guide can provide the operator with the earliest indication of a developing problem and positive useful information on the serviceability of his equipment. 1.1 Scope This guide applies to operating mineral oil immersed transformers that are newly energized and have no previous combustible gas history; and to operating mineral oil immersed transformers that have been operating long enough to develop a combustible gas database. The guide addresses: General Theory, Sampling and Laboratory Procedures, Variability and Precision Detection, Operating Guidelines, This is an unapproved IEEE Standards Draft, subject to change. 5

6 Development of Norms, Diagnostic Techniques, Fixed and Portable Monitoring Devices, Normative Annex. 1.2 Limitations The scientific principles of hydrocarbon and cellulose decomposition and the analytical procedures described in this guide are well established. However, application of these principles to the determination of faults in transformers was established by empirical studies relating measured combustible gas concentrations with observed disturbances and failures. The empirical nature of these studies confers a degree of variability and uncertainty that must be recognized particularly in numerical fault diagnosis. The variables involved are equipment type, the materials involved in the fault, and the extraction and analytical procedures themselves. Equipment variables are transformer design, location, service and operating temperatures. Material variables are solubility related to oil, degree of saturation and the types of materials involved at the fault. Analytical variables associated with the sampling, extraction and measuring process affect to some degree the confidence in diagnostic procedures especially between laboratories. Also, gas concentrations below 10 times the minimum detection level may invalidate Ratio (numerical) diagnosis. Replicate samples are recommended before developing an operating decision or fault diagnosis. Because of this inherent variability, it is difficult to obtain a consensus on guidelines for detection norms or diagnostic Ratios. Exact fault types or degrees of fault intensity may not be inferred from a single gas concentration, or even after repeated sampling. Rather, development of concentration trends will give more reliable insight into developing faults. In addition, diagnosis of the primary or failure-initiating fault from a sample taken after failure is invalid because of the saturation quantities of all gas species that are generated by the heat released from the transformer failure itself. In the light of these variables, this guide is offered as an advisory document, providing guidelines to assist the transformer operator and manufacturer in deciding on the status and continued operation of a transformer that exhibits combustible gas formation. This guide applies only to operating transformers. It does not include transformers where arcing switch contacts or expulsion tubes are exposed to the main oil volume. No attempt should be made to relate these guidelines to new transformers during factory temperature rise tests. Finally, the operators development of a combustible gas database accompanied by actual confirmation of predicted fault by failure investigation will allow each individual operator to establish his own norms unique to his size and type of transformer and operating conditions. 2. References This guide shall be used in conjunction with the following standards. When the following standards are superseded by an approved version, the revision shall apply. ASTM D (2003) e1, Standard Test Method for Gas Content of Insulating Oils 1 ASTM D (1999) e1, Standard Practice for Sampling Small Gas Volume in a Transformer ASTM D , Standard Test Method for Analysis of Gases Dissolved in Electrical Insulating Oil by Gas Chromatography ASTM D , Standard Practice for Sampling Insulating Liquids for Gas Analysis and Determination of Water Content 1 ASTM publications are available from the American Society for Testing and Materials, 100 Barr Harbor Drive, West Conshohocken, PA , USA ( This is an unapproved IEEE Standards Draft, subject to change. 6

7 ASTM D , Standard Test Method for Combustible Gases in the Gas Space of Electrical Apparatus Using Portable Meters ASTM D , Standard Practice for Sampling Gas from a Transformer under Positive Pressure IEC Mineral oil impregnated electrical equipment in service- Guide to the interpretation of dissolved and free gases analysis 2 IEEE Std C , Terminology for Power and Distribution Transformers 3 3. Definitions 3.1 Caution surveillance range: Combustible gas concentrations above the normal surveillance range and below the warning surveillance range. 3.2 Dissolved gas analysis (DGA): The extraction, detection and analysis of gases dissolved in oil. 3.3 Fault diagnosis: A numerical diagnostic procedure using Rogers ratios to determine specific transformer fault types for combustible gas concentrations at or above the warning surveillance range. 3.4 Key gas: A characteristic combustible gas generated in oil filled transformers can be used for qualitative determination of fault types based on gases, which are typical or predominant at various temperatures. 3.5 Minimum detection level (MDL): The lowest level at which a gas can be detected in oil; described in ASTM D n% Probability norm: A statistically determined value, which signifies the probability that n% of all normally operating transformers have individual or total combustible gas concentrations at or below that value. 3.7 Normal surveillance range: A combustible gas concentration level at or below which individual or total combustible gas concentrations in transformers are considered normal. 3.8 Partial discharge (PD): An electric discharge which only partially bridges the Insulation between conductors, and which may or may not occur adjacent to a conductor. Depending on intensity, Partial discharges are often accompanied by emission of light, heat, or sound, separately or in combination. The term Corona has also been used to describe partial discharges. This is a non-preferred term since it has other non-related meanings. 3.9 Screening diagnosis: A diagnostic procedure using Key gases to obtain a preliminary estimate of lowlevel incipient fault development in the caution surveillance range. 2 IEC publications are available from the Sales Department of the International Electrotechnical Commission, Case Postale 131, 3, rue devarembé, CH-1211, Genève 20, Switzerland/Suisse ( IEC publications are also available in the United States from the Sales Department, American National Standards Institute, 11 West 42 nd Street, 13 th Floor, New York, NY 10036, USA ( 3 IEEE documents in the reference list are available from the Institute of Electrical and Electronics Engineers, Service Center, 445 Hoes Lane, P.O. Box 1331, Piscataway, NJ , USA. This is an unapproved IEEE Standards Draft, subject to change. 7

8 3.10 Total combustible gas (TCG): The sum (%) of all combustible gases including carbon monoxide and excluding oxygen reported as a % of the transformer gas space Total dissolved combustible gas (TDCG): The sum of all combustible gases that are dissolved in the insulating oil Type 1 Service category: Type 1 Transformers have never received DGA before or do not have significant gas databases Type 2 Service category: Type 2 Transformers have had previous DGA tests or have a gas database large enough for statistical derivation of surveillance norms. Type 2 transformers may also have been assigned surveillance norms because of their similarity to other Type 2 transformers in design and loading Warning surveillance range: Combustible gas concentrations above the Caution surveillance range. Combustible gas concentrations within this range generally result in failure of the transformer and should result in immediate investigation. 4. General decomposition theory The principal causes of gas formation within an operating transformer are thermal disturbances such as conductor losses and electrical disturbances and other temperature producing effects that produce combustible gases from mineral oil, and cellulosic decomposition. Generally, there is little or no heat associated with low energy partial discharges where decomposition gases are formed principally by lowlevel ionization. Mineral oil decomposition is the predominant mechanism utilized in the detection and diagnostic procedures applied to transformers. 4.1 Cellulose decomposition The thermal decomposition of oil-impregnated cellulose insulation produces carbon oxides (CO, CO 2 ) and some hydrogen or methane (H 2, CH 4 ) from the oil (note: CO 2 is not a combustible gas). The rate at which they are produced depends exponentially on the temperature and directly on the volume of material at that temperature. A large, heated volume of insulation at moderate temperature may produce about the same quantity of gas as a smaller volume at a higher temperature. 4.2 Mineral oil decomposition Mineral transformer oils are mixtures of many different hydrocarbon molecules, and the decomposition processes for these hydrocarbons in thermal or electrical faults are complex. The fundamental steps are the breaking of carbon-hydrogen and carbon-carbon bonds by the heat evolved from the fault mechanism. Active hydrogen atoms and hydrocarbon fragments are formed. These free radicals can combine with each other to form gases, molecular hydrogen, methane, ethane, etc., or can recombine to form new, condensable molecules. Further decomposition and rearrangement processes lead to the formation of products such as ethylene and acetylene and, in the extreme, to modestly hydrogenated carbon in particulate form. These processes are dependent on the presence of individual hydrocarbons, on the distribution of energy and temperature in the neighborhood of the disturbance, and on the time during which the oil is thermally or electrically stressed. These are chemical transformations; therefore, the specific degradations of the transformer oil hydrocarbon ensembles and the fault conditions may not be predicted reliably from purely chemical kinetic considerations. An alternative approach is to assume that all hydrocarbons in the oil are decomposed into predictable products and that each product is in equilibrium with all the others. In addition, the presence of certain metals and coatings can have a catalytic effect on mineral oil decomposition. It is well known that combustible gases are formed by the thermal decomposition of mineral oil. And the proportions of certain combustible gases are unique for a specific temperature of decomposition. These relationships, coupled with evidence that thermal faults in a transformer generally produce lower temperatures in the oil than electrical or arcing faults are the basis for the empirical This is an unapproved IEEE Standards Draft, subject to change. 8

9 development of the fault diagnostic procedures given in this guide. Some examples of these relationships are: the presence of methane suggesting a relatively low energy discharge or thermal fault; and the presence of acetylene suggesting that a high-energy arc has occurred. The thermodynamic approach has limits; it must assume an idealized but nonexistent isothermal equilibrium in the region of a fault, and there is no provision for dealing with multiple faults in a transformer. Much work has been done to correlate these predictions from thermodynamic models with the actual behavior of transformers. 5.0 Combustible gas generation in equipment All transformers generate gases to some extent at normal operating temperatures. But occasionally gasgenerating abnormalities known as faults occur within an operating transformer such as a local or general overheating, electrical problems, or a combination of these. These types of faults produce characteristic gases that are generally combustible. Their concentrations are determined by extraction and Gas Chromatography Detection of combustible gases dissolved in the oil or found in the inert gas space or relay may indicate the existence of any one, or a combination of thermal or electrical faults. Furthermore, the ratios of certain gases have been found to suggest fault types using the diagnostic procedures in 7.0. Interpretation by the individual gases can become difficult when there is more than one fault, or when one type of fault progresses to another type, such as an electrical problem developing from a thermal one or vice versa. Analytical laboratories should report the MDL for their particular extraction procedures and equipment. Gas concentrations greater than ten times MDL are preferred for reliable numerical ratio diagnoses. Finally, operators should avoid attempts to assign greater significance to gas measurements than justified by the natural variability of the generating and analytical processes. While new logical procedures may reduce diagnostic uncertainty, increased sampling and observation of gas generating trends should also enhance the correlation between Fault and numerical Diagnosis. Carefully following the rules and procedures given here will provide valuable assistance to the operator and manufacturer. 5.1 Thermal fault At oil temperatures from 150ºC to 300ºC relatively large quantities of the low molecular weight gases, such as hydrogen (H 2 ) and methane (CH 4 ), and trace quantities of the higher molecular weight gases ethylene (C 2 H 4 ) and ethane (C 2 H 6 ) are produced by poor cooling or stray losses in winding or leads; between core laminations or in core, tank or supporting structures. As the decomposition temperature in mineral oil increases from 300ºC to 700ºC, the hydrogen concentration exceeds that of methane and is accompanied by significant quantities of higher molecular weight gases, first ethane and then ethylene. Beyond 700ºC (the upper end of the thermal fault range), increasing quantities of hydrogen and ethylene and traces of acetylene (C 2 H 2 ) may be produced. In general, non-electrical thermal faults produce temperatures below 700ºC; however, welding on oil filled equipment or oily surfaces produces acetylene due to the very high temperature. Also, thermal faults may evolve into electrical faults, which may be a source of error in diagnosis. 5.2 Thermal faults involving cellulose The thermal decomposition of cellulose paper and solid insulation produces mostly carbon monoxide (CO), carbon dioxide (CO 2 ), and water even at normal operating temperatures. Also, moisture generated by cellulose decomposition will accelerate further decomposition especially in sealed units. Gaseous byproducts of cellulose decomposition are found at normal operating temperatures in the transformer. Diagnosis of thermal faults involving cellulose is given in this guide. 5.3 Electrical faults Electrical faults range in energy and temperature from intermittent low energy partial discharges to steady discharges of high energy (arcing). As the discharge progresses from low energy to higher energy, the acetylene and ethylene concentrations rise significantly This is an unapproved IEEE Standards Draft, subject to change. 9

10 5.3.1 Partial discharges (PD) The very low temperature of partial discharges (150ºC to 300ºC) produces mainly hydrogen, with lesser quantities of methane and trace quantities of acetylene. Note that this temperature range also produces some thermal decomposition of oil and cellulose. The gases produced at these low temperatures may suggest either an electrical or thermal fault or both Low and high energy discharge (arcing) As the intensity of the electrical discharge reaches arcing or continuing discharge proportions, producing temperatures from 700ºC to 1800ºC, the quantity of acetylene becomes pronounced. Note that these temperatures are also produced by welding, complicating diagnosis. 5.4 Sampling and laboratory procedures Establishing a reference point (baseline) for statistical norms for gas concentrations in new and repaired transformers in service, and developing a routine surveillance sampling program are critical elements in the application of this guide. ASTM Standard methods are utilized to obtain oil or gas samples for extraction and analysis by gas chromatography. Monitoring devices are available which directly and continuously or periodically extract and measure combustible gas components Surveillance sampling Regular surveillance monitoring of combustible gas concentrations in operating transformers can begin anytime. Generally, daily or weekly sampling is recommended after start -up followed by monthly or longer intervals, which may vary depending on application and individual system requirements. After detection of increasing amounts of generated gas, shorter sampling intervals are suggested. When a possible source of gas generation has been determined, good engineering judgment should be applied to determine sampling interval and operating procedure Sampling and analytical method All samples of oil from electrical apparatus for DGA should be taken in accordance with ASTM D3613. TCG samples from the gas space above the oil may be taken in accordance with ASTM D 2759 or D3305 if the gas volume is small. The combustible gases contained in oil samples for DGA are extracted from the oil and analyzed by gas chromatography per ASTM D3612. TCG samples of the gas from the inert gas space may be analyzed per ASTM D3612 or equivalent industry standards. (See also Annex A, B). The syringe or cylinder method is preferred for all oil or gas samples. 5.5 Variability and precision of analysis Minimum detectable levels of gas concentration measurements for gas species dissolved in oil are given in ASTM D3612. Gas concentrations near MDL have a high probability of large error. Gas concentrations at ten times MDL or greater have typical inter-lab precision of approximately 5% and typical intra -lab precision about 10% which should be recognized when establishing a DGA program. Equal intralaboratory precision is recommended when analysis is performed by more than one laboratory. Wide variations in gas concentrations within a single sample are cause for concern. 6. Detection and operating guidelines Much information has been acquired on the utilization of combustible gas data from surveillance oil samples for detecting and determining incipient fault conditions in operating transformers. This guide seeks This is an unapproved IEEE Standards Draft, subject to change. 10

11 to assist the operator in applying this gas data to Type 1 transformers per clause 6.1 and Type 2 transformers per clause 6.2 following a step-by-step process described in the following Flow Chart Figure 1. FIGURE 1 Flow chart MAINTENANCE Derive 90% Probability Norms Sample Compare to Database Norms Sample Compare to Range Guidelines Table 1, 6.1 Sample Compare to Assigned Norms Normal Range Table 1 RANGE DETERMINED Key Gas Procedure Table 3, Caution Range Table 1 Warning Range Table 1 Key Gas Procedure Table 3, Generation Rate Determine Sampling Interval and Operating Procedure Table 2, Ratio Diagnosis Table 4, Trend & Graphics 8.0 Database ACTION & OPERATING DECISION The Flow Chart process begins with taking oil samples containing dissolved gases from Type 1 or Type 2 transformers during routine maintenance surveillance. The samples are sent to the laboratory where dissolved gases are extracted and analyzed. The gas concentrations from Type 1 transformers are compared per clause 6.1 to the guidelines in Table 1 to determine their surveillance range. Alternatively, the concentrations from Type 2 transformers are compared per clause 6.2 to statistically derived norms or assigned norms from similar units and their corresponding surveillance range is determined. Gas generation rates are determined per clause and the range and rate information is applied to Table 2 to determine the guidelines for new sampling intervals and operating procedures. This is an unapproved IEEE Standards Draft, subject to change. 11

12 6.1 Detection procedure for transformers in Type 1 category Referring to Flow Chart Fig. 1, the operators of Type 1 transformers only, for which no previous gas data exists, must initially rely on Table 1 for guidance that indicates whether their transformer is operating in the normal surveillance range or in the caution or warning surveillance ranges. The Table 1 guidelines used for this purpose were surveyed from many laboratories, manufacturers and operators representing industry practice for normally operating transformers. The Type 1 transformer is then assigned to the particular surveillance range within which either TDCG or any one or more component gas concentrations fall. For example, if TDCG is >700 mg/kg (ppm) or C 2 H 2 is 4 mg/kg (ppm), the transformer is in the caution surveillance range. TABLE 1 - Guidelines for surveillance range for Type 1 transformers having no previous combustible gas tests Generated Gases Hydrogen Methane Acetylene Ethylene Ethane Carbon Monoxide TOTAL a Surveillance range Normal mg/kg (ppm) H 2 CH 4 C 2 H 2 C 2 H 4 C 2 H 6 CO TDCG <100 <120 <2 <50 <65 <350 <700 Caution mg/kg (ppm) 100 to to to 5 50 to to to to 1900 Warning b mg/kg (ppm) >700 >400 >5 >100 >100 >570 >1900 a Total of all combustible gases b Any component or their total gas concentrations in warning range indicates a severe problem generally requiring immediate intervention or removal Combustible gas generation rate When any combustible gas is detected in any surveillance range, it is essential to determine whether the source is active or passive by taking a second sample St at T days later and computing the generation rate R s as a % per day of the first sample S o. As a rule of thumb, a generating rate exceeding 0.5% per day for any single component or their sum total suggests a possible problem; generation rates exceeding 3% per day suggest an abnormality exists. The surveillance range determined from Table 1 and the generation rate determined from Eq.1 are then applied to Table 2 (clause 6.1.2) to determine the suggested operating guidelines with the exception of C 2 H 2, where a generation rate > MDL is cause for concern and daily surveillance. Rs = 100 S o ( St So) T (1) Where: R s S o S t T is the Rate (% of So per day) is the first sample (ppm) is the second sample (ppm) is the time in days This is an unapproved IEEE Standards Draft, subject to change. 12

13 6.1.2 Determining the operator guidelines from surveillance range and generation rate Table 2 gives suggested operator guidelines for TDCG in the normal, caution and warning ranges found from Table 1 for given rate limits. The gas concentrations for individual components can be substituted for TDCG in column 2, Table 2 to determine operator guidelines from just the components. TABLE 2 - Guidelines for surveillance range and generation rate Surveillance range TDCG mg/kg (ppm) Warning > 1900 Generatio n rate R s (%/day) Suggested Operator Guidelines Sampling Interval > 7 Daily < 7 Weekly Operating Procedure Extreme caution: Diagnostics plan outage; Advise manufacturer or insurer. Caution 700 to 1900 > 7 Daily => 3 < 7 Weekly > 0.5 =< 3 Monthly Caution; Screening diagnosis; check load dependence. Advise manufacturer or insurer. Normal < 700 = > 0.3 <= 0.5 Monthly Caution check load dependence < 0.3 Normal surveillance Continue normal operation Transformers with TDCG in the normal surveillance range with generating rates > 0.3% per day, have reduced sampling intervals, therefore stricter operating procedures are suggested. If their generating rates are < 0.3% per day the operator may continue with normal surveillance sampling and operating procedure. In addition, with rates < 0.3% per day, normal transformers may have their normal maintenance sampling intervals lengthened. Transformers with TDCG in the caution or warning surveillance range will have reduced sampling intervals and stricter operating procedures. 6.2 Detection procedure for Type 2 transformers Referring again to the Flow Chart, Fig. 1, transformers with previous combustible gas history may have databases that are sufficient for statistical derivation of 90% probability norms for individual components or their total and operators can then establish their own equivalent to Table 1. A procedure for determining norms from a database is given in clause The norms derived in this manner from a database must reflect data from fault-free equipment. The newly derived 90% probability norms may be substituted for the normal surveillance range values given in Table 1. The caution and warning surveillance range norms will be proportional to those in Table 1. These new surveillance range norms and the generation rates calculated for the given sample gas concentrations are then applied in Table 2 to determine the suggested operating guidelines and surveillance sampling intervals. Both the statistically derived range values and the proportional caution and warning surveillance ranges are used as trial values which are refined by continued experience and should be agreed upon by user, manufacturer and insurer as appropriate. Transformers with statistically insignificant numbers of tests in the database continue to be evaluated per clause 6.1. Special cases may occur where the equipment operator, insurer and manufacturer agree on values called assigned norms that have been assigned to a particular type This is an unapproved IEEE Standards Draft, subject to change. 13

14 of equipment or service based on experience. Procedures for utilizing assigned norms are given in clause Procedure for statistical development of surveillance range norms from a gas database a. A Statistically sufficient number of normal, fault free, transformers of similar design and service are collected into a database. b. Values for each component Combustible gas (H 2, CH 4, C 2 H 2, C 2 H 4, C 2 H 6, CO) and their total (TDCG) determined from each transformer in the database are tabulated and sorted in decreasing order. A separate Frequency Distribution Chart (probability chart) is then constructed for each component and TDCG for all the transformers in the database using well-known statistical procedures such as those available in most spreadsheet programs. From the plotted frequency distribution for each component, the 90% probability norm for that component is the unique 90% probability value for the database and it signifies that 90% of all the normal transformers in this database have values at or below that value. c. You now can construct a table similar to Table 1 in this guide for your own particular case using the derived 90% norms for the normal surveillance range guidelines. The caution and warning values in your new table will be proportional to those in Table 1. This method of deriving individual norms from a transformer population will yield values, which are statistically representative of that population and preferable to the survey values in Table 1. Continued analysis of the database will refine the estimated norms and increase the users confidence in their reliability. The reliability of derived norms depends on the number of transformers in the database Procedure using assigned norms Referring to the Flow Chart Fig.1, the step-by-step procedure may employ assigned norms which are values for the normal, caution and warning surveillance ranges that have been developed as part of a proprietary surveillance plan; either from data developed within the operators company or from recommendations from outside contractors. For example, banks of transformers of duplicate design and loading may all utilize the same set of surveillance norms based on sound engineering judgment and agreement between user, manufacturer and insurer. Assigned norms are substituted for the values in Table 1. Then sample gas concentrations from transformers with assigned norms are utilized to determine the surveillance ranges and calculated generation rates which are then applied to Table 2 in the same manner as statistically determined norms in clause Diagnosis of transformer faults Referring to the Flow Chart, Fig.1, after completing the detection procedures in clause 6, the final steps of this process are the Key gas method in the caution surveillance range and Ratio diagnosis. The results and data from these investigations are then placed in a database record for each unit to enhance the reliability of subsequent tests. When the concentrations or generating rates of any single dissolved combustible gas component or the sum of the components (TDCG) extracted from the oil or gas space of an operating transformer rise into the caution range, the operator should begin to consider the possible sources of these gases and their generation progress by utilizing analysis in the caution range per clause 7.1. When the values reach the warning range, fault diagnosis per clause 7.2 employs an array of Ratios attributed to Rogers et al given in Table 4. Diagnosis of cellulose decomposition per clause 7.3 may also be applied. These diagnostic procedures can allow the operator to focus his subsequent investigation, utilizing load studies or thermal imaging to pinpoint thermal problems in the windings or cooling system or discharge monitoring to pinpoint electrical This is an unapproved IEEE Standards Draft, subject to change. 14

15 trouble. Generally, thermal faults produce lower temperatures and electrical faults except low-level partial discharge produce higher temperatures. The diagnostic Ratios used in Table 4 are: Ratio 1 (R1) = CH 4 / H 2 Ratio 2 (R2) = C 2 H 2 / C 2 H 4 Ratio 3 (R3) = C 2 H 4 / C 2 H 6 Graphical 2 or 3 dimensional plots of the three Rogers ratios provide an excellent physical representation of the limit envelopes circumscribing each fault type. Ratio diagnosis applies to gases extracted from the oil. For diagnoses of components extracted from the transformer inert gas space or relays see Annex B. 7.1 Analysis in the caution surveillance range In addition to increased surveillance for generation rate and trends, transformers with TDCG in the caution range may receive preliminary or screening diagnosis by the Key gas procedure Key gas procedure The Key gas procedure permits a tentative determination of possible fault types empirically determined from their unique gas species. The Key gas method may also be useful for benchmarks in the normal range and may also help to confirm diagnoses in the warning range. Table 3 relates fault types with typical proportions of their Key gas indicators. Table 3 Key gases KEY GAS FAULT TYPE TYPICAL PROPORTIONS OF GENERATED COMBUSTIBLE GASES Ethylene (C 2 H 4 ) Carbon- Monoxide (CO) Hydrogen (H 2 ) Hydrogen and Acetylene (H 2, C 2 H 2 ) Thermal Oil Thermal Oil and Cellulose Electrical Low Energy P.D. Electrical High Energy (arcing) Predominantly Ethylene with smaller proportions of Ethane, Methane, and Hydrogen. Traces of Acetylene at very high fault temperatures. Predominantly Carbon Monoxide with much smaller quantities of Hydrocarbon Gases in same proportions as Thermal faults in oil alone. Predominantly Hydrogen with small quantities of Methane and traces of Ethylene and Ethane. Predominantly Hydrogen and Acetylene with minor traces of Methane, Ethylene, and Ethane. Also Carbon Monoxide if cellulose is involved. 7.2 Diagnosis in the warning surveillance range When the dissolved gas concentrations and generating rates progress from the caution into the warning range, they are high enough to minimize the effects of sampling and analytical error. In the warning range, the Ratio diagnostic procedures given in clause are significant and provide relatively accurate fault diagnosis. The fault types defined in clauses 4.1, 4.2, and 4.3 were empirically derived from considerable experience of several European investigators who correlated gas diagnoses on many units with fault types subsequently associated with observed disturbances or failures. Trends per clause 8.0 can confirm Ratio diagnosis. The ratios of CO 2 /CO in clause 6.3 can provide diagnostic information on cellulosic degradation associated with thermal faults. This is an unapproved IEEE Standards Draft, subject to change. 15

16 7.2.1 Rogers Ratio diagnosis This procedure utilizes the three ratios R1, R2, and R3. When gas concentrations yield a ratio outside the Table 4 limits for any given fault type, applying the Key gas method and graphical analysis of gas generating trends can help to clarify the situation. Ratio diagnosis begins with extraction of the combustible gases from the oil sample and chromatographic analysis per clause Ratios are calculated from the reported gas concentrations. Then the Ratios are compared to values in Table 4 providing fault diagnosis. Table 4 - Diagnostic ratios for fault determination CASE C 2H 2 (R2) C 2 H 4 CH 4 (R1) H 2 C 2H 4 (R3) C 2 H 6 Suggested Diagnosis 0 < 0.01 < 0.1 < 1.0 Normal 1 =>1.0 =>0.1 < 0.5 => 1.0 Low energy Discharge 2 => 0.6 < 3.0 => 0.1 < 1.0 => 2.0 High energy Discharge 3 < 0.01 => 1.0 < 1.0 Low Temp. Thermal 4 < 0.10 => 1.0 =>1.0 < 4.0 Thermal < 700 C 5 < 0.2 => 1.0 => 4.0 Thermal > 700 C Case 0 = Normal Unit Case 1 = Low Energy Discharge Case 2= High Energy Discharge Case 3 = Low Temp. Thermal (< 300 C) Case 4 = Thermal (300 C C) Case 5 = Thermal (> 700 C) 7.3 Diagnosis of cellulose decomposition The ratio of CO 2 /CO may sometimes be used as an indicator of the thermal decomposition of cellulose. This ratio for normal cellulosic decomposition is usually between 7 and 10. Also, CO 2 /CO ratios below 3 and significantly greater than 10, suggest excessive thermal decomposition. The magnitudes of the concentrations of CO 2 and CO should exceed 5000 ppm and 500 ppm respectively, in order to improve the certainty factor. Ratios are sensitive to minimum values. In addition to the CO 2 /CO ratio, the Furanic Series per ASTM D5837 is also an indicator of cellulose decomposition. 8.0 Trend and graphical analysis Trends in the generation of individual gas concentrations, key gases, generation rate, or Ratios may be used to evaluate the temporal development of faults. Graphical analysis of trends may help to indicate when This is an unapproved IEEE Standards Draft, subject to change. 16

17 faults progress from relatively benign thermal faults into more critical electrical faults. It is suggested that the report of a possible fault by either the Key gas procedure, or Ratio diagnosis, should include a numerical value for an increasing or decreasing trend. The trend for the magnitude of CO to increase and CO 2 to decrease indicates advancing degradation of cellulose insulation. Graphical analysis may be applied using two-dimensional (2D) or three-dimensional (3D) plots of the values for the pertinent ratios. Two-dimensional plots are used to determine the correlation between the ratio CH 4 /H 2 on the x axis vs. the ratio C 2 H 2 /C 2 H 4 on the y axis; or the ratio CH 4 /H 2 on the x axis vs. the ratio C 2 H 4 /C 2 H 6 on the y axis with the coordinates for the Rogers ratio limits for particular cases. However, the 3D plots are more revealing. Three-dimensional graphical analysis may be accomplished by developing a three-dimensional composite plot of all five Rogers fault case envelopes from Table 4. A simpler alternative is to develop separate envelopes for each of the five Rogers fault cases. Then gas ratios from a transformer sample may be plotted for either the 2D or 3D representations. A repeated result within a specific fault region or volume confirms a diagnosis. A trend for the results to move toward the boundaries of the envelopes suggests another fault or more than one fault exists. For example, in the 3D case say the three calculated ratios for a particular analysis converge at a point within a Case 3 fault case envelope (Low temperature Thermal fault). Then, if the solution points for successive samplings show a trend for the diagnosis to move out toward the boundaries of the Case 3 envelope, the conclusion may be drawn that the fault is changing in intensity or it may be progressing to another fault type. An example would be the progression of a thermal fault case into an electrical fault case that suggests the movement of gas bubbles from a thermal fault location into nearby areas of high electrical stress. 8.1 Construction of a 3 D graph The construction of a 3D Ratio graph is accomplished by plotting the limit envelopes from Table 4 for ratio R1 vs. R3 (on the xy plane); ratio R1 vs. R2 (on the xz plane); and ratios R3 vs. R2 (on the yz plane) using appropriate scales. All 5 cases from Table 4 may be shown together on the same graph, or each case may be graphed separately as in Figs. 2a and 2b for simplicity. Then the actual values of the ratios from the transformer sample test data are located within their respective envelopes. Figure 2 3D ratio graph Fig. 2a, Case 3 Fig. 2b, Case 1 R3 R3 Xb Xa Ya Zb Yb Za R1 R1 R2 R2 Low Temperature Thermal Fault Low Energy Discharge This is an unapproved IEEE Standards Draft, subject to change. 17

18 Fig. 2a is an example of a 3D Rogers Ratio envelope for a Low Temperature Thermal fault (Case 3 Table 4). The line Xa represents the interval between 1.0, the minimum Table 4 value of R1, and 4.0 an arbitrary maximum limit (see note) used for plotting the envelope. The line Ya represents the interval between 0, the arbitrary minimum (see note) value of R3, and its Table 4 limit 1.0. The line Za represents the interval between 0, the arbitrary minimum value of R2, and its Table 4 limit Fig. 2b is an example of a 3D Rogers Ratio envelope for a low energy discharge fault (Case 1 Table 4). The lines Xb, Yb, and Zb represent the Table 4 intervals for plotting the Case 1 envelope. Arbitrary values of 4.0 were used for maximum R2 and R3 limits for plotting the envelope. Note: In order to describe a graphical envelope an arbitrarily chosen maximum limit was chosen for a Table 4 ratio that does not contain a minimum limit. Zero values are not obtained with ratios. 9.0 Detection and analysis of gases in gas space, relay or other fixed or portable devices The on-site detection and determination of combustible gases in an operating transformer, using a portable or on-line combustible gas meter can be the earliest and most easily obtained indication of a thermal or electrical fault. And it may form the basis for further testing. Other portable or on-line devices are useful for periodic testing of operating transformers without removing them from service. These devices normally accept a small volume of oil or gas depending on the technology used. They will provide a measurement of hydrogen or composite values of hydrogen and carbon monoxide, or in the case of a portable GC, a measurement of all combustible gases. Another type of gas monitoring system samples the transformer gas space at fixed intervals to provide a measure of individual combustible gases. Periodically, a sample of the oil is pumped from the transformer into the monitoring system that extracts the gases and passes them through a micro GC for measurement. This system uses helium as the carrier gas so careful attention should be given to ensure proper operation. Gases collected in the gas space or gas relays can be sampled per clause 5.4. ASTM Method, D-3612, can determine the individual components of the gas samples. Since the sample is already in the gas phase the extraction step in the ASTM method is not applicable. Annex A describes a TCG procedure for evaluating the transformer operating procedure given TCG percentages and generating rates. The procedure in Annex A is similar to the DGA procedure in Table 2 clause Combustible gases may be explosive. Strict precautions should be observed when sampling directly from the transformer. Annex B describes procedures for converting concentrations from gases from relays or sampled from gas spaces into dissolved gas-in-oil equivalents that can then be diagnosed per clause 7.2. This is an unapproved IEEE Standards Draft, subject to change. 18

19 Annex A (Normative) TCG detection and operating procedure When combustible gases are first found in the transformer gas space, sampling is repeated to determine the rate of increase of % TCG per day. Then the % combustibles (TCG) and the rate of increase (%/day) are applied to Table A1 to determine guidelines for sampling intervals and operating procedures in the normal, caution and warning surveillance ranges similar to the procedures for dissolved gas. Table A1 utilizes values for TCG % limits and generation rates which can be used to determine the surveillance ranges and the suggested sampling intervals and operating guidelines similar to Table 2 in clause Table A1 TCG surveillance ranges and suggested guidelines Surveillance Range TCG % Generation Rate % TCG / day Suggested Operator Guidelines Sampling interval Operating procedure Warning > 5 Caution.5 to 5.>.03 Daily <.03 Weekly > 0.03 Weekly.01 to.03 Monthly < 0.01 Quarterly Extreme Caution DGA per 7.2 a Caution DGA per 7.1 b Normal <.5 >0.01 Monthly Caution c < 0.01 Normal Surveillance Continue Normal operation a b c Plan outage; advise manufacturer or insurer. Check load dependence; advise manufacturer or insurer. Check load dependence. This is an unapproved IEEE Standards Draft, subject to change. 19