A Nomograph for Interpretation of LTC A Nomograph for Interpretation of LTC A Nomograph for Interpretation of LTC A Nomograph for Interpretation
of LTC- -- -DGA Data DGA Data DGA Data DGA Data
Fredi Jakob, Karl Jakob, Simon Jones Weidmann-ACTI Rick Youngblood, Cinergy
I. Introduction I. Introduction I. Introduction I. Introduction
Dissolved Gas Analysis, DGA, is based on the fact that the release of energy in oil filled electrical equipment results in a partial destruction of insulating fluids and/or solid insulation. The number of molecules destroyed is relatively small but detectable amounts of low molecular weight gases are produced. The quantification of the low molecular weight fault gases is the basis of all DGA.
In the main tank of a transformer very little, if any fault gases other than carbon oxides should be produced. Carbon oxides are continuously produced by unavoidable decomposition of cellulose insulation. In contrast, equipment that involves conductor- switching operations produces fault gases during normal operation. Initially it was assumed that the gases produced by the arcing process associated with switching would mask gases associated with problems such as carbon deposition on the contacts, which results in excessive heating.
Youngblood 1 was one of the first investigators to recognize that the gassing pattern for a healthy load tap changer, LTC, would be significantly different from that observed in a problem LTC. Specifically, he suggested that acetylene and hydrogen are generated during the normal arcing process and the hydrocarbons, methane, ethane and especially ethylene are generated when overheating occurs in a problem LTC. These three hydrocarbons are often called hot metal gases since they are produced when a heated conductor is in contact with mineral oil based dielectric fluid.
II. Data Interpretation II. Data Interpretation II. Data Interpretation II. Data Interpretation A. Concept A. Concept A. Concept A. Concept The interpretation of DGA data for transformers, LTCs and OCBs is empirical in nature. The development of interpretation protocols for OCBs and LTCs parallels the development of DGA diagnostics for the main tanks of transformers. Key gases associated with heating problems are methane, ethane and ethylene. These gases, in the order listed, require increased energy for their production. The energy requirements were calculated by Halstead 2 . Initially Youngblood ignored the levels of arcing gases, acetylene and hydrogen that developed whenever an LTC operated. Subsequent work by Youngblood 5 indicated that arcing gases are also diagnostically significant. For example, increased acetylene levels were often followed by increased heating gas concentrations. The increased acetylene is due to changes in the arc duration and/or characteristics as the contacts are eroded or covered with carbon.
The next step in the development of diagnostic protocols for LTCs was the empirical determination of normal or threshold values. The gas retention rate in an LTC is very dependent on breathing configuration, so this is a major factor in determining threshold levels. Free breathing LTCs rapidly lose gases to the environment while sealed LTCs retain much of the gas produced.
Threshold levels have been determined for specific models and types of LTCs by Doble 3 and Baker 4 . Generic levels have been set by Youngblood 5 , and are useful when threshold values have not yet been determined. Fault gas ratios, which are discussed below, are applicable for unit evaluation when threshold values are reached or exceeded. We are investigating the value of these ratios even if individual gases are below the threshold values.
B. B. B. B. Fault Gas Concentration Ratios Fault Gas Concentration Ratios Fault Gas Concentration Ratios Fault Gas Concentration Ratios 1. Normal Operation Arcing, which generates very high temperatures, occurs with each switching operation in an LTC. Arcing produces mainly acetylene and hydrogen. When arcing gases are being produced, heating gases are also produced since the oil temperature varies with distance from the arcing contacts. At lower oil temperatures, heating gases are produced more abundantly than acetylene.
If one accepts this hypothesis, one would expect that the ratios of heating to arcing gases in a problem free unit would remain fairly constant with operational count. Since the contact surface is being eroded and additional deposits are being formed with each operation, these ratios will predictably change slightly with operation count, providing there is not a large change in load current. Although the load on a distribution circuit in a sub station does remain nearly constant over a period of time with minor ups and downs, anything that causes a significant change in average constant load will change the ratios. The data presented in a paper by Duval 6 , shown in Table 1 supports this hypothesis. Note that all of the significant ratios in these problem free units are well below those found in a LTC that has developed a heating or other problem. Gas ratios that are useful in detecting LTC problems are listed in Table 2.
Table 1. Gas Formation as a Function of Operation Co Table 1. Gas Formation as a Function of Operation Co Table 1. Gas Formation as a Function of Operation Co Table 1. Gas Formation as a Function of Operation Count unt unt unt. (Duval (Duval (Duval (Duval 6 66 6 ) )) ) Operations: 500 3600 49000 Gas produced/operation Gas/500 Gas/3600 Gas/49000 Hydrogen 6870 12125 14320 13.74 3.37 0.29 Methane 1028 5386 10740 2.05 1.50 0.22 Acetylene 5500 35420 53670 11.00 9.84 1.10 Ethylene 900 6400 35839 1.80 1.78 0.73 Ethane 79 400 3944 0.16 0.11 0.08 Ratio: Acetylene Ethylene
0.16 0.18 0.66
*Note: Some gas is always lost with time. Therefore, the gas concentration per operation is expected to decrease with operation count. Duval did not provide breathing configurations for this data.
The ratio is fairly consistent and independent of operation count. When the operation count is very high, 49,000, the ratio increase is probably related to changes in the contact condition.
2. Heating Problems Initially a resistive film develops on contacts, which results in an increase in contact resistance, increased heating and an increase in heating gas concentrations. Since more heating gases are produced, the ratio of heating to arcing gases, ethylene to acetylene, increases. This change in gas concentrations and gas concentration ratios indicates problems. The concentration ratios of ethane to methane and ethylene to ethane are temperature dependent, so both are expected to increase with increasing temperature, in a problem unit. These two heating gas ratios should also reflect increased contact resistance and heating.
Table 2. Important Gas Concentration Ratios. Table 2. Important Gas Concentration Ratios. Table 2. Important Gas Concentration Ratios. Table 2. Important Gas Concentration Ratios. Heating to Arcing Ratios Heating to Arcing Ratios Heating to Arcing Ratios Heating to Arcing Ratios Hydrogen Acetylene Ethylene +
Acetylene Ethylene
Hydrogen Acetylene Ethane Ethylene Methane + + +
Acetylene Ethane Ethylene Methane + +
Temperature Dependent Ratios Temperature Dependent Ratios Temperature Dependent Ratios Temperature Dependent Ratios Methane Ethane
Ethane Ethylene
C. Application of Gas Concentration Ratios C. Application of Gas Concentration Ratios C. Application of Gas Concentration Ratios C. Application of Gas Concentration Ratios As is the case with main tank DGA, ratios of fault gas concentrations are most valid if at least one of the fault gases exceeds its threshold value. Threshold values used by Weidmann-ACTI are model specific, whenever this data is available. If model specific data is not available, we use the monthly watch levels developed by Youngblood 5 as threshold values. Normal values used for the ratios represent the 90 th
percentile of fault gas concentrations from a very large number of DGA results (~2500 units). These 90 th percentile values, which are listed in Table 3, are generic and do not take into account the difference in gassing rates or breathing configurations of specific units. Ratios are, as expected, model specific and we hope to refine our results as more data becomes available. Table 4 lists the ratios calculated for a Westinghouse LTC and for a McGraw Edison LTC, which according to Bakers 4 threshold levels require attention.
Table 3. Generic 90 Table 3. Generic 90 Table 3. Generic 90 Table 3. Generic 90 th th th th Percentile Fault Gas Ratios Percentile Fault Gas Ratios Percentile Fault Gas Ratios Percentile Fault Gas Ratios
Note the difference between the model specific ratios listed in Table 4, and the generic values which are listed in Table 3. The differences in critical ratios for the R1 R2 R3 R5 R5 R6 Acetylene Ethylene
Acetylene Ethane Ethylene Methane + + 0.36 0.24 0.38 0.31 7.51 0.56 Westinghouse and McGraw Edison units illustrate how model specific the gassing characteristics can be.
Table 4. Unit Specific Ratio Comparison. (Based on values set by C. Baker) Table 4. Unit Specific Ratio Comparison. (Based on values set by C. Baker) Table 4. Unit Specific Ratio Comparison. (Based on values set by C. Baker) Table 4. Unit Specific Ratio Comparison. (Based on values set by C. Baker) WESTINGHOUSE LTC UTT Ethylene Ethane Methane Acetylene Hydrogen LT3 3,000 250 1,000 5,000 5,000
Ratio 1 Acetylene Ethylene
Ratio 2 Hydrogen Acetylene Ethylene +
Ratio 3 Hydrogen Acetylene Methane Ethane Ethylene + + +
Ratio 3 Hydrogen Acetylene Methane Ethane Ethylene + + +
Ratio 4 Methane Ethane
Ratio 5 Ethane Ethylene
5.00 2.22 3.11 1.00 5.00
LT3 is Bakers designation of a unit requiring inspection.
D. Graphical Methods D. Graphical Methods D. Graphical Methods D. Graphical Methods Graphical methods have been widely used as an aid for the interpretation of DGA data for transformer main tanks. A nomograph developed by Church et al. 7 has been successfully used to graphically calculate and interpret fault gas ratio data for power transformers. A similar nomograph can be applied to interpretation of DGA data for both LTCs and OCBs. The nomograph consists of adjacent parallel logarithmic scales on which fault gas concentration in parts per million, ppm, is plotted. A simple example will illustrate the concept, shown in Figure 1. Rose 8 , uses the ethane/methane ratio as an indicator of a heating problem. He has stated that the critical value of this ratio occurs when it exceeds unity. Each of these gases is plotted below on a logarithmic scale. The two scales are aligned, without any offset. CH 4 C 2 H 6 1 2 3 4 5 6 7 8 9 10 2 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 10 4 9 8 7 6 5 4 3 2 10 3 10 5 1 2 3 4 5 6 7 8 9 10 2 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 10 4 9 8 7 6 5 4 3 2 10 3 10 5
Figure 1. Logarithmic Representation of Methane/Ethane Gas Concentration Ratios.
A straight line connecting the methane concentration of 800 ppm to the ethane concentration of 1200 ppm, a ratio of 1.5 would have a slope greater than one. Any set of equal gas concentrations would have a zero slope. A sample with less methane than ethane will be connected by a line with a negative slope. The slope of the line is thus a clear indication of the LTC condition. In cases where the transition for a selected ratio, from normal to a problem condition occurs at any value other than one, the relative position of the scales is adjusted so that a horizontal line indicates the transitional ratio. Lines with positive slopes indicate ratios greater than the transitional value. Lines with negative slopes correspond to ratios below the critical values.
Threshold levels are also indicated on each scaled gas concentration. At least one of the gas concentrations should be above these values before the ratios are considered valid. As previously stated, this concept requires further study.
Figure 2 is a copy of the proposed nomograph for the empirical interpretation of LTC DGA data. The threshold values indicated are generic in nature. Unit specific information should be used whenever sufficient data is available. The relative position of each scale is base on a statistical determination of the 90 th percentile gas ratios from our extensive database. These threshold values are shown in Table 5. These values are also generic in nature and should be replaced if model specific data is available. The nomograph is constructed so that any line with a positive slope indicates a LTC heating problem. Table 5. 90 Table 5. 90 Table 5. 90 Table 5. 90 th th th th Percentile Values for Different Breathing Configurations. Percentile Values for Different Breathing Configurations. Percentile Values for Different Breathing Configurations. Percentile Values for Different Breathing Configurations. C2H2 C2H4 C2H6 CH4 H2 Configuration Free Breather 2733 851 107 379 1418 FB/Desicant 2811 473 130 141 467 Sealed 3744 1337 178 572 1781 Vacuum 42 49 79 61 72
Figure 2. Proposed Nomograph for the 90th Percentile Gas Thresholds.
III. Case Studies III. Case Studies III. Case Studies III. Case Studies
Two case studies for Cinergy data are presented below. This data, from Cinergys initial work, was interpreted without consideration of fault gas concentration ratios. Case history I is of particular interest since serious damage occurred in the six month interval between scheduled tests. The data indicates that critical gas ratios such as the ratio of ethylene to acetylene doubled from March 92 to Feb 93. This points out the necessity of trending both ratios and gas concentrations. We believe that the extent of the coking would have been less severe if the unit had been inspected in Feb 93 based on the doubling of the significant ratios, rather than to have waited for six additional months.
12-Mar-92 This unit indicated the early stages of mechanical difficulties. While the Acetylene and Hydrogen levels are elevated, the level of Ethylene is less than 100 ppm. Indicating a continuance of annual monitoring
LTC Case Study I LTC Case Study I LTC Case Study I LTC Case Study I
Federal Pacific TC-25 69KV x 12KV 20MVA Desiccant Breather Date Comments C2H2 CH4 C2H6 C2H4 H2 3/12/1992 Annual DGA Test Cycle 589 60 2 89 144 2/1/1993 6 Month Test Cycle 1625 342 70 534 3099 8/12/1993 Thermal Runaway 1633 53434 55535 253024 2217 Date Comments Ratio 1 Ratio 2 Ratio 3 Ratio 4 Ratio 5 3/12/92 Annual DGA Test Cycle 0.15 0.12 0.21 - 0.15 2/1/93 6 Month Test Cycle 0.32 0.11 0.2 0.21 0.33 8/12/93 Thermal Runaway 155 66 94 1.03 155 C2H4 / C2H6 Ratio 5 C2H6 / CH4 Ratio 4 (CH4 + C2H4 + C2H6) / (C2H2 + H2) Ratio 3 C2H4 / (C2H2 + H2) Ratio 2 C2H4 / C2H2 Ratio 1 1-Feb-93 At this time the unit was placed on a 6 month monitoring schedule, due to the elevated Acetylene, Hydrogen, and Ethylene levels. At 534 ppm Ethylene immediate removal from service was not indicated.
Nomographic representations of Case Study II data on 8/31/92 is shown in Figures 3-5. In nearly all graphs, the various threshold values were either met or exceeded, and nearly all ratios have a slope greater than one. This illustrates that this approach is capable of identifying unit problems even in the absence of the available unit specific information. As long as the correct breathing configuration is given, the correct set of threshold values can be applied, and units can then be correctly diagnosed.
Cinergy Comment Cinergy Comment Cinergy Comment Cinergy Comments ss s:
Date Comments C2H2 CH4 C2H6 C2H4 H2 8/31/1992 Removed from Service 8527 3279 1135 9606 9083 12/17/1993 Post Repair 501 387 16 375 2883 5/1/1994 Normal 541 534 9 313 3800 8/17/1995 placed on 6 month watch 648 590 52 836 3995 LTC Case Study II LTC Case Study II LTC Case Study II LTC Case Study II
Westinghouse UTT-A 138KV x 69KV x 13.8KV Sealed Date Comments Ratio 1 Ratio 2 Ratio 3 Ratio 4 Ratio 5 8/31/1992 Removed from Service 1.13 0.55 0.8 0.35 0.15 12/17/1993 Post Repair 0.75 0.11 0.23 - - 5/1/1994 Normal 0.58 0.07 0.2 - - 8/17/1995 Placed on 6 Month Watch 1.29 0.18 0.32 0.06 0.09 C2H4 / C2H6 Ratio5 C2H6 / CH4 Ratio4 (CH4 + C2H4 + C2H6) / (C2H2 + H2) Ratio3 C2H4 / (C2H2 + H2) Ratio2 C2H4 / C2H2 Ratio 1 31-Aug-92 Based on the DGA result this unit was immediately removed from service. The fault was determined to be due to contact misalignment. 12-Aug-93 Too late, by August, the unit was in thermal runaway, as indicated by the extremely high level of Ethylene at 253,024 ppm. Repairs included a Tap Shaft board, Slip Rings and a New Reversing Switch assembly.
Figure 5. Nomographic Representation of Case II Utilizing 90th Percentile Threshold Values.
IV. Indicating Problem Severity IV. Indicating Problem Severity IV. Indicating Problem Severity IV. Indicating Problem Severity
The severity of a problem, or condition of the unit, may be designated as Normal, Caution or Warning. A proposed nomograph has been developed by taking the 90 th
percentile values as a baseline, and aligning the concentration scales so that any connecting line, with this ratio, will have a slope of zero. Units having a zero slope and/or a slope that resides below these threshold values are diagnosed as Normal. A slope of the line residing within one standard deviation of the threshold values is diagnosed as a Caution, and at two standard deviations is a Warning. A section of the proposed nomograph is illustrated in Figure 6.
1 0 5 10 3 2 3 4 5 6 7 8 9 10 4 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 10 2 9 8 7 6 5 4 3 2 1 1 0 5 1 0 3 2 3 4 5 6 7 8 9 10 4 2 3 4 5 6 7 8 9 9 8 7 6 5 4 3 2 10 2 9 8 7 6 5 4 3 2 1 C 2 H 2 C 2 H 4 N o r m a l C a u t i o n W a r n i n g S l i d i n g L e g e n d
Figure 6. Nomograph Diagnosis of LTC Condition.
IV. IV. IV. IV. Future Work Future Work Future Work Future Work The empirical analysis of DGA data for LTCs is well developed. Gas concentration levels and gas concentration ratios can differentiate between normal and problem units. We believe that both the concentration and ratio values will work best if they are model specific. The compilation of these values requires user feedback on problem units. Trending of both the gas concentrations and ratios is always the best method to identify incipient problems. Another concept under investigation is normalization of fault gas data. We believe that during normal switching operations, the ratio of ethylene to acetylene is fixed. This ratio should remain constant for different numbers of operations. Furthermore, since these two gases have approximately the same solubility in mineral oil and approximately the same escape rates from the oil, the ratios should remain fairly constant even for free breathing units. For example if the concentration of both ethylene and acetylene are 100 ppm then the ratio is one. If a heating problem is superimposed on the normal arcing process and the gas levels are 175 ppm for ethylene and 150 ppm for acetylene there would be an additional 25 ppm of ethylene due to the heating problem. One could thus normalize our results using a ratio of:
Ethylene (due to heating) Acetylene (due to arcing) = Total Ethylene - (Ethylene produced during normal arcing) Acetylene
References: References: References: References:
1. Youngblood, R., Jakob, F., Haupert, T.J. Application of DGA to Detection of Hot Spots in Load Tap-Changers, Minutes of the Sixtieth Annual International Conference of Doble Clients, 1993, Sec. 6-4.1.
2. Halstead, W. D., A Thermodynamic Assessment of the Formation of Gaseous Hydrocarbons in Faulty Transformers, Journal of the Institute of Petroleum, Vol. 59, Sept. 1959, pp. 239-241.
3. Doble Client Transformer Committee Subcommittee Report on Transformer Load Tap Changer Dissolved Gas Analysis September 24, 2001.
4. Baker, Charles. Personal correspondence. 2002.
5. Youngblood, R., Baker, C., Jakob, F., Perjanik, N., Application of Dissolved Gas Analysis to Load Tap Changers.
6. Duval, Michel. , A Review of Faults Detectable by Gas-in-Oil Analysis in Transformers, IEEE Electrical Insulation Magazine May/June 2002, Vol. 18 no. 3, pp. 8-17.
7. Church, J.O., Haupert, T.J., and Jakob, F., Electrical World, Vol. 201, No. 10, October 1987, pp. 40-44.