IEC 60567 SAMPLING OF GASES AND OF OIL FOR ANALYSIS OF FREE AND DISSOLVED GASES
IEC 60567 OIL-FILLED ELECTRICAL EQUIPMENT - SAMPLING OF GASES AND OF OIL FOR ANALYSIS OF FREE AND DISSOLVED GASES - GUIDANCE
9 Quality control
9.1 Verification of the entire analytical system
The preferred method of ensuring that the entire system (extraction and chromatography) gives stable and accurate results is by analysing a known quantity of gas-in-oil standards (containing all the gases listed in 8.1 in proportions resembling those found in oils taken from transformers) in place of an oil sample and carrying out all the procedures detailed above.

In the case of the Toepler pump extraction method, it has been confirmed that the extraction efficiency and performance of the whole analysis equipment can be verified by the daily analysis of an air-saturated oil standard (see Annex D) and one standard gas mixture with average gas concentrations in the range usually found in transformer oil samples.

It is recommended to use at least two gas-in-oil standards, one containing low concentrations of gases (resembling oils in factory tests) and the other containing higher concentrations (resembling oils from equipment in the field) to check the quality of the results produced by the analytical system in the entire linearity range. The chromatogram peak areas or heights can then be related directly to the quantities of gases present in the oil, compensating automatically for incomplete extraction and other operational factors.

It is good practice to repeat this procedure at intervals of calibration of not more than 6 months or following changes in apparatus or operating conditions.

9.2 Limits of detection and quantification
The basic requirement is that the analytical system, consisting of degassing equipment and gas chromatograph, shall have adequate sensitivity for the task for which it is designed.

Not only the detection limit, but also the sensitivity, the repeatability and the accuracy of results depend on many details of the overall procedure, in particular the method of extraction and the design and method of operation of the gas chromatograph and its ancillaries.

For example, by using a methanator combined with a flame ionization detector, instead of using a thermal conductivity detector, the sensitivity and detection limit for the carbon oxides are considerably improved.

Tests on equipment in service where the dissolved gas concentration is often high do not require the same detection limits as factory tests where gas concentrations are usually very low.

In the case of head space, extraction efficiency is low, and the concentration of gases in the extracted gas phase typically 100 times lower than with the other extraction methods. At low gas-in-oil concentration levels, detection limits in the nl/l range in the gas phase of the headspace are therefore necessary, requiring the use of more sensitive equipment and procedures (see 7.5.4.3).

It is necessary that each laboratory determines overall procedures which will give suitable sensitivities for all gases. For guidance, experience indicates that the entire analytical system should be capable of detecting gases dissolved in oil at the concentrations indicated in Table 5.

For the analysis of gas samples taken from gas-collecting relays, a detection limit equivalent to that obtained for the analyses of the gases extracted from the oil for service tests adequate.

NOTE The limit of detection is the lowest concentration that can be identified. The limit of quantification is the lowest concentration that can be quantified with a reasonable precision and accuracy.

9.3 Repeatability, reproducibility and accuracy
Repeatability (r), reproducibility (R) and accuracy are defined in detail in ISO 5725.

9.3.1 Repeatability
Repeatability is related to the differences which are observed when the same oil sample is analysed several times by the same laboratory over the same day or a short period of time.

Having established methods which give adequate overall detection limits, each laboratory must confirm that these methods give adequate repeatability following the procedure described in ISO 5725.

For gas concentration levels greater than 10 ul/l the repeatability of a laboratory shall be considered as acceptable if the absolute difference (A - B) of two measurements A and B satisfies the following equation:
r = (A - B) < k x (A + B) /2
which means that the repeatability of the laboratory, at 95 % confidence limit, is lower than k times the mean concentration of the gas analysed. The k coefficient depends on the nature of the gas analysed.

A general acceptable value, calculated from an international IEC inter-laboratory test is:
k = 0,07 for concentrations >10 ul/l and <1000 ul/l, and k = 0,10 for concentrations >1000 ul/l.

For low gas concentrations (for example, <10 ul/l), the required repeatability is given by the following equation: r = S (where S = detection limit), whatever the concentration, as deduced from the same international IEC inter-laboratory test.

It is recommended that each laboratory check its own individual repeatability at both concentration levels at appropriate time intervals or after major changes on its system (gas extractor and gas chromatograph). To do that, a sufficient number of analyses on multiple samples of the same oil are run within a short period of time (less than one day), then the repeatability of results r is determined according to ISO 5725.

9.3.2 Reproducibility
Reproducibility is related to the differences which are observed when the same oil sample is analysed by different laboratories (inter-laboratory reproducibility), or when it is analysed by the same laboratory over long periods of time (after several days, weeks or months) (intra-laboratory reproducibility).

Inter-laboratory reproducibility has been evaluated by CIGRE as around +/-20 % at medium concentration levels.

It is recommended that each laboratory check its own intra-laboratory reproducibility at different concentration levels at appropriate time intervals or after major changes on its system (gas extractor and gas chromatograph). To do that, analyse multiple samples of the same oil at regular intervals of time, for instance each week or each month over a period of several months, then determine the reproducibility of results R according to ISO 5725.

A good conservation of the samples is necessary between analyses. Storing the samples in a fridge is recommended to avoid reactions with light and oxygen present in the oil.

9.3.3 Accuracy
Accuracy is related to the differences which are observed between the values analysed by a laboratory and the true values of dissolved gases contained in the oil sample.

Inaccurate DGA results may lead to wrong fault diagnoses, especially if gas ratios are close to a fault zone boundary, or to inappropriate actions on the equipment if concentration values are close to the typical or alarm values defined in IEC 60599.

To be able to determine accuracy the nominal values of dissolved gas concentrations shall be known. The mean of several measured values has been shown by inter-laboratory tests to be different from the nominal values. The determination of accuracy shall be carried out with gas-in-oil standards prepared according to Clause 6, or through participation to round robin tests using such standards or certified gas-in-oil standards prepared according to ISO 5725.

Examples of accuracies which can be obtained using the overall experimental procedure are given in Table 6. These values are deduced from IEC and CIGRE inter-laboratory tests made on two gas-in-oil standards (prepared according to 6.1) and involving 44 laboratories world wide. One standard sample contained medium gas concentration levels (hydrocarbons between 9 ul/l and 60 ul/l, CO and CO2 between 100 ul/l and 500 ul/l). The other one contained low gas concentration levels (hydrocarbons between 1 ul/l and 10 ul/l, CO and CO2 between 30 ul/l and 100 ul/l).

It is recommended that each laboratory determine its own accuracy, which may differ from the values in Table 6. To do that, analyse a gas-in-oil standard according to Clause 6, then determine the accuracy according to ISO 5725.

10 Report of results
The report for DGA shall include
- a reference to this standard;
- information on the oil sample (see Clause 5);
- the sampling procedure used (for example, syringe, bottle) (see Clauses 3 and 4);
- the extraction procedure used (for example, Toepler, head space) (see Clause 7 and Annex C);
- the detection limits for each gas with the analysis procedure used (see 9.2);
- for each gas analysed, the results in ul/l or in umoles/l (see 8.7).

NOTE 1 When available, it may be useful for diagnosis purposes to indicate the average accuracies obtained by the laboratory at these gas levels with the analysis procedure used (see 9.3.3).

NOTE 2 Guidelines for drafting the report in terms of quality assurance can be found in ISO/IEC 17025.