By Michael Belanger
The key to analyzing transformer test results is to compare them with other existing reference data. Two types of reference data are available: evaluation by threshold witness group or existing historical test data.
If it is the first time that a measurement campaign has been carried out on equipment and therefore no historical data is available, we try to obtain reference data elsewhere to which we can compare our measurement results. This is called evaluation by threshold witness group. The evaluation is more accurate if we use a reference value from a similar group of equipment and the probability of making a correct diagnosis is better. Exceptions are the values on the nameplate such as transformer ratio or C1 and C2 from the capacitance tap of equipped bushings.
Where this is not the first measurement campaign undertaken on the equipment, the previously measured data allows for direct comparison and enables us to carry out a close follow-up of the measured results. This is called follow-up by direct reference. As a back-up, we would also use the reference values from the witness group.
Reference data is generally established by design or by a measurement campaign. An example of the norm established by design is the transformer turn ratio where the difference between the measurement and the calculated value, from the nameplate, should be less than 0.5 per cent. An example of the established norm by a measurement campaign is that of the power factor. Figure 1 illustrates statistical data of the power factor measured between the high and low voltage windings (CHL). This data was compiled by DOBLE. According to the graph, CHL should have a power factor lower than 0.5 per cent.
We will list some of the reference data as it is used in the electrical industry.
The reference information generally comes from private or public groups which collect large quantities of information on equipment such as transformers, notably IEEE, CEI, DOBLE, manufacturers, suppliers of electricity, etc.
This section contains the norms which are actually in use for new oil and oil-in-service.
When purchasing replacement oil, it is convenient to use the norm ASTM D3487 for type II oil or the norm DOBLE's "TOPS". Type I oil contains only 0.08 per cent antioxidant versus 0.3 per cent for type II. The type of antioxidant added should be indicated so as to add the same type in the future. This norm is an indication of the durability of the oil but, most of all, will provide compatibility of the materials within the transformer.
When very large quantities of oil are involved, we recommend obtaining for your records the refining process and the type of crude. The degree of distillation and refining of the crude and the kind of processes involved may greatly change the constituents of the final product .
Other supplementary tests could be added: resistivity D1169, foaming tendency D892, electrostatic charge tendency ECT and the dielectric of oil in movement. 
Before the oil can be introduced into the transformer, it should have the properties of the norm represented in Table 2.
For oil in service, some criteria merit being examined individually.
It is well known that acidity and interfacial tension are indicators of oxidation of oil. Oxidized oil, in its turn, accelerates the aging of insulation. A study conducted by ASTM, during 1946 and 1957 on 500 transformers, indicated that if the interfacial tension was superior to 24 Dynes/cm and the acidity was inferior to 0.1 mg/KOH/g, then sludge would not have formed in the oil. These values should be the boundaries. This study made no correlation between sludge formation and the amount of water in the insulation and the operating temperature.
Within the last five years, the author has conducted a survey of 251 transformers used by small and medium sized manufacturing industries. In this survey, the acidity and the interfacial tension were monitored. This allowed us to conclude that with minimal intervention, it was possible to maintain the boundary values as mentioned in the preceding paragraph. These results are reproduced in the Figures 2 and 3. We can see that the distribution of the interfacial tension and the acid number properties moves to the right with age. This could suggest putting in place a criterion system for this category of transformer users, based on years of service. This would allow us to follow the transformer's properties more closely. The interfacial tension and the acidity seem to stay unchanged until the transformer is 30 years old. At that point, the properties change rapidly. This could suggest oil degradation following the exhaustion of the antioxidant.
From the ASTM survey and the author's survey, we could be led to conclude that oil decay products start to be significantly produced when the interfacial tension and the acid number reaches the 35-30 and 0.06-0.08 range respectively, but sludge only starts to generate when the interfacial tension or the acid level is at around 24 and 0.1 respectively . If this fact is confirmed, then the reference values, the interfacial tension or the acidity, should be set to the values suggested in the first part of this paragraph.
The amount of water present in the oil will differ when the oil is in the bulk reservoir than when it is added to the transformer. In bulk, the oil should meet the ASTM D3487 norm. But, in the transformer, the measurement of dissolved water in oil is affected by the quantity of water already contained in the insulation. Since water is one of the principal catalysts of insulation aging, a norm for the amount of water in insulation should be applied along with the current oil dielectric strength test. The concentration of water is expressed in per cent per volume. The recommended value for in service transformers (more/larger) than 500 kv is 0.25 per cent, for the range of 230kv to 69 kv - 0.5 per cent, and 1 per cent-5 per cent for 35kv and less transformers. Several means are available to evaluate the relative humidity ( per cent) in the insulation .
Once the concentration of dissolved water in the oil and the temperature of the sample is known, it is possible to evaluate the dielectric rigidity of the oil and the relative humidity (RH) in the paper. To evaluate the RH, make sure the transformer is in a steady state condition for at least two weeks, otherwise the result is useless.
The evaluation of the dielectric rigidity should be undertaken according to the D1816 2mm norms. This norm specifies the use of the hemispheric electrode which is more sensitive to polar contaminants, including water, than the flat electrodes of D877 [6, page 265].
Table 1 summarizes the recommended values for the maintenance of oil filled transformers.
We can see in Table 2 that the greater the voltage of increases, the more demanding are the criteria.
In the case where the equipment history is available, we can establish thresholds of troublesome concentrations of dissolved gas.
When there is no historical data, we can use the information in Tables 1 and 2.
The levels in Table 1, from IEEE, Std C57.104-1991, are commonly used in North America. This table establishes a criterion on four levels, with a probability norm of 90 per cent, to determine the risks of weakness in a transformer.
Note 1: The quantities shown in Table 3 are based on a large power transformer (more than 100 MVA) containing thousands of gallons of oil. For equipment with a lesser volume of oil, faults will generate a larger concentration of gas (generally in the inverse ratio of their oil volume). Values in this table are based on equipment having 8333 imp gal of oil.
Interpretation of conditions in Table 3
Condition 1: Levels are normal.
Condition 2: Levels within this range are higher than normal.
Condition 3: Levels within this range indicate an elevated level of decomposition.
Condition 4: Levels within this range indicate excessive decomposition.
Although not as common, you could also use the age of the transformer in service, or the age since it was last degassed, in the selection of limits of normal concentrations. The older the transformer, higher the limits should be increased according to the following table.
Note about Table 4: Danger levels are reached when concentrations are about 5 - 10 times higher than the values in the table.
For the bushings you should use the values in Table 5.
Power factor 
The value of CHL in the oil is independent of the bushings' condition and is therefore a better representation of the condition of the windings. Two sets of data are a compilation of power factors (PF) measured on site. The first set represents the values for new transformers, whereas the second set is for used transformers. It is possible to obtain from DOBLE a compilation of power factors by manufacturer. When the equipment is new, this value can be obtained directly from the manufacturer.
According to DOBLE, this compilation indicates that the maximum power factor that can be measured for a new transformer is 0.5 per cent. A higher power factor should be justified by the manufacturer. For transformers in service the power factor should not have a value greater than 1.5 per cent.
Transformer insulation is a function of the voltage class and the basic impulse level of the winding.
Minimal value that a transformer should have when tested is a step function that can be approximated by the following equation:
Solid particles in suspension in insulating liquid affect the liquid's dielectric strength depending on the particlesÍ concentration, size and composition. Particles can be composed of metal: ferrous or non-ferrous; cellulose; porcelain; carbon etc., although most are composed of cellulose fibers and small carbon particles (<2 µm).
A particle count does not give any indication of their shape or nature.
It is possible to perform an evaluation of the metal in oil, but there is no standard on an acceptable amount of metal.
Estimation of life expectancy 
The evaluation of the life expectancy of a transformer is a key reason for having follow-up and diagnosis systems. This preoccupation is closely related to the need of the suppliers of electricity to predict the time of replacement in order to maximize the useful life of the equipment, as well as minimizing the risks of failure leading to power reliability problems.
The ultimate question to answer is how many years are left before the equipment has a failure?
The evaluation of the life expectancy is often subject to a number of erroneous interpretations [10, pages 92-98]. First, it is important to define what we agree upon as end of life.
The end of life is attained when the transformer is incapable of fulfilling its functions. Certain organizations distinguish between technical, planned and economical end of life. The tendency is to give too much importance to the technical end of life. It is rare that a transformer is replaced for only technical reasons; the main reasons to retire a transformer from service are related to costs. The operational expenses must be minimized. These reasons are of a planning nature (modification to load profile, voltage changes, etc.)
Second, we should distinguish between the life expectancy of the insulation and that of the transformer. It has often been the case where the transformer was kept in service several years after the insulation was classified obsolete. It is implicit that the life expectancy of the insulation is not that of the life expectancy of the transformer.
The technical life expectancy of a transformer is determined by several factors. It depends upon design, historical events, operating conditions, its actual state and future conditions.
Most of the present methods put too much emphasis on the condition of the insulating material. We could easily appreciate that not only temperature, load and water content have an effect on the capacities of a transformer to fulfill its functions but also the number of short-circuits, over-voltage, design weakness, repairs and moving, etc.. To be able to use a multi-factor evaluation, it is necessary to have an indepth understanding of the interrelations between the internal components. Once this is acquired, the historical information of the transformer will be needed. It is, therefore, important to gather the information as quickly as possible at the time in order to easily access it.
The eternal question is, "How long will my transformer last?". In order to answer this question we have extracted data from a survey of 251 transformers used by small and medium sized industries. From this survey, we have the transformer size and age profile with which we can estimate the life span of your transformer.
Figure 4 represents the transformer size profile of the survey. It indicates that most transformers in use by small to medium sized industries are in the 500- 2500 kVA range.
From Figure 5 the variations which we observe are probably due to cycles in the economy. Characteristically, these small to medium sized industries are more prone to these economical cycles.
The decrease in the number of transformers more than fifty years old is probably due to the closing of small and medium sized industries. If the decrease was cause by a mechanism failure, the curve would have been less abrupt. Instead, the decrease would be spread over two decades [11, Actuarial concept, pages 13-14]. The author is warning you not to use the curves from that reference for estimating the probable life span of your transformers.
From these we can conclude that transformers can operate for fifty years and probably more...
 Power Factor Test Data book, PFTDRB-871. Doble Engineering Co.
 Diagnostic du transformateur. Recherche du dysfonctionnement, Méthodologie de l'entretien, Expertise. Michel Belanger, June 1999.
 PERCEPTION 4.1, Follow-up and diagnosis software for oil insulated equipment, SEIDEL, www.seidel.qc.ca .
 Insulating Oil Qualification and Acceptance Tests from a user's Perspective", ASTM Publication STP 998 "Electrical Insulating Oil", H. G. Erdman editor, 1988.
 Moisture Equilibrium in Transformer Paper-Oil System, Y. Du, M. Zahn, B.C. Lesieutre, A.V. Mamishev, S.R. Lindgren, IEEE, DEIS Jan/Feb 1999 - Vol. 15, No. 1.
 A Guide to Transformer Maintenance, Transformer Main-tenance Institute, Kelly J.J., SD Myers, 1981.
 Bulletin MS-25, A Guide to the Interpretation of Transformer Fault Gas Data, J. E. Morgan, Morgan & Schaffer Co.
 IEC 61464:1998, Insulated Bushing, Guide for the Interpretation of Dissolved Gas Analysis (DGA) in Bushings Where Oil is the Impregnating Medium of the Main Insulation (Generally Paper).
 Status and Trends in Transformer Monitoring. C. Bengtsson, ABB Transformer, Ludvika, Sweden, IEEE transaction on Power Delivery, Vol. 11, No. 3, July 1996.
 IEEE Guide for Loading Mineral-Oil-Immersed Transformers, C57.91-1995
 Distribution Reliability Centred Maintenance: Quantifying Common Sense, F. L. Kaempffer, Electricity Today, April 1999.
 Electrical Insulating Oils Part I: Characterization and pre-treatment of New Transformer Oils. per centA. Sierota, J. Rungis, IEEE Electrical Insulation Magazine, January /February 1995-Vol. 11, No.1.
Michel Belanger is the owner of SEIDEL Inc. He is an electrical power systems consultant and designer of diagnosis systems for the industry. You can reach him at (418) 822-3561. ET