(Tech al Trends ] Yasuyuki Akashi Expected Residual Life Estimation Techniques for High Voltage Rotating Machines Rotating machine, Insulation slagnosis, Expected estimation of residual te After more than twenty years of operation, many electrical facilities are reaching their time of renewal. However, the general trend of such capital invest- ment has faced the problem of budget cuts. In recent years, the load to the elec- trical facilities has increased and the maintenance and inspection of such facili ties have become increasingly important, We have developed a technique of estimating expected residual life for elec- trical facilities. For high voltage and extra high voltage class rotating machines Abstract 06 (6kV, 11kV class), we are collecting data regarding nondestructive insulation breakdown. Through the database of stat imate the residual life expectancy for rotating machines. possible to e: iagnosis and tical diagnostic data, 1 Preface Many electrical facilities manufactured after the 1970s (high economic growth period) are stil in operation in Japan. However, financial investment in plant and equipment has tended to be restrained. Recently, the load to the electrical facilities has increased further because of an increase in the facility availability. Under such circumstances, the needs for maintenance and inspection are further increased for electrical facilites. High voltage or extra high voltage class gener- ators and motors (“rotating machines” hereafter) are essential to electrical facilities. If rotating machines, become inoperable, the result cannot be limited to only the shutdown of a plant and the remaining life- line may be seriously affected, Most rotating machines have been used for more than twenty years and their users are required to take adequate counter measures such as extension of life for rotating machines or renewal This paper introduces the development of our diagnostic techniques (estimation of expected reside ual life) for the stator windings of 6.6kV rotating machines. 2. Insulation of Windings for Rotating Machines 24 Diagnosis Most causes of failure in rotating machines are related to their windings. When winding deteriorates and insulation breakdown occurs, social and eco- nomic losses are substantial. It is therefore essen- tial to assess the state of deterioration in winding insulation so that adequate countermeasures can be taken, For early detection of deterioration in windings, we are promoting diagnostic services for winding insulation of rotating machines. Fig.1 shows our (ur diagnosis ear ig shown, One complete set of diagnostic ‘equipment is loaded onto this car, A single car is capable of slagnosing in the feldiagnostic car. Table 1 shows the items of diagno- sis. 2.2 Factors of Deterioration Deterioration can be caused by any one, or combination, of the following factors: thermal, elec- trical, mechanical, and environmental. Due to these factors, voids and/or cracks are generated within the layers of winding insulation. In a location where voids and/or cracks are produced, partial discharges [Dy tems of insulation Diagnosis Test items and contents of diagnosis are shown, These are the items of ciagnos's applicable to general tating machines. Each iagnostic item indicates that there is an effect of evaluation in the contents of related diagnosis, Major contents Testitem err DC absorption |insuationresistance fer [Contamination test minute (a) land moisture Jabsormion Polarity index Tan Stest [tan dy (tan 6 of base) Contamination [c, (oase capacitance) [absorption tans oid ana Jextolaton ACI ‘AC curentiest | a1 voi ana Jextoion P, (Sudden-inerease point of primary curent Pz (Suaden-inerease point of [secondary curent) Partial scharge [Q, (Maximum electic charge [Local test dant) eteroration lv Power-‘requency sparkover vattage) Wining Insulation "ayer (a) Cross-section of Getorioratoniree sample (0) Cross-section ofa sample Stor dateroration ‘A cross-section of rotating machine cols is shown for investiga tion. Compared with @ deterioraion-free tem, a sample after Undergoing thermal deterioration shows voids In insulation layer. ‘Such a condition suggests that there has been a progression of deterioration. are generated. Partial discharges give rise to ero- sion in the insulation layers, thus leading finally to insulation breakdown Fig. 2 shows cross sections of a coil observed before and after thetmal accelerated deterioration test. These cross sections were viewed as part of the deterioration mechanism study we promoted for rotating machines. After the occurrence of deterio- ration, a large void can be perceived. Starting with such a void, partial discharges are generated and ‘we surmise that the insulation layer is then eroded by these partial discharges 3. Investigation into Expected Residual Life Estimation Approach 3.1. Expected Residual Life Estimation Study We investigated the estimation formula for residual breakdown voltages that will be introduced below. The residual breakdown voltage denotes a insulation breakdown voltage to be measured after the deterioration in windings of rotating machines. The breakdown voltage in windings has sufficient performance for the operation of a rotating machine for a long time. There is, however, some dispersion according to the difference in insulation configura- tion or specifications. Therefore, the residual breakdown voltage Used in this section should mean the rate of break- down voltage before and after deterioration. It is ditt ficult to obtain test data of the breakdown from an’ actual machine. This is reason, we have tried to obtain data from actual machines and also data from accelerated deterioration testing with the use of @ model coil as shown in Fig. 3. Based on this data, we could draw an estimation formula for resid- al breakdown voltages, for example on the basis ‘An external appearance of model coils, used for the testing is ‘shown, These cols are manutactured in the same manner as ‘those of actual machines. In order to set up the conaition more like that ofan actual machine, mimic core slots are mounted.08 of, multiple regression analysis and a neural net work method. Fig. 4 shows an outline of the accel- erated deterioration test for model coils. 3.2. Correlation between Diagnostic Parame- ters and Residual Breakdown Voltage Fig, 5 shows the correlation between the max: imum electric charge quantity “Qm” (operating volt- age) of discharge and the residual breakdown volt age. Value Qm is one of the insulation diagnostic parameters and this value is believed to have the closest relationship with the residual breakdown Pe] [LD Acceterated Deterioration Test for Model Coils (Outlined accelerated deterioration testis chown for the model calls, Both thermal and electrical sresses are applied to the ‘model coils. Ths testis carried out for the presumption of oper- ational life by examining changes in electrical characteristics and those in breakdown voltages, both considered to be caused by deterioration. 100 20 [Che Rpprox. 205 elden cautionary standard S000p0| S00 “00 "a0 100,000 Maximum electric earge quaity of charge Gm (pC) Correlation between Maximum Electric Charge ‘Quantity of Discharge and Ri down, vol Correlation between maximum electric charge quantlly and re sidual breakdown voltage is shown because itis one of the pa- rameters for insulation diagnosis, As the maximum electric charge quantity of discharge increases, there is a tendency of a decrease in residual breakdown volage. If our judgment stand- ard (caution needed) of maximum electric charge quantity of dis- charge 5000pC is exceeded, breakdown is possible. voltage of coils. Along with an increase in Qm, a tendency can be confirmed in the reduction of resid- al breakdown voltage. Generally, the breakdown voltage needed to assure operation of rotating machines Is 2E + 1kV (E: rated voltage) and we define the machine life with this value, The diagnos- tic judgment standard with Value Qm comes in a variety of figures among manufacturing companies. At our company, 5000pC is regarded as the judg- ment standard (caution needed) although there are some differences among various machine types. According to the result shown in Fig. 5, we recog- nize that the residual breakdown voltage is lowered to a level of the standard value when Value Qm is greater than 5000pC. 3.3. Estimation Formula for Residual Breakdown Voltages of High-Voltage Rotating Machines Using various diagnostic parameters acquired, we have established an estimation formula for resid- val breakdown voltages of high-voltage rotating machines. When we use this estimation formula, it is possible to estimate a residual breakdown volt age without being influenced by environmental fac- tors such as humidity. If a comparison is made between data from actual machines and values from estimation formula, itis possible to expect the evaluation of expected residual life. Fig. 6 shows an example of the estimation for- mula, There isa favorable correlation between actu- ally measured values and theoretical values led from the estimation formula. ‘Theoret value Lie: Approx 20% Actually measured value Resi! bresksonm vag ( Tntveductory Diagram of Estimation Formula This diagram shows a comparison between col breakcown vole age estimated from diagnosis and actual breakown voltage.4 Postscript This paper introduced processes of expected residual lite estimation with the use of multiple re- gression analysisandaneuralnetwork. Improvement of accuracy is an important subject for the estima- tion formula. In this connection, we will continue to collect diagnostic data from actual machines and make efforts to optimize the estimation formula. Rotating machines may involve some peculiar prob- lems such as layer short, slot discharges, vibration sparking, and so on. They are significant phenome- na that can give rise to serious insulation breakdown in windings in a short amount of time. Since no pre- liminary sign can be indicated in advance, itis ditfi- cult to carry out preventive maintenance. We have been developing the diagnostic techniques to re- solve these challenges. Going forward, we will make every effort to offer high-quality services to our customers through further reinforcement of diagnostic techniques. All product and company names mentioned inthis paper ar the trademarks andior service marks of ther respective owners.