review-of-condition-assessment-of-power-transformers-in-service
advertisement
F E A T U R E A R T I C L E Review of Condi Re diti tioon Ass sseess ssm men entt of Power Transformers in Service Key Words: Transformer insulation, condition assessment, failure statistics, oil testing, dissolved gas analysis, partial discharge (PD), power factor, dielectric spectroscopy, recovery voltage, winding movement detection T ransformers are required throughout modern interconnected power systems. The size of these transformers ranges from as low as a few kV kVA A to overr a few hun ove hundre dred d MV MVA, A, wit with h rep replac lacemen ementt cos costs ts ran rangin ging g from a few hundred dollars to millions of dollars. Power M. Wang and A.J. Vandermaar transformers are usually very reliable, with a 20-35 year design life. In practice, the life of a transformer can be as long as 60 years with appro appropria priate te maint maintenance enance.. However, the in-service failure of a transformer is potentially dangerous to utility personnel through explosions and fire, fir e, pot potent ential ially ly dam damagi aging ng to the env enviro ironme nment nt thr throug ough h oil leakage, is costly to repair or replace, and may result in signif sig nifica icant nt los losss of rev revenu enue. e. In a lar large ge pub public lic pow power er uti utilit lityy, the number of transformers in the subtransmission and transmission network (excluding the lower-voltage distribution network) can be from a few hundred to over one thousand (69 kV to 500 kV). As transfor transformers mers age, age, their their internal internal condit condition ion degrade degrades, s, which increases the risk of failure. Failures are usually trig- The University of British Columbia Vancouver,, Canada Vancouver Cana da ger gered by sever severeeshort-circuits, conditio cond itions, ns, such as lig lightni htning ng stri strikes, kes, switchswit chinged transients, short-cir cuits, or other incidents. When the transformer is new, it has sufficient electrical and mechanical strength to withstand unusual system conditions. As transformers transfor mers age, their insulati insulation on strength can degrade to the point that they cannot withstand system events such as short-circuit short-ci rcuit faults or transient overvoltages. overvoltages. To pre preven ventt the these se fa fail ilure uress an and d to ma maint intai ain n tra transf nsform ormers ersin in good operating condition is a very important issue for utilities. utilitie s. Traditio raditionally nally,, routine preventati preventative ve maintena maintenance nce programs combined with regular testing were used. With deregula dere gulation tion,, it has beco become me inc increas reasingl inglyy nece necessar ssaryy to redu reduce ce maintenance maintena nce costs and equipment inventori inventories. es. This has led to reductions in routine maintenance. The need to reduce costs co sts has hasal also so res result ulted ed in re reduc ductio tions ns in spa spare re tra transf nsfor ormer merca ca-- The cha change nge to cond conditi ition-b on-based asedmain maintena tenance nce has resul resulted ted in the re reduc ductio tion, n, or eve even n eli elimi minat nation ion,, of rou routi tine ne ti timeme-bas based ed maint ma intena enance nce.. Ins Instea tead d of doi doing ng ma main inten tenan ance ce at a reg regula ularr interval, maintenance maintenance is only carried out if the condition of the equipment requires it. Hence, there is an increasing need for better nonintrusive diagnostic and monitoring tools too ls to as asses sesss the int intern ernal al con condit dition ion of the tra transf nsform ormers ers.. If ther th eree is a pr prob oble lem, m, th thee tr tran ansf sfor orme merr ca can n th then en be re repa pair ired ed or replaced before it fails. Many Ma ny tes testin ting g an and d mon monito itori ring ng tec techni hnique quess hav havee bee been n use used d by uti utili litie ties. s. Thi Thiss ar artic ticle le rev revie iews ws the exi existi sting ng mon monito itori ring ng an and d diagnosticc methods and future trends. diagnosti pacity paci ty an and d inc increa reases sesin in ave averag ragee loa loadi ding ng.. The There re is al also so a tr trend end in the theind indust ustry ry to mov movee fr from om tra tradi ditio tionaltimenaltime-bas based ed mai mainte nte-nance programs to condition-based maintenance. These Transformer failure can occur as a result of different causes and conditions. Generally, transformer failures can be defined as follows [1]-[2]: 12 Powertech Labs Inc. Powertech Surrey, B.C., Canada K.D. Srivastava There is an increasing need for better nonintrusive diagnostic and monitoring tools to assess the internal condition of transformers. changes occur at a time when the average age of the transformers form ers in service service is increas increasing ing and approa approachi ching ng the end of nominal design life. Power Transformer Failures and Problems 0883-7554/02/$17.00©2002IEEE IEEE Electrical Insulation Magazine Tablee I. Typ Tabl Typical ical Cau Causes ses of Tran Transfor sformer mer Fail Failure uress Internal External Insulation deterioration Lightning strikes Loss of of wi winding cl clamping System sw switching op operations Overheating System overload Oxygen System faults (short circuit) Moisture Solid cont contamin aminati ation on in the insu insulat lating ing oil Partialdischarge Partial discharge Design gn & manu manufact facture ure defec defects ts Windingresonance Winding resonance any fo any forc rced edou outa tage gedu duee to totr tran ansf sfor orme merr da dama mage gein inser servi vice ce (e.g., winding damage, tap-changer failure) troub tro uble le th that at re requ quir ires es re remo mova vall of th thee tr tran ansf sfor orme merr fo forr re re-turnn to a rep tur repair airfac facili ility ty,, or whi which ch req requi uires resext extens ensive ivefie field ld repair rep air(e. (e.g., g.,ex exces cessiv sivee gas gaspro produc ductio tion, n,hig highh moi moistu sture relev lev-els). Transformer failures can be broadly categorized as elec trical, mechanical, or thermal. a failure can be internal or external. Table I The Table lists cause typicalofcauses of failures. In addition to failures in the main tank, failures can also occur in the bushings, in the tap changers, or in the transformer accessories. The failure pattern of transformers follows a “bathtub”curve,asshowninFig.1.Thefirstpartofthecurveis failure due to infant mortality; the second part of the curve is the constant failure rate; and the last part of the curve is failure due to old age. In addition to normal aging, a transformer may develop a fault that results in faster-than-normal aging, resulting in a higher probability of failure. Power transformers have proven to be reliable in normal operation with investment a global failure rate of 1 capacity – 2 percent per year. The large in generating after the Second World World War War that continued into the early 1970s has resulted in a transformer population that, in theory,, is fast approaching the end of life [3]. The end of theory life li fe of a tr tran ansf sfor orme merr is ty typi pica call llyy de defi fined ned as the lo loss ss of me me-chanical strength of the solid insulation in the windings. These power transformers are at the last stage of the “bathtu “ba thtub” b” cur curve. ve. The Theyy are exp expect ected ed to hav havee an inc increa reasin singg failure rate in the next few years. A survey [4] reports that the main main causes (51 percent of transformer failures in a five-year period) were due to the following problems: moisture, contamination and aging which caused the damagetothetransformerbushingscausedbylossofdielectric strength of the internal insulation. An American utility reported four single-phase EHV autotr aut otrans ansfor former mer fai failur lures es due to tra transf nsform ormer er win windin dingg res res-onance [5]. All of the failures involved the breakdown of the no-load tap changers immediately after the transmission system was energized. The utility also experienced three 25/765 kV, 500 MVA generator step-up transformer failures and two 765 kV, 80 MVA reserve auxiliaryy tra iar transf nsform ormer er fai failur lures; es;allof allof the thefai failu lures reswer weree die dielec lectri tricc in nature [6]. Another survey done by a CIGRÉ working group on failures in large power transformers [1] found that about 41 percent of failures were due to on-load tap changers (OLTC) (OL TC) and about 19 percent were due to the windings. The failure origins were 53 percent mechanical and 31 percent dielectric. On transformers without on-load tap changers, 26.6 percent of failures were due to the windings, 6.4 percent were due to the magnetic circuit, 33.3 percent were due to terminals, 17.4 percent were due to thee ta th tank nk an andd di diel elec ectr tricflu icfluid id,, 11 pe perc rcen entt we were re du duee to ot othe herr accessories, and 4.6 percent were due to the tap changer. changer. Figure 2 shows the with percentage distribution for power transformers on-loadfailure tap changers. Another report presents transformer failure data in South Africa [7]. This failure analysis was based on 188 Typical Transformer Failure Pattern s re lui a F f o r e b m u N transformer’s strength damage dam age to theinternal windin win dinggdielectric or decompr decompressi ession ontoofdecrease, the win the windding under short circuit forces, or November/December 2002 — Vol. 18, No. 6 Years in Service Figu re 1. Bat htu b fai lur e cur ve. 13 power transformers in the voltage range of 88 kV to 765 kV with ratings from 20 to 800 MVA. The failure modes are shown in Fig. 3. Failure statistics for large transformers that had been inservicebetween15and25yearsareshowninFig.4[4]. The above surveys and research results indicate that load tap changers, windings, insulation aging, and contamination are the key sources of transformer failures. Another paper [8] indicates that the average number of fa fail ilur ures es ov over er a fo four ur-y -yea earr per perio iod d (1 (197 975 5 to 19 1979 79)) wa wass 2. 2.6 6 failures per year per 100 transformers. Thee co Th cost st an and d ti time me to re repa pair ir an and d re repl plac acee a po powe werr tr tran anssformer for mer is ver veryy subs substan tantia tial. l. The rep repair air and replac replaceme ement nt of a 345/138 kV transformer normally requires about 12 15 months. If a spare is available, available, the time needed for replacementofafailedunitisintherangeof8-12weeks. Transformer Life Management Transformer life management has gained an increasing ac ing acce cept ptan ance ce in th thee pa past st 10 - 15 ye year ars, s, du duee to ec econ onom omic ic and technical reasons. The fundamental objective is to Core Terminal Accessories On Load Tap Changers Tank/Fluid Windings Fig ure 2. Perc ent age fa fail ilure ure of po wer tr an sf or mer s (CIGRE survey) [1]. Core Aging Others Short Circuit Figu re 3. Perce nta ge of fai lur es of pow er tra nsf orm ers (South Africa) [7]. 14 ° Lightning/Switching Transients Tap Changer promote the longest possible service life and to minimize lifetime operating costs. The importance of this issue [9]-[15] has led to a lot of research in this area. In general, transformer transformer life is equal to the insulation life, which whi ch dep depend endss on mec mechan hanic ical al str streng ength th and ele electr ctrica icall integrity. Insulation degradation consists of hydrolytic, oxidative, and thermal degradation. The aging and life of a tr tran ansf sfor orme merr ha hass be been en de defi fine ned d as th thee li life fe of th thee pa pape perr insulation [10]. Several aging mechanisms were identified as follows: applied mechanical forces thermal aging (chemical reactions) voltage stresses contamination. The transformer is subjected to mechanical forces due to transportation, electromagnetic forces caused by system sho short rt cir circui cuits,and ts,and inr inrush ush cur curren rent. t. Vi Vibra bratio tion n and the therrmalforcess gen malforce genera erated ted by dif differ ferent ent ther thermalexpans malexpansion ion rat rates es in different materials cause long-term degradation of the paper pap er.. The even eventua tuall die dielec lectri tricc fai failur luree may occ occur ur whe when n the mechanical forces rupture the insulation. The compressive mechanical forces on the cellulose paper can cause material flow and cause clamping pressure to reduce. Thus the aging of paper insulation determines the ultimate life of the transformer, although other factors may contribute to earlier failure. Thermal aging of transformer insulating materials is associated with the chemical reactions occurring within the mat materi erials als.. The These se che chemic mical al rea reacti ctions ons are cau caused sed by pyrolysis oxidation and hydrolysis, and are accelerated by increased levels of temperature and of the oxygen and moistur moi sturee con content tents. s. Ass Associ ociate ated d wit with h the che chemic mical al rea reacti ction on of the cellul cellulose ose paper is a reduction in the mecha mechanical nical properties. The paper insulation becomes brittle to the point of almost falling apart, but it still retains an acceptable level of dielectric strength. The temperature of a transformer has a major impact on the life of the insulation. Continuous on-line monitoring of the transformer oil temperature along with a thermal model of the transformer can give an estimate of the loss of life of the transformer due to overheating. Current industr ind ustryy stan standar dards ds lim limit it max maximu imum m all allowa owable blehot hot spot spottem tem-peratures in transformers to 140 C with conventional oil/paper insulation. End of life may be dictated by any one factor or by a combination of factors. Much attention has been given to paper paper ag aging ing as a cau cause se of transf transform ormer er fai failur lure. e. Whi While le it is un undo doub ubte tedl dlyy a fa fact ctor or in re redu duci cing ng li life fe,, it do does es no nott au auto to-matically lead to failure; some other influence is normally required, such as mechanical shock. In industry loading guides (e.g., IEC, ANSI, and IEEE) the principal factor for end of life relates only to the transformer’ss the former’ therma rmall fac factor tor.. A cla classi ssical cal met method hod of calculating the remaining life life of a transformer has been the Arhennius-Dakin formula: IEEE Electrical Insulation Magazine Remaining life = AeB//TT Miscellaneous where A = initial life; B = constant, depending on the proper pro pertie tiess of the mat materi erial al stud studied ied;; and T = abs absolu olute te tem tem-perature in K. A more comprehensive approach is clearly clearly needed to evaluate the remaining life of a transformer as a whole. The other factors affecting the probability of failure are not as easily quantified as thermal aging. To assess the overall condition of a transformer reliably, reliably, several moni° toring techniques are used and are under investigation. The most common monitoring/testing methods used for transformer condition assessment are given in [11], [16]-[76]. The traditional routine tests are: transformer ratio measurement, winding resistance, short-circuit impedance and loss, excitation impedance, and loss dissipation factor fac tor and cap capaci acitan tance,as ce,as wel welll as app applie lied d and ind induce uced d potentia ten tiall test tests. s. The These se test testss usua usually lly giv givee inf inform ormati ation on on fau faults lts in windings, winding conductor and joint problems, winding deformation, oil moisture and contamination, and dielectric problems. Special tests include partial discharge char ge measur measurement, ement, freque frequency ncy respon response se analy analysis, sis, vibra vibra-tion analy analysis, sis, infr infrared ared examin examination ation,, voltag voltagee recov recovery ery,, and degree of polymerization. These detect problems such as local partial discharge, winding looseness and displacement, slack winding and mechanical faults, hot spot on conn co nnec ecti tion on,, mo mois istur turee in pa pape perr an and d ag agin ing g of pa pape perr, as we well ll as insulation degradation. Oiltests are use used d exte extensi nsivel velyy. The Theyy con consis sistt of dis dissol solved ved gas ana analys lysis is (DG (DGA) A) wit with h rat ratio io ana analys lysis, is, fur furan an ana analys lysis, is, water content, resistivity resistivity,, acidity, acidity, interfacial tension t ension (IFT), and dissipation factor (DF). These detect oil incipient faults, overheating, aging of paper, dryness of oil-paper, and aging of oil. Life assessment of large transformers may be performed for the following reasons [12]: to monito monitorr the condition of transf transformers ormers and provide an early warning of faults to diagnose problems when transformers exhibit signs of distress or following the operation of protection equipment to de dete term rmin inee wh wheth ether er a tra transf nsfor orme merr is in a sui suita tabl blee co conndition to cope with unusual operating conditions to ob obta tain in re refe fere renc ncee re resu sult ltss to as assi sist st in th thee in inte terp rpre reta tati tion on of subsequent tests to ass assist istin inpla planni nning ngthe therep replac lacemen ementt stra strateg tegyy fora pop popuulation of transformers to satisfy the requirements for insurance coverage. Testi esting ng and mon monito itorin ring g met method hodss are rev review iewed ed in det detail ail in the next section. Insulation Aging Overvoltage Core Insulation Failure Winding Deformation Due to Short Circuit Forces Contamination of Insulation Figur e 4. Fail Failure ure of tra nsf orm ers 15 to 25 yea rs old [4] . analysis, analys is, suc such h as an expe expert rt syst system em cap capabl ablee of pro provid viding ing an assessment of equipment condition and suggested actions. There are a variety of tools available to evaluate the condit con dition ion of tra transf nsform ormers ers [2 [25], 5], [5 [55], 5], [6 [64]4]-[6 [65], 5], [77]-[ [77 ]-[81] 81].. The Theyy canbe sepa separat rated ed int into o tra tradit dition ional al dia diagno gnosstic met method hodss tha thatt hav havee seen wid widesp esprea read d use for man manyy yea years rs and nontraditional methods that range from methods that ar that aree st star arti ting ng to be us used ed to me meth thod odss th that at ar aree st stil illl in th thee research stage. Traditional Diagnostic Methods OIL TESTING Testing of the winding insulating insulating oil is one of the most common tests used to evaluate the condition of transformtransformerss in se er serv rvic ice. e. Th Ther erma mall an and d el elec ectr tric ical al fa faul ults ts in th thee oi oill le lead ad to degradation degrada tion of the oil. Dissolved Gas Analysis Insulati Insul ating ng oils under abn abnorma ormall elec electric trical al or ther thermal mal stress str esses es br breakdown eakdown to li liber berat atee sma small ll qua quanti ntitie tiess of ga gases ses.. The Monitoring and Diagnostic Methods composition ofof these gases is sdependent upon, itthe type of fault fa ult.. By mea means ns disso di ssolve lved d ga gas analy ana lysis(DGA) sis(DGA), is pos possib sible le to dist distingu inguish ish faul faults ts such as part partial ial disc discharg hargee (co (corona rona), ), over over-heati hea ting, ng, an and d ar arci cing ng in a gr grea eatt var variet ietyy of oil fi fill lled ed equ equipm ipment ent.. A numb number er of samp samples les must be take taken n over a peri period od of time to discern trends and to determine the severity and progresprogression of incipient faults. The gases in oil tests commonly evaluate the concentration of hydrogen, methane, acetyacetylene, ethylene, ethane, carbon monoxide, carbon dioxide, nitrogen, and oxygen. The relative ratios and the amount of gas detected in the sample are used to detect problems with the insulation structure [82]-[90]. Cellulosic Cellul osic Decompo Decomposition sition—The ther thermal mal deco decompos mpositi ition on of oil oil-im -impreg pregnate nated d cell cellulos ulosee insu insulati lation on prod produces uces car carbon bon oxides id es (CO (CO,, CO2) an and d som somee hyd hydrog rogen en andmetha andmethane ne (H2, CH4) Generally Gener ally speaki speaking, ng, the term “moni “monitoring toring”” descri describes bes a basicparameter measur measurement ement with thresho threshold ld alar alarms. ms. The term “dia “diagnosti gnostics” cs” indic indicates ates the additi addition on of sophis sophisticat ticated ed due to the oil. —Minerall transformer oils are mixOil Decompos Decomposition ition—Minera tures tur es of man manyy di diffe fferen rentt hyd hydroc rocar arbon bon mol molecu ecules les,, an and d the de- November/December 2002 — Vol. 18, No. 6 15 comp composit osition ion proc processes essesfor for thes thesee hydr hydrocar ocarbons bonsin in therm thermal al or electrical electric al faults are complex. Heating Heating the oil produces ethylene (C2H4) as the principa principall gas. Inform Inf ormati ation on fro from m the ana analys lysis is of gas gasses ses dis dissol solved ved in insula su lati ting ng oi oill is on onee of th thee mo most st va valu luab able le to tool olss in ev eval alua uati ting ng the health of a transformer and has become an integral part of preventive maintenance programs. Data from DGA can provide: advanc adv anced ed wa warni rning ng of dev develo elopi ping ng fau faults lts moni mo nito tori ring ng th thee ra rate te of fa faul ultt de deve velo lopm pmen entt conf co nfir irm m th thee pr pres esen ence ce of fa faul ults ts a mea means ns for co conve nveni nient ently ly sch schedu eduli ling ng rep repai airs rs monito mon itori ring ng of con condit ditio ion n dur during ing ove overlo rload. ad. DGA DG A dat dataa by it itsel selff doe doess not al alwa ways ys pr provi ovide de suf suffic ficie ient nt in in-formation on which to evaluate the integrity of a transformer system. Information about its manufacture and the history of a transformer in terms of maintenance, loading prac pr actic tice, e, pre previo vious us fau faults lts,, an and d so on ar aree an in integ tegra rall par partt of the information informa tion required to make an evaluatio evaluation. n. Gener Ge neral ally ly,, the there re ar aree thr three ee ste steps ps in invol volved ved.. The fi first rst ste step p is to establish whether or not a fault exists. In-service transformers always always have some fault gases dissolved in their oil. Only when these levels exceed some threshold value is a fault fau lt suspe suspected cted.. Seve Several ral reco recommen mmended ded safevalues have havebeen been ity of the fault is established by comparison of the levels of gas gases es wit with h thr thresh eshold old lev levels els and the their ir rat ratee of gen genera eratio tion. n. At least two consecutive samples are needed to calculate rates of fault generation. A list list of key gases gases and their relate related d faults faults are shown in in Table III. Fo Forr a deta detaile iled d disc discussi ussion, on, cons consult ult IEEE Std. C57.104C57. 104-1991 1991,, “IEE “IEEE E Guid Guidee for the Inte Interpre rpretati tation on of Gases Generated in Oil-Immersed Transformers.” published. Somestep of these are listed Table II. of fault. Two The second is to determine the type Two methods most commonly used are the key gases and gas ratios [17]-[18], [21]-[23], [27], [29]-[30], [36], [39], [45], [56], [58], [60], [76]. The first involves plotting all the total dissolved combustible gas (TDCG) as a percentage of their total in a histogram. Each fault type will give a distinctive pattern characterized by a key gas, generally era lly the mos mostt abun abundan dant. t. For exa exampl mple, e, hig high h lev levels els of hydrog dr ogen en wi with th lo low w le leve vels ls of oth other er ga gase sess ar aree ch char arac acte teri rist stic ic of partial discharge. The ratio method requires the calculation ti on of ra rati tios os of ga gase sess am amon ong g ea each ch ot othe herr, su such ch as me meth than anee to hydrogen. Three or four such ratios are used for diagnosi no sis. s. Th Thee mo most st wi wide dely ly use used d ar aree Ro Roge ger’s r’s ra rati tios os;; the sev sever er-- pounds are produced andisdissolved inthe thestrength oil. Theof presence of these compounds related to the paper as measured by its degree of polymerization (DP). Furan and phenol measurement in oil is a convenient, noninvasive method to assess the condition of the paper insulation. Transformer oil samples should be analyzed for furans and phenols when one or more of the t he following conditions exist: overh ove rheat eating ing or ove overl rload oadin ing g of the tra transf nsform ormer er high hi gh le leve vels ls of ca carb rbon on mo mono noxi xide de or ca carb rbon on di diox oxid idee rapid rapi d decr decrease easeof of inter interfaci facial al tensi tension on with without out a corr correspo espondnding in g in incr crea ease se in ac acid id nu numb mber er sudd su dden en da dark rken enin ing g of th thee oi oill an and d a su sudd dden en in incr crea ease se of th thee mois mo istu ture re co cont nten entt of th thee oi oill Insulating Oil Quality The condition of the oil greatly affects the performance and the service service life of transformers. transformers. A combina combination tion of elecelectrical, physical, physical, and chemical tests is performed to measure the ch chang angee in the ele electr ctric ical al pro proper pertie ties, s, ext extent ent of con contam tamina ina-tion, and the degree of deterioration in the insulati insulating ng oil. The res result ultss ar aree use used d to est establ ablish ish pre preven venti tive ve ma maint intena enanc ncee procedure proc edures, s, to avoi avoid d costl costlyy shutdo shutdowns wns and premature equipment failure, failure, and extend the service life of the equipment. There is a multitude of tests available for insulating oil. The most commonly used, and their significance, are list li sted ed in Tab able le IV. Thr Thresh eshold old lev levels els for the these se test testss are spec speciified in ASTM D3487 for new oils and IEEE Guide 637-1985 for service oils. As paper degrades, a number of specific specific furanic com- Table Tab le II. Rec Recomm ommend end Lim Limits its of Diss Dissolv olved ed Gase Gasess Gas Dornenburg/Stritt IEEE Bureau of Reclamation Age A ge Compensated Hydrogen 200 100 500 20n + 50 Methane 50 120 125 20n + 50 Ethane 35 65 75 20n + 50 Ethylene 80 50 175 20n + 50 Acetylene 5 35 7 5n + 10 Carbon Monoxide 500 350 750 25n + 50 500 0 TDCG* (total of above) Carbon Dioxide n 720 6000 2500 110n + 71 710 0 10000 100n + 15 1500 00 = yea years rs in ser servic vicee *Total *Tot al disso dissolved lvedcomb combusti ustiblegas ble gas 16 IEEE Electrical Insulation Magazine transf tran sfor orme mers rs ov over er 25 ye year arss ol old. d. Furan Fur an measur measurement ement is still a relat relatively ively new techni technique, que, and its interpretation is dependent on many operational and his histor torica icall fac factor tors. s. How Howeve everr, the gui guidel deline iness in Tabl ablee V provide some assistance. The degree of polymerization (DP) estimated from fura fu ran n an anal alysi ysiss re rela lates tes to th thee av aver erag agee va valu lue. e. Pa Pape perr in tr tran anssformers forme rs usuall usuallyy does not age unifor uniformly mly,, and there will be areas where degradation is more severe. POWER FACTOR TESTING The insulation power factor is the ratio of the resistive current component to the total leakage current under an applied voltage. Power factor measurement is an important ta nt sou sourc rcee of da data ta in mon monit itor orin ing g tr tran ansfo sform rmer er an and d bus bushi hing ng conditions. In general, power factor measurement equipment comes with three basic modes of operation: a) grounded specimen test (GST); b) GST guard; and c) ungrounded specimen test (UST). The three measurement modes allow measurement of the current leaking back to thee tes th testt set on ea each ch le lead ad,, in indi divi vidua duall llyy an and d tog togeth ether er.. In ge genneral, a power factor of less than 1 percent is considered good; 1-2 percent is questionable; and if it exceeds e xceeds 2 percent, action should be taken. Practically, the evaluation is not only based on a single power factor data point but is also based on the history of the change in power factor. Measurement of a transformer’s capacitance and pow po werfa erfact ctorat orat vo vollta tag ges upto 10kV (a (att 50or 60 Hz)has long been used as both a routine test and for diagnosis. The acceptance value should be less than 0.5 percent. Reference Refe rence [60] categ categorize orizess the interw interwindin inding g power facto factorr as the following: dry < 0.5 percent; medium < 1.5 percent ce nt;; an and d we wett > 1. 1.5 5 pe perc rcen ent. t. Theeval Theevalua uati tion on al also so ta takesackesaccount of the transformer’s power factor history. history. The test requires an outage and isolation of the transformer. The tests can be done, respectively respe ctively,, on high-voltage winding to ground, high- to low-voltage winding, low-voltage winding to ground, high- to tertiary-voltage winding, low- to tertiary-voltage tertiary-voltage winding, and the tertiary-voltage winding to ground insulation. It is used to detect problems with the transformer bushings and to evaluate the condition of the oil/paper insulation structure [17]-[18], [22], [35], [39], [45], [56]-[57], [60], [91]. WINDING RESISTAN RESISTANCE CE Winding resistance is used to indicate the winding conWindingresistance ductor and tap changer contact condition. The test requires an ohmmeter capable of accurately measuring resi re sista stanc ncee in th thee ra rang ngee of 20 Ω do down wn to fr frac acti tion onss of an Ω. Winding Windi ng resistance varies with oil temperature. During the test, the temperature should be recorded. For future comparison compa risons, s, the resistance resistance should be conver converted ted to a reference temperature. Measurement of transformer winding resistance requires an outage and isolation of the transf tra nsform ormer er.. Vari ariati ations ons of mor moree tha than n 5 per percen centt may ind indiicate a damaged conductor in a winding [22]. Table Tab le III III.. Key Gase Gasess Gene Generat rated ed by Part Particul icular ar Fau Fault lt Key Gas Characteristic Fault H2 Partiall Discharge Partia C2H6 Thermal Ther mal Fau Faultlt <30 <300 0 ºC C2H4 Thermal Ther mal fault 300 300ºCºC-<70 <700 0 ºC C2H2, C2H4 Therm Th ermal al Fa Faul ultt > 70 700 0 ºC C2H2, H2 Dischar Dis charge ge of Ener Energy gy Tablee IV. Insu Tabl Insulati lating ng Oil Test Testss Type of Test ASTM Method Significance/Effects Dielectric Br Breakdown D877, D1 D1816 Moisture, pa particles, ce cellulose fifibers/lower di dielectric strength Neutralization Nu Number D644, D9 D974 Acidic pr products fr from oi oil ox oxidation/ sl sludge, co corrosion Interfacial Te Tension (I(IFT) D971 Presence of of po polar co contaminants, ac acids, so solvents, va varnish Color D1500 Darkening indicates contamination or deterioration Water Co Content D1533 Excessive pa paper de decomposition/lower di dielectric st strength Power Fa Factor D924 (1 (100, 25 25 C) C) Dissolved me metals, pe peroxides, ac acids, sa salts/overheating Oxidation In Inhibitor (D (DBPC*) D2668, D1 D1473 Low le levels re results in in ac accelerated oi oil ag aging Metals in Oil Indicative of pump wear, arcing or sparking with metal *DBPC—Dibutyl *DBPC— DibutylParacreso Paracresoll November/December 2002 — Vol. 18, No. 6 17 Table Tab le V. Gu Guide idelin lines es for Deg Degrad radatio ation n 2-Furaldehyde (ppm) Degree of Polymerization Extent of Degradation 0 – 0.1 800 – 1200 Insignificant 0.1 – 0.5 700 – 550 Significant 1.0 – 2.0 500 – 450 Cause for concern >10 <300 End of life WINDING RATIO RATIO There are two commo commonly nly used PD detection methods: turns test isduseful to op determine whThe wheth ether erwinding or no nott the there re ar areeratio anyy sh an shor orte ted turns tur ns or open en wi wind nd-ing cir circui cuits. ts. The mea measur sured ed rat ratio io shou should ld be wit within hin 0.5 per per-cent of the ratio of the rated voltages between the windin win dings, gs, as not noted ed on the tra transf nsform ormer er nam namepl eplate ate.. All tap posi po siti tion onss an and d al alll ph phas ases es sh shoul ould d be me meas asur ured.The ed.The te test st ca can n be performed at a very low voltage. detection of the acoustic signals and measurement of the elec el ectr tric ical al si sign gnal alss pr prod oduc uced ed by the PD [2 [27] 7].. PD ca can n al also so be detected indirectly, using chemical techniques such as measur mea suring ing the deg degrad radati ation on pro produc ducts ts pro produc duced ed by the PD. The acceptable PD limits for new transformers are dependent on the voltage and size of the t he transformers and range from < 100 to < 500 pC. PD pul pulses ses gen genera erate te mec mechan hanica icall str stress ess wav waves es tha thatt pro proppagate aga te thr throug ough h the surrou surroundi nding ng oil (in the the range range of 100 to 300 kHz) [35]. To detect these waves, acoustic emission sensors are mounted either on the transformer tank wall or inthe oi oill in insi side de th thee tr tran ansf sfor orme merr ta tank nk inthe oi oil.If l.If mu mult ltiiple sensors are used, the PD can be located based on the arri ar riva vall ti time me of th thee pu puls lses es at the se senso nsors rs.. The se sens nsit itiv ivit ityy of THERMOGRAPHY Infrared emission testing is used to check the external surface temperature of the transformer on-line. It is use us eful for det detecti ecting ng the therma rmall pro proble blems ms in a tra transf nsform ormer er,, suc such h as cooling system blockages, locating electrical connection tio n pro proble blems, ms, andfor loc locati ating ng hot spo spots ts [32 [32], ], [39 [39], ], [42 [42]. ]. Infrar Inf rared ed ima imager gerss “se “see” e” the sur surfac facee hea heatt rad radiat iation ion fro from m objects. It cannot look “inside” the transformer tank. Black and white thermograms (heat pictures) show hot area ar eass in wh whit itee an and d co cold ld ar area eass in bl blac ack, k, un unle less ss st stat ated ed oth other er-wise. For color thermograms, white and red areas are usually hotter, while black and blue areas are colder. colder. Infrared thermography provides the heating patterns forr theloa fo theload d th that at wa wass on theequi theequipm pmen entt at th thee ti time me tha thatt th thee scan sc an wa wass pe perf rfor orme med. d. An Anyy ab abno norm rmal al co cond ndit itio ions ns ca can n be lo lo-cate ca ted d fr from om th thee sc scan an.. Th Thee sev sever erit ityy of ov over erhe heat atin ing g fr from om th thee scan can be categorized as follows: Clas Cl assi sifi fica cati tion on Tem empe pera ratu ture re Ex Exce cess ss* * Attention: 0 - 9°C Intermediate: 10 - 20 °C Serious: Critical: 21 - 49 °C >50 °C *Temperature *Tempera ture excess is defin defined ed as the diff difference erence in temperature between a reference point on the transformer at normal temperature and a higher temperature point. Nontraditional Transformer Transformer Monitoring Techniques Techniques Theree ha Ther hass be been en a gr grea eatt de deal al of new de deve velo lopm pmen entt in te test st-ing in g an and d mon monito itorin ring g tec techni hnique quess in re recen centt yea years,and rs,and the these se ar aree finding increasing increasing use on transformers. IN-SERVICE IN-SERVI CE PD TESTING PD in transformers degrades the properties of the insulating materials and can lead to eventual failures [23]. 18 the test is dependent on the location of the PD, since the sign si gnal al is at atten tenua uated ted by th thee oi oill an and d wi wind ndin ing g st stru ruct ctur ure. e. Th This is means that the deeper inside the winding the PD is located, the greater the attenuation. Piezoelectric sensors and an d fi fiber beropt optic ic sen sensor sorss ca can n mea measur suree the thePD.Rece PD.Recent nt res resea earch rch shows sho ws tha thatt opt optic ical al sen sensor sorss ha have ve a pot potent entia iall sen sensit sitivi ivity ty muc much h higher than normal external tank-mounted piezoelectric sensor sen sorss fo forr PD det detect ectio ion n [93 [93]. ]. Fi Fiberoptic beropticsen sensor sorss al also so cou could ld potentially potential ly be placed inside the winding winding.. PD causes highhigh-freque frequency ncy lowlow-ampli amplitude tude distur disturbance bancess on th thee ap appl plie ied d vo volt ltag agee an and d cu curr rren entt wa wavef vefor orms ms th that at ca can n be detected electrically. Electrical PD signals can be measured at a number of different locations, including bushing tap current or voltage and neutral current [17]-[18], [23]-[24], [26]-[27 [23]-[24], [26]-[27], ], [35]-[3 [35]-[36], 6], [39], [45], [50]-[5 [50]-[51], 1], [55], [62], [68], [70], [73], [76]. Techniques Techniques using detection of ultra-high-frequency signals (typically 1–2 GHz) havee beendevelo hav beendeveloped ped to dete detect ct PD in gas gas-in -insul sulate ated d subs substatations. The method has been applied to transformers and shows some promise [44], [61]. Acoustic methods of PD detection are limited by signal attenuation, and electrical measurements are limited by electromagnet electro magnetic ic interfe interference rence problem problems. s. Equipme Equipment nt is commercially commerc ially available available to continu continuously ously monitor and evalua eva luate te int intern ernal al PD onon-lin linee usi using ng bot both h aco acousti usticc and ele elecctrical methods. Investigations are also proceeding on improving acousti aco usticc det detect ection ion of PD, as wel welll as fur furthe therr wor work k on on ele elecctrical dete trical detecti ction on for in ser servic vicee mon monito itorin ring g [94 [94]. ]. The goa goall is to be able to detect and ideal ideally ly locate PD levels with a minimum sensitivity of at least 100 pC. IEEE Electrical Insulation Magazine RECOVERY VOLTAGE MEASUREMENT The recovery voltage measurement (RVM) [95]-[98] method meth od is us used ed to de dete tect ct th thee co cond ndit itio ions ns of oi oill-pa pape perr in insu su-lation and the water content of the insulation. The RVM relies on the principle of the interfacial polarization of composite dielectric materials; that is, the buildup of space spa ce cha charge rgess at the int interf erface acess of oil oil-pa -paper per ins insula ulatio tion n due to impurities and moisture. A dc voltage is applied to the insulation for a time. The electrodes are then short-circuit cu ited ed fo forr a sho short rt pe peri riod od of ti time me,, af afte terr wh whic ich h th thee sh shor ortt ci cirrcuit is re cuit remo move ved d to ex exam amin inee th thee ra rate te of th thee vo volt ltag agee bu buil ildup dup or the polarization profile. The time constant associated with wit h thi thiss peak recover recoveryy vol voltag tagee giv gives es an indicat indication ion of the state sta te of the ins insula ulatio tion. n. The mai main n par parame ameter terss der derive ived d fro from m the polarization spectrum are the maximum value of the recovery voltage, the time to peak value, and the initial rate of rise of the recovery voltage. The test results give an indication of the state of the oil/paper insulation structure of the transformer. It requires a transformer outage to carry out the test [18], [23], [23 ], [40 [40], ], [45 [45], ], [60 [60], ], [66 [66]-[ ]-[67] 67],, [99 [99]. ]. Thi Thiss met method hod is ver veryy controversial as to its suitability for direct measurement of the moisture content in oil, due to the strong dependence of the results on the geometry, geometry, and construction of the insulation system of a transformer. Figures 5 and 6 show sh ow typ typic ical al RVM cu curv rves es fo forr ol old d tr tran ansf sfor orme mers rs th that at ar aree in good and poor condition. The drawbacks of this test are that a long outage may be requi required red and the the unrel unreliabi iability lity in the inte interpre rpretatio tation n of the results. rentscan be mo rentscan moni nitor tored ed to obt obtai ain n a si sign gnat atur uree ev ever eryy ti time me th thee tap changer moves. Changes in this signature are used to detect problems in the tap changer. Bearing monitors are used use d to todet detectbeari ectbearing ng wea wearr on ontra transf nsform ormer eroilpumps oilpumps[26 [26]. ]. INTERNAL TEMPERATURE MEASUREMENT The tra tradit dition ional al met method hod to mea measur suree the tem temper peratu ature re of a trans transforme formerr windi winding ng is to measur measuree the transformer’s transformer’s top to p an and d bo bott ttom om oi oill tem tempe pera ratu ture re an and d es esti tima mate te th thee ho hott sp spot ot temperature. New fiber optic equipment has been developed that is able to monitor the temperature two different ways. One is a distributed temperature measurement along the entir entiree lengt length h of the windi winding ng by a fiber optic cable. The temperature of the complete winding could be monitored if a fiber optic cable can be laid along the transformer winding during construction of the transformer. There are drawbacks of this method, however. High cost and high mechanical stresses on the fiber (squeezing and buckling) are a major concern. The fiber optic needs to be handled with extreme care. It would have to be installed during transformer construction [23], [100]. The application of the fiber optic sensor so far has bee been n mai mainly nly for lab labora orator toryy res resear earch ch and pri princi ncipal pal design studies. The technology used in the fiber optic temperature sensors is capable of measuring the full range of temperatures encountered on transformers. 1000 ) 100 WINDING INSULAT INSULATING ING OIL TESTING/MON TESTING/MONITORING ITORING V ( e In add additi ition on to the win windin ding g ins insula ulatin ting g oil test testss rou routin tinely ely carried out, as already described, there are other oil tests that can provide information on the condition of the transformer.. These include particle count, metals in the transformer oil, furan furananal analysis, ysis, anil aniline ine point,corrosive sulfur sulfur,, and oxi oxi-dation stability. Equipment to continuously monitor oil condition in service is increasingly beingsystems installed on transformers. The most widely installed measure hydrogen content, although systems that measure moisture and other gases are also available. available. The hydrogen and composition sensors use semiconductor or fuel cell technology; and more complex sensors, which make use of infrared technology and gas chromatography, can detect several or all of these gases. g tal o V 10 1 0.01 100 1000 10,000 1000 100 V November/December 2002 — Vol. 18, No. 6 10 Figu re 5. Typica Typicall RVM cur ve for a trans forme r i n g ood condit condition. ion. ( TAP CHANGER/MOTOR MONITORING detect pro detect proble blems, ms, suc such h as con contac tactt ove overh rheat eating ing,, whi while le ac acous oustic tic analys ana lysis is of the swi switch tching ing ope operat ration ioncan can det detect ect fau faults lts in the selector lect or and diverter diverter switches [99]. Tap Tap changer motor cur- 1 Charge Time (Seconds) ) The use of oil testing has been extended to the testing of the tap changer oil. The oil tests are used as an indicator of contact deterioration [19], [27], [35], [39], [42], [50]-[51], [55], [59], [62], [63]-[64], [100]. Monito Mon itori ring ng of the thetap tapcha change ngerr tem temper peratu ature re can canbe be use used d to 0. 1 e g tal o V 10 1 0.01 0.1 1 10 100 1000 Charge Time (Seconds) Figu re 6. Typica Typicall RVM cur ve for a trans forme r i n poor p oor condit condition. ion. 19 The other type of system uses fiber optics for point temperature measurement. Since the sensors and associated cables are insulated, they can be installed directly at the tra transf nsform ormer er hot spo spots. ts. The bes bestt tim timee to in insta stall ll the these se sen sen-sorsis sors is duri during ng tran transfor sformer mer cons construc truction tionat at the loc locatio ations ns indi indi-cated by thermal modeling of the transformer; however, theyy ca the can n be ret retrof rofit itted ted to an exi existi sting ng tra transf nsfor ormer mer,, but thi thiss is difficult difficu lt to do. Temperature systems are being installed in on-load tap changers. Monitoring the temperature and temperature trends has been found to be a useful indicator of degradadegradationof tion of tap chan changer gercont contacts acts[23] [23],, [27] [27],, [35] [35],, [49] [49]-[51 -[51], ], [53] [53].. ON-LINE POWER FACTOR MEASUREMENT Systems to measure bushing power factor on-line are now available. Manufacturers have made available two systems for monitoring the condition of bushings, based on detecting changes in their capacitance and power factor. Both systems use sensors on the bushing capacitance tapss to mea tap measur suree the bush bushingleaka ingleakage ge cur curren rents. ts. One syst system em uses an electric field sensor to measure the bus voltage phase angle, and calculates the capacitance and dissipation factor from the measured data. The other technique sumss the bus sum bushin hing g cur curren rents ts fro from m the thr three ee pha phases ses andplots them on a polar plot. Any shift in the resultant currents indicates a change in capacitance or dissipation factor of one of the bushings. These measurements can give sufficient warning of an impending bushing failure to allow replacement of the bushing before a catastrophic failure occurs. POWER FACTOR VS. FREQUENCY MEASUREMENT (DIELECTRIC SPECTROSCOPY) The measurement of power factor over a broad range of frequencies from a low of 1 mHz to 1 kHz or higher has been used to evaluate the insulation condition [17], [40], [48], [57]. Interference can be easily detected as an irregulari irreg ularity; ty; the transf transformer ormer insulation insulation usuall usuallyy has a smooth power factor factor-frequ -frequency ency chara characterist cteristic. ic. Po Power wer −30 −40 )f ) ( V/ f)( −50 I ) ( B −60 d( e d tui −70 n g Reference Transformer Transformer Under Test a M −80 −90 0.5 1 1.5 Frequency (Hz) Figu re 7. FRA tes t resu lts comp ari son . 20 2 2..5 2 3 x 106 factor-frequency characteristics allow for a more complete diagnosis of the examined insulation. At the lower frequency range, pressboard dielectric loss is the main factor fac tor;; at med medium ium fre frequen quency cy ran range,the ge,the oilconduc oilconductiv tivity ity is the dominant contributor; and at the higher frequency range, the pressboard and the oil volume determine the dielectric diele ctric loss. Different aging mechanisms mechanisms can be detected tec ted andidenti andidentifie fied d at the their ir res respec pectiv tivee fre frequen quency cy ran ranges ges.. WINDING MOVEMENT MOVEMENT DETECTION DETECTION A very serious problem problem that is particularly particularly difficult difficult to detect is movement or distortion of the transformer winding. Forces Forces on the winding during short circuits on the transformer can cause winding distortion. The other source sou rce of win windin ding g mov moveme ement nt is red reduct uction ion or los losss of win windding clamping. This can result in a transformer fault that will cause damage to the transformer and may result in explosive failure of the transformer. Traditionally, the only way to evaluate the winding condition of a large power transformer is to drain the oil from the transformer and carry out an internal inspection. Some research work has focused on using the transformer vibration signal to detect winding looseness and on deve develop lopingthe ingthe ana analys lysis is tec techni hnique quess for int interp erpret retingthe ingthe vibrat vib ration ion dat dataa [10 [101]1]-[10 [104]. 4]. The meth method od is bas based ed on loo lookkingfor cha change ngess in the tra transf nsform ormer’s er’s vib vibrat rationsigna ionsignatur turee to dete de tect ct mo movem vemen entt in thewind thewindin ing. g. Th This is me meth thod od is no nott us used ed as widely as frequency response analysis tests for detectdet ecting winding movement. In the frequency response analysis test (FRA), the transformer is isolated from the system and the impedance or admittance of the transformer is measured as a function of frequency (typically to at least 2 MHz). This gives a “fingerprint” of the transformer. The test is repeated over time and the “fingerprints” from two or more tests are compared. There are two different test methods commonly used to ca carr rryy ou outt th thee FR FRA A tes test: t: th thee sw swep eptt fr frequ equen ency cy te test st an and d th thee pul pulse se tes test. t. The swept swe pt fre frequen cy meth method od app applie liess a var variab le frequency voltage or quency a white noise voltage toiable the high-voltage winding and records the response in another oth er win windin ding g or ter termin minal. al. Thi Thiss tec techni hnique que is mor moree wid widely ely used in Europe than in North America. A similar techniquemorecomm ni quemorecommon only ly use used d in No Nort rth h Am Amer eric icaa is th thee pu puls lsee FRA test. With this technique a pulse signal s ignal is applied to the hig high-v h-volt oltage age win windin ding, g, and the res respon ponse se is rec record orded ed in another winding or terminal. Research indicates that the pulse method is more sensitive to detect small winding moveme mov ement nt and win windin ding g cla clampi mping ng loo loosen seness ess [10 [105]. 5]. Fig Figure ure 7 sh show owss an FR FRA A te test st re resu sult ltss co comp mpar aris ison on fo forr a tr tran ansf sfor orme merr with wi th so some me mo movem vemen entt co comp mpar ared ed to a tr tran ansf sfor orme merr in go good od condition. In general, the greater the difference between the two “signatures,” the greater movement in the transformer. The test requires experienced personnel to compare the two signatures and evaluate the severity of the movement. IEEE Electrical Insulation Magazine conven ventio tional nal FR FRA A tes testt req requir uires es a tra transf nsform ormer er out out-The con age to ca age carr rryy ou outt th thee tes test. t. Wor ork k ha hass be been en ca carr rrie ied d ou outt in Eu Eu-rope and North America to use the transient voltages generated gener ated durin during g switching switching opera operations tions as the driving driving signal to mea measur suree the tra transf nsform ormer er adm admitt ittanc ancee [75 [75], ], [10 [106]. 6]. If the on-line FRA test could be developed, it could reduce or el elim imin inat atee theneedfor ou outa tage gess to ca carr rryy outan FR FRA A tes test. t. The FR FRA A tes testt has bee been n use used d ext extensi ensivel velyy. The dra drawba wbacks cks of the test are that it requires an outage, an initial reference testt wit tes with h the tra transf nsform ormer er in goo good d con condit dition ion,, gre great at con consis sis-tenc te ncyy in th thee te test st se setu tup p fr from om on onee te test st to th thee ne next xt,, an and d it re re-quires experienced personnel to interpret the data. Despite Despi te these drawbacks this has been found to be the most effective test in detecting winding movement. Another technique used to detect det ect winding w inding displacedis placement is the frequency response of stray losses (FRSL). This Th is te test st is do done ne ov over er a ra rang ngee of fr freq eque uenc ncie iess fr from om 20 Hz to over 600 Hz [17]-[19], [21], [23], [25], [27], [34], [38], [45], [49], [52], [54], [56], [58], [72], [76], [107]. The FRSL test has not been extensively used or studied. It is th thou ough ghtt no nott to be as se sens nsit itiv ivee to wi wind ndin ing g mo move veme ment nt as the FRA test due to its lower measurement frequency frequ ency range. testing at reduced voltages. The use of other tests (both offof f-li line ne an and d on on-l -lin ine) e) is in incr crea easi sing ng bu butt is li limi mite ted d by a nu nummber of factors. Cost: Co st:Th Thee hi high gh co cost st of te testi sting ng an and d mo moni nito tori ring ng ca can n ma make ke it diff difficu icult lt to to justif justifyy the test tests. s. The purc purchas hasee pri price ce of the the equi eq uipm pmen entt is on only ly on onee co cost st fa fact ctor or li limi miti ting ng the their ir use use.. Th Thee costt of iso cos isolat lating ingthe the tra transf nsform ormer er and andper perfor formin ming g the thetest test canbe can be subs substan tantia tiall for foroff off-li -line ne test tests. s. The Thelon long g out outage agetim timee requir req uired ed by test tests, s, such as the rec recove overy ry vol voltag tagee meth method, od, can make them difficult to carry out. The installation cost co stss fo forr onon-li line ne mon monit itor orin ing g equ equip ipme ment nt ca can n be a ma major jor cost factor. Dataa int Dat interp erpret retati ation: on: The Theint interp erpret retati ation on of test testss oft often en requires experienced expert personnel. Incorrect interpret pr etat atio ion n of th thee da data ta ca can n le lead ad to fa fals lsee co conc nclu lusi sion onss ab abou outt the transformer condition. Reliability: The degradation of a transformer occurs overr seve ove several ralyea years. rs. Sens Sensors orsand andele electr ctroni onicc equi equipme pment nt instal st alle led d on onthetran thetransf sfor ormer merss mus mustt be beab able leto tope perf rfor orm m ove overr many years with minimal maintenance. Compatibility: The compatibility of the many on-line monitoring systems now available is a major concern. Typically systems from one supplier are completely incompatible with those of other suppliers. Diagnostic Software and Expert Systems Use of non nontra tradit dition ional al dia diagno gnosti sticc and mon monito itorin ring g tec techhniques is expected to increase on the aging transformer population. The cost of the equipment will fall and reliabilit abi lityy wil willl inc increa rease se wit with h inc increa reased sed usa usage. ge. The int interp erpret retaation and understanding of the test data obtained from tests such as FRA, RVM, and vibration testing will improve. In particular, standard analysis techniques are being developed that will enable field personnel to more easily use the test results and will reduce the need for interpretation by experts. Multiple test software that combines the results of different tests and gives an overall assessment of condition is expected to find increasing use. The use of continuous on-line monitoring of transformers is increasing. The cost of the equipment is decreasing and the sensors are improving. This makes it easierr to justify the installation easie installation of sophisticated sophisticated monitormonitoring sys systems tems on tra transf nsform ormers ers.. Sta Standa ndardi rdizat zation ion wil willl mak makee it easier eas ier to int integr egrate ate sys system temss and dat dataa fro from m dif differ ferent ent sup suppli pli-ers. The use of wireless technologies within the substation for communication between the transformer and control room will make it easier to install monitoring equipment. The ultimate goal of transformer monitoring monitoring and diagnosti nos ticc tec techni hnique quess is to hav havee a set setof of dev devic ices/ es/sys system temss to mon moniitor and ant antici icipat patee the tra transf nsform ormer er fai failur lure, e, so tha thatt appropriate action can be taken before forced outage occurs. The organizational organizational culture of a power utility signifi signifi-cantly impacts on the operational practices in the use of condition-based maintenance. Diagnosti Diagno sticc sof softwa tware, re, whi which ch giv gives es mor moree def defini inite te ind indica ica-tions of transformer problems than conventional analysis, is under investigation by many researchers and util ut ilit itie iess [6 [63] 3],, [7 [71] 1],, [7 [76] 6].. Th Thee us usee of so soft ftwa ware re ca can n im impr prov ovee the rel reliab iabili ility ty and rep repeat eatabi abilit lityy of the ana analys lysis is of test dat data. a. It ca can n al also so be us used ed to ext extra ract ct in info form rmat atio ion n tha thatt is no nott av avai aillable from the data directly. A great deal of research has been done on software to interpret transformer oil test data such as gas, moisture content, and dielectric strength and correlating the data with the transformer insulation condition. Expert systems te ms ha have ve be been en dev devel elop oped ed th that at gi give ve an al alar arm m si sign gnalto alto sy sysstem operators. Some systems have been developed to detect PD signals in transformers [94]. Equipment using us ing acoustic emission sensors and specialized software has been be en su succ cces essf sful ul in de detec tecti ting ng PD an and d lo loca cati ting ng th thee or orig igin in of the dis discha charge rge.. The sen sensor sorss are mou mounted nted exte externa rnally lly on the transformer tank wall and three-dimensional location techni tec hniquesare quesare app applie lied d to loc locate ate the sour source ce of the det detect ected ed signals. The present advancement in artificial intelligence (AI) mode mo deli ling ng tec techn hniq iques ues ha hass en enab able led d po power wer en engi gine neer erss an and d re re-sear se arch chers ers to de devel velop op po powe werf rful ul an and d ver versa sati tile le AI sof softw twar aree to diag di agno nose se tr tran ansfo sform rmer er fa fault ults. s. Th Thee use of exp expert ert sys system temss of of-ferss the pote fer potenti ntial al of reducing reducing the man manpow power er and financia financiall overhead over head requ require ired d by util utiliti ities es to asse assess ss tra transfo nsforme rmerr con condidition; however, this potential has not yet been realized. Discussion and Concluding Remarks The most widely used tests to diagnose the condition of tra transf nsform ormers ers are sti still ll oil test testss and off off-li -line ne pow power er fac factor tor November/December 2002 — Vol. 18, No. 6 received ived the B.Sc B.Sc.. deg degree ree in elec electric trical al engi engineer neering ing M. Wang rece from fr om Xi Xian an Ji Jiaot aotong ong Uni Univer versit sityy, Xi Xian an,, Chi China na and the 21 M.A.Sc. degree in electrical engineering from the University of British Columbia, Vanc ancou ouver ver,, Ca Canad nada, a, in 198 19822 an andd 199 1991, 1, re re-spectively. From 1982 to 1988, she was withthe with the Wuha uhann Hig Highh Volta oltage ge Re Resear search ch Institute as a research engineer. engineer. In 1991, she joined Powertech Labs Inc. as a senior research engineer. Her research interests are in tran transfor sformer mer cond conditio itionn moni monitori toring, ng, tran transfor sformer mer frequency response analysis, and high-voltage engineering. She is an active IEEE member and is a registered professional engineer in the province of British Columbia. John Vande Vanderma rmaar ar re rece ceiv ived ed hi hiss B. B.Sc Sc.. in En En-- gineeri ginee ring ng fr from om the Uni Univer versi sity ty of Ma Manit nitob obaa in 1975. From 1975 to 1980 he was with the Operations and the Engineeri Engineering ng Divisions of BC Hydro. In 1980 he joined the Resear Re search ch and andDeve Developm lopment ent Divi Division sionof of BC Hydro (now Powertech Labs Inc.). He has been responsible for many research projects jec ts in the thear areas easof of hig highh-vol volta tage ge ins insul ulati ation, on,eq equi uipme pment nt con con-ditio di tionn mon monito itori ring, ng, an andd equ equipm ipment ent lilife fe as asses sessme sment. nt. Current Cur rently ly,, he is the Mana Manager ger of the the Hig Highh Volta Voltage ge Gro Group up at References [1] CIGRÉ Working Working Group 05, “An international survey on failures failure s in large power transformers in service,” Electra, no. 88, May 1983. [2] V.I. Kogan, et al., “Failure analysis of EHV transformers,” IEEE Trans. Power Delivery, vol. 3, no. 2, pp. 672-683, 1988. [3] B. Sparling, “Transformer monitoring and diagnostics,” in Proc. IEEE Power Engineering Society 1999 Winter Power Meeting,vol.2, New York, 1999, pp. 978-980. [4]] O.N [4 .N.. Gr Grech echko ko an andd I. Ka Kala lach chev eva, a, “C “Cur urre rent nt tr tren ends ds in th thee development of in-service monitoring and diagnostic systems for 110-750 kV power transformers,” Applied Energy: Russian Journal of Fuel, Power, Power, and Heat Systems, vol. 34, no. 5, pp. 84-97, 1996. [5] A.J. McElroy, “On the significance of recent EHV transformer failure fai luress inv involv olving ing wind winding ing res resona onance nce,” ,” IEEE Tran rans. s. Po Power wer Apparatus and Systems, vol. 94, no. 4, 1975, pp. 1301-1306. [6] K.H K.H.. Lee and J.M. Sch Schneid neider er,, “Ro “Rockpo ckport rt tra transi nsient ent vol voltag tagee monitoring system: analysis and simulation of record waveforms,” IEEE Trans. Trans. Power Delivery, vol. 4, no. 3, 1989, pp. 1794-1805. [7] M. Minhas, J.P. Reynders, and P.J. de Klerk, “Failure in power system transformers transformers and appro appropriate priate monitoring techni techniques,” ques,” th presen pre sented ted at the 11 Int. Symp. High Voltage oltageEnginee Engineering, ring, Londo London, n, U.K., 1999. [8] R. Sahu, “Optimization of the transmission and subtransmission nd Powe owerte chghLa Labs Inc. In c. Mr Mr. .tVTan ander derma maar ars an is the thesec retar tary of IEC TC 42rtech High Hi Vbs olta ol tage ge Tes est echn ec hniq ique ues and d a secre memb me mber ery of IEC IE C Working Group 12 on Voltage Measurement by Means of Air Gap Gaps. s. He is also acti active ve on var various ious IEEE stan standar dards ds committees. K.D. Sriv Srivasta astava va (M’67-SM-81-F’85-LF’00) received the B.E. (Honors) degree in electrical engineering from the University of Roor oorkee kee,, Ro Roork orkee, ee,Ind India ia and andthe thePh. Ph.D. D. degree gr ee fr from om the Uni Univer versit sityy of Gl Glasg asgow ow,, Gl Glas as-gow, U.K. in 1952 and 1956, respectively. He wa wass a Res esea earc rchh En Engi gine neer er at A. Rey eyro roll llee and Co., Newcastle, U.K. from 1957 to 1958. From 1958 to 1960, he was with the University of Roorkee Roor kee and the University of Jodhpur as a Senior Faculty Member Mem ber.. Fr From om 1961 to 1966, he wor worked ked at Brush Electric Electric Co., Loughborough, U.K., as a Senior Research Engineer, and then at the Ru Ruther therfor fordd Hig Highh Ener Energy gy Labo Laborato ratory ry,, Harwell, U.K. as a Principal Scientific Officer. In 1966, he emigrated emigrat ed to Canada and was appointed Professor of electrical engineering at the University of Waterloo, Waterloo, ON,, Ca ON Cana nada da,, an andd fr from om 19 1972 72 to 19 1978 78,, he wa wass Ch Chai airm rman an of the Department. In 1983, he was appointed Professor and Head of the Department of Electrical Engineering at the Universi Uni versity ty of Bri British tishCol Columbi umbiaa (UBC (UBC), ),V Vanco ancouver uver,, BC, BC,Can Can-ada,, and from ada from 1986 1986 to 1994, 1994, he was Vice Vice Pre Preside sident nt of StuStudent and Academic Services at the same University. His researc rese archh inte interest restss are arein in gas gaseous eousinsu insulati lation on and andhig high-vo h-voltag ltagee engineering. 22 transformer spares on the AEP system,” in Proc. 42 American Power Conf. Conf., Chicago, IL, 1980, pp. 701-705. [9] CIGRÉ Working Group 09 of Study Committee 12, “Lifetime evaluation of transformers,” Electra, no. 150, pp. 39-51, 1993. [10] L. Pettersson, “Estimation of the remaining service life of power transformers and their insulation,” Electra, no. 133, pp. 65-71, 1990. [11] M. Darveniza, et al., “Investigations into effective methods for assessing the condition of insulation in aged power transformers,” IEEE Trans. Trans. Power Delivery, vol. 13, no. 4, pp. 1214-1223, 1998. [12]] J.A. Lap [12 Lapwor worth, th, P.N. Jar Jarman man,, and I.R. Fu Funne nnell, ll, “Co “Condi nditio tionn assessment assess ment techniques for large power transf transformers ormers,” ,” Second International Conf. Reliability of Transmission and Distribution Equipment, Cov Coventr entryy, U.K U.K.,., Con Conf.Publ.no. f.Publ.no. 406 406,, pp.85-90 pp.85-90,, 199 1995. 5. [13] M. Kazmierski, R. Sobocki, and W. Olech, “Selected elements of lifee ma lif mana nagem gemen entt of la larg rgee po power wer tr tran ansf sfor orme mers rs—A —A Pol olish ish experience,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [14]L. Pet Pettersso tersson, n, N.L. Fa Fantana ntana,, and U. Sunder Sundermann,“Life mann,“Life asses assessment: sment: Rank Ra nkin ingg of po powe werr tr tran ansf sfor orme mers rs us usin ingg co cond nditi ition on ba based sed evaluation—A evalua tion—A new appro approach,” ach,” presented at the Intern Internationa ationall Council on Large Electric Elect ric Systems (CIGRÉ), Paris, France, 1998. [15] P.N. Jarman, J.A. Lapworth, and A. Wolson, “Life assessment of 275 and 400 kV transmission transformers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [16] G. Breen, “Essential requirements to maintain transformers in service,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1992. IEEE Electrical Insulation Magazine [17] S. Agou, et al., “Increasing the reliability of online insulation condition assessment of MV-HV equipment,” Electricity Today, vol. 12, no. 6, pp. 22-27, 2000. [18] T.H. Aschwanden, et al., “Development and application of new condition assessment methods for power transformers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [19]] A. Bas [19 Basak, ak, “Co “Condi ndition tion mon monito itorin ringg of pow power er tra transf nsform ormers ers,” ,” Engineering Science and Education Journal, vo vol.8, l.8, no no.. 1,pp.411,pp.41-46 46,, 1999. [20]] M. Bela [20 Belange nger, r, “T “Tran ransfo sforme rmerr dia diagno gnosis sis:: Pa Part rt 1: A sta statis tistica ticall justification for preventative maintenance,” Electricity Today, vol. 11, no. 6, pp. 5-8, 1999. [21] M. Belanger, “Transformer diagnosis: Part 2: A look at the reference data for interpreting test results,” Electricity Today, vol. 11, no. 7, pp. 19-26, 1999. [22]M. Belange Belanger, r,“T “Transf ransformerdiagnosis:Part ormerdiagnosis:Part 3: Detecti Detection on techniq techniques ues Electricity Today and frequency of transformer testing,” , vol. 11, no. 8, pp. 19-26, 1999. [23] C. Bengtsson, “Status and trends in transformer monitoring,” IEEE Trans. Trans. Power Delivery, vol. 11, no. 3, pp. 1379-1384, 1996. [24] T.R. T.R. Blackburn, et al., “On-line partial discharge measurement on instrum ins trument ent tra transf nsform ormers ers,” ,” pre presen sented ted at the Int Intern ernati ationa onall Symposium on Electrical Insulating Materials, Toyohashi, Japan, 1998. [25] L. Bolduc, et al., “Detection “Detect ion of transformer winding displacement by the frequency response of stray losses (FRSL),” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [26] P. Boss, et al., “Economical aspects and practical experiences of power pow er tra transf nsform ormers ers onon-line line mon monitor itoring ing,” ,” pre presen sented ted at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [27] A.J.M. Cardoso and L.M.R. Oliveira, “Condition monitoring and Int. t. J. CO COMA MADE DEM M, diagno dia gnosti stics cs of pow power er tra transf nsform ormers ers,” ,” In Sunderland, U.K., 1999, pp. 5-11. [28] D. Carreau, et al., “Condition monitoring diagnostics expert systems: A project roadmap,” Electricity Today, vol. 12, no. 4, pp. [35] K. Feser, et al., “The transfer function method for detection of windingdisplacements windingdisplaceme nts on powertransforme powertransformers rs aftertransport,short circuit or 30 years service,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [36] [3 6] E. Go Gock cken enba bach ch an andd H. Bo Bors rsi, i, “D “Dia iagn gnos osti ticc me meth thod odss fo forr transformer transf ormerss on-si on-site,” te,” in Proc Proc.. Int. Symp. Ele Elec. c. Ins Insul. ul. Mat Matls. ls., Toyohashi, Japan, 1998, pp. 2-36. [37] [3 7] P. Gu Guui uini nic, c, “P “Pro rogr gres esss re repo port rt of st stud udyy co comm mmit itte teee 12 (Transformers),” Electra, no. 178, pp. 17-21, 1998. [38] E. Hanique, H.F. Reijnders, and P.T.M. Vaessen, “Frequency response analysis as a diagnostic tool,” J. Elektotechniek, vol. 68, no. 6, pp. 549-558, 1990. [39] J.W. Harley, “CIGRÉ Working Group 12.18 TF02, ‘Survey on diagnostics & monitoring techniques transformer subsystems,’” presented at CIGRÉ SC 12, Sydney Colloquium, Colloquiu m, Sydney, Sydney, Australia, 1997. [40] J.W. J.W. Harley and V.V V.V.. Sokolov S okolov,, “Diagnostic techniques for power transformers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [41]] X. Hua [41 Huang ng and Z. Yan, “Stu “Study dy on tren trendd ana analysi lysiss met method hod for onon-line line insulation diagnosing of capacitive-type equipment,” presented at the International Symposium on Electrical Insulating Materials, Toyohashi, Japan, 1998. [42] T.R. Hyde, “On-lin “On-linee condit condition ion monito monitoring ring technology and 8-14, 2000. [29] T.H. Crowley, Crowley, “Automated diagnosis of large power transformers using adaptive model-based monitoring,” B.S. and M.S. thesis, M.I.T.,., Cambridge, MA, 1990. M.I.T [30] J.B. DiGiorgio, et al., “Cut failures by analyzing transformer oil,” Electrical World, vol. 192, no. 7, pp. 52-54, 1979. [31] N. Dominelli, “Furanic and non-furanic analysis as a transformer diagno dia gnosti stic,” c,” pre presen sented ted at the EPR EPRII Sub Substa station tion Equ Equipm ipment ent Diagnostics Conference IV, IV, New Orleans, LA, 1996. [32]] G. Duk [32 Duke, e, “Pr “Predic edictiv tivee mai mainte ntenan nance ce a cas casee stud studyy in inf infrar rared ed thermography,” Electrical Maintenance, pp. 11-12, 1998. [33]] R. Farq [33 arquha uharso rsonn and K. Cai Caird, rd, “An ove overvi rview ew of sub substa statio tionn equipment monitoring and diagnostics: Part 1,” Electricity Today, vol. 12, no. 5, pp. 23-26, 2000. applications,” ERA Technology, Rep. No. 95-0546R, 1995. [43] [4 3] A. Ja Jaks ksts ts,, et al al., ., “A ma majo jorr br brea eakth kthro roug ughh in tr tran ansf sfor orme merr technology,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [44] M.D. Judd, et al., “Transformer monitoring using the UHF techniq tech nique,” ue,” pre presen sented ted at the 11th Intern Internationa ationall Sympos Symposium ium on High Voltage Engineering, London, U.K., 1999. [45] A.J. Kachle Kachler, r, “Diagnostic “Diagnostic and monitoring technology technology for large power pow er tra transf nsform ormers ers,” ,” pre presen sented ted at CIG CIGRÉ RÉ SC 12, Sydn Sydney ey Colloquium, Sydney, Australia, 1997. [46] S.R. Kannan, M. Tech, and N. Rao, “Generator loading limits for impulse testing low-inductance windings,” Proc. IEE, vol. 122, no. 5, pp. 535-538, 1975. [47]] J.L.Kirtle [47 J.L.Kirtleyy, et al. al.,, “Mo “Monito nitorin ringg the hea health lth of pow power er tra transf nsform ormers ers,” ,” IEEE Computer Applications in Power, vol. 9, no. 1, pp. 18-23, 1996. [48] M.F. Lachman, W. Walter, and P.A. von Guggenberg, “On-line diagnostics diagno stics of high-v high-voltage oltagebushin bushings gs and curren currentt trans transformer formerss using the sum current method,” IEEE Trans. Power Delivery, vol. 15, no. 1, pp. 155-162, 2000. [49] P. P. Leemans, et al., “Control, diagnostic and monitoring of power transformers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [50]T. Leibfr Leibfried, ied, “Onlinemonitors keep transf transformersin ormersin service service,” ,” IEEE Computer Applications in Power, vol. 11, no. 3, pp. 36-42, 1998. [51] [5 1] T. Le Leib ibfr frie ied, d, et al al., ., “O “Onn-li line ne mo moni nito tori ring ng of po powe werr transformers—Trends, new developments and first experiences,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [34] K. Feser, et al., “General trends in condition monitoring of electrical insulation,” in Proc. Stockholm Power Tech. Int. Symp. Electric Power Engineering, vol. 1, Stockholm, Sweden, 1995. [52] T. Leibfried and K. Feser, “Monitoring of power transformers using the transfer function method,” IEEE Trans. Power Delivery, Paper PE-053-PWRD-1-08-1998, 1998. November/December 2002 — Vol. 18, No. 6 23 T.. Leibfried, “Online monitoring of power transformers—System [53] T technol tech nologyand ogyand dat dataa eva evalua luation tion,” ,” pre presen sented ted at the11 th International Symposium Sympos ium on High VoltageEngineering oltageEngineering,, Londo London, n, Englan England, d, 1999. [54] P. P. Macor, et al., “The “ The short-circuit resistance of transformers: The feedba fee dback ck in Fr Franc ancee bas based ed on tes tests, ts, ser servic vicee and cal calcula culatio tionn approaches,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [55] R. Malewski, J. Douville, and G. Belanger, “Insulation diagnostic system for HV power transformers in service,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1986. [56] R. Malewski and M. Kazmierski, “Diagnostic techniques for power transformers,” presented at the International Council on Large Electric Systems Syste ms (CIGRÉ), Paris, France, 2000. [57] R. Malewski, et al., “Instruments for HV insulation testing in substations,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [58] M. Mizokami, M. Yabumoto, and Y. Y. Okazaki, “Vibration analysis of a 3-phase model transformer core,” Electrical Engineering in Japan, vol. 119, no. 1, 1997. [59]] A. Mol [59 Mollma lmann nn and B. Pa Pahla hlavan vanpou pour, r, “Ne “New w gui guideli delines nes for interpretation of dissolved gas analysis in oil-filled transformers,” Electra, no. 186, pp. 30-51, 1999. [60] T. Noonan, “Power transformer on-site condition assessment testing,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [61]] A.J. [61 A.J.M. M. Pe Pemen men,, et al. al.,, “On“On-line line par partial tial dis discha charge rge mon monito itorin ringg of HV th components,” presented at the 11 International Symposium on High Voltage Voltage Engineering, London, U.K., 1999. [62] T.D. Poyser, et al., “On-line monitoring of power transformers,” IEEE Trans. Trans. Power Apparatus and Systems, vol. 104, no. 1, 1985. [63 63]] J. J.H. H. Pro rova vanz nzan anaa, et al al., ., “T “Tra rans nsfo form rmer er co conndi diti tion on monitoring—Realizing an integrated adaptive analysis system,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1992. [64]] L. Ru [64 Ruijin ijin,, et al. al.,, “On “On-lin -linee det detecti ection on of gas gases es dis dissol solved ved in transf tra nsform ormer er oil and the fau faults lts dia diagno gnosis sis,” ,” pre presen sented ted at the Internationa Intern ationall Sympos Symposium ium on Electri Electrical cal Insula Insulating ting Materia Materials,” ls,” Toyohashi, Japan, 1998. [65]] B.G [65 B.G.. Ru Rushf shford ord,, “Bus “Busines inesss cas casee dev develo elopme pment nt for onon-line line equipment,” Electrical Maintenance, pp. 5-10, Oct. 1998. [66]] T.K. Sah [66 Saha, a, et al., “Electrica “Electricall and chemical chemical dia diagno gnosti stics cs of transformers transf ormers insula insulation—P tion—Part art A: Aged transf transformers ormers sampl samples,” es,” IEEE Trans. Trans. Power Delivery, vol. 12, no. 4, pp. 1547-1554, 1997. [67]] T.K. Sah [67 Saha, a, et al., “Electrica “Electricall and chemical chemical dia diagno gnosti stics cs of transformers transf ormers insulation—Part insulation—Part B: Accele Accelerated rated aged insulat insulation ion samples,” IEEE Trans. Pow Power er Deliv Delivery ery, vo vol. l. 12, no. 4, pp. 1555-1561, 1997. [68] R.J. Schwabe, et al., “On-line diagnostics of oil paper insulated instrument transformers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. [69] S. Tenbohlen, et al., “Enhanced diagnosis of power transformers [70] Y. Tian, et al., “PD pattern identification using acoustic emission measur mea sureme ement nt and neu neural ral net networ works,” ks,” pre presen sented ted at the 11th International Symposium on High Voltage Engineering, London, U.K., 1999. [71] K. Tomsov omsovic, ic, M. Tapper apper,, and T. Ingvas Ingvasrsson, rsson, “Performance “Performance evaluation of a transformer condition monitoring expert system,” presented at the International Council on Large Electric Systems (CIGRÉ), Berlin, Federal Republic of Germany, Germany, 1993. [72]] A.J [72 A.J.. Vand anderm ermaar aar and M. Wang ang,, “T “Tran ransfo sforme rmerr con condit dition ion monitoring monito ring by frequen frequency cy respo response nse analy analysis” sis” in Proc. 10th Int. Symp. High Voltage Eng., Montreal, Canada, 1997, vol. 4, pp. 119-122. [73] C. Wang, Wang, et al., al. , “Analysis and suppression of continuous periodic interference interfe rence for on-lin on-linee PD monito monitoring ring of power transformers,” transformers,” th presented at the 11 International Symposium on High Voltage Engineering, London, U.K., 1999. [74] M. Wang, A.J. Vandermaar, and K.D. Srivastava, “Condition monitoring of transformers in service by the low voltage impulse test method,” presented at the 11th International Symposium on High Voltage Engineering, London, U.K., 1999. [75] M. Wang, A.J. Vandermaar, and K.D. Srivastava, “Transformer condition monitoring by the low voltage impulse test method,” presented at the CIGRÉ Third South African Regional Conference, Johannesburg, South Africa, 1998. [76] Z. Wang, Y. Liu, and P.J. Griffin, “Neural net and expert system using on- and off-line methods: Results, examples and future trends,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 2000. 1978. [86] M. Duval, “Dissolved gas analysis: It can save your transformer,” IEEE Electrical Insulation Mag., vol. 5, no. 6., pp. 22-27, 1989. 24 Applications in diagnose diagno se transf transformer ormer faults faults,” ,” IEEE Computer Applications Power, vol. 13, no. 1, pp. 50-55, 2000. [77] N. Al[77] Al-Kha Khayat yat and L. Hay Haydoc dock, k, “Swe “Swept pt fre freque quency ncy response response test for cond co nditi ition on mo moni nito torin ringg of po powe werr tra trans nsfo form rmer er,” ,” in Proc. Electrical/Electronics Insul. Conf., Chicago, IL, 1995. [78] A. Drobishevskey, E. Levitzkaya, and M. Filatova, “Application of LV impulses for diagnostics of transformers during tests and in service,” presented at the 8 th Intern Internationa ationall Sympos Symposium ium on High Voltage Engineering, Engi neering, Yokohama, Japan, 1993. [79] A.J. Kachler, et al., “High voltage impulse tests on power transformer transf ormerss using a digita digitall measur measuring ing system,” in Proc. 5th Int. Symp. High Volt. Eng., Braunschweig, Stadthalle, 1987, vol. 3, paper 72.05. [80] E.L. White, “Monitoring transformers in service for winding displa dis placem cement ent usi using ng the low vol voltage tage imp impuls ulsee met method hod,” ,” ERA Technology, Rep. No. 81-62R, June 1981. [81] P. Tantin, P.V. Goosen, and P. Christensen, “CIGRÉ SC 12 Power Transformers Special Report 1998,” presented at the International Council on Large Electric Elect ric Systems (CIGRÉ), Paris, France, 1998. [82]] “Gui [82 “Guide de forthe Sam Samplin plingg of Gas Gases es andOil fro from m OilOil-Fi FilledElectr lledElectrical ical Equipment and for the Analysis of Free and Dissolved Gases,” IEC Publication 567, 1992. [83]] IEE [83 IEEE E Guid Guidee for the Int Interp erpreta retatio tionn of Gas Gases es Gen Genera erated ted in Oil-Immersed Transformers, Transformers, IEEE Std. S td. C57.104-1991, 1991. [84] R.R. Roger, “IEEE and IEC codes to interpret incipient faults in transf tra nsform ormers ers,, usin usingg gasin oil ana analys lysis,”IEEE is,”IEEE Tran rans. s. Ele Elec. c. Ins Insul.,vol. ul.,vol. 13, no. 5, pp. 349-354, 1978. [85] “Interpretation of the analysis of gases in transformers and other oil-filled electrical equipment in service,” IEC Publication 599, IEEE Electrical Insulation Magazine [87] PP.J. .J. Burton, et al., “Recent “ Recent developments by CEGB to improve the prediction predict ion and monito monitoring ring of trans transformer former perfo performanc rmance,” e,” presen presented ted at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1984. [88] B. Fallou, “Detection of and research for the characteristics characteristi cs of an incipient fault from analysis of dissolved gases in the oil of an insulation,” Electra, no. 42, pp. 31-52, 1975. [89]] A. De Pa [89 Pabloand bloand A. Mol Mollma lmann, nn, “Newguidel “Newguideline iness forfuran forfuranss ana analys lysis is as well as dissol dissolved ved gas analys analysis is in oil-fi oil-filled lled transf transformers ormers,” ,” presen presented ted at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1996. [90]] A. Mol [90 Mollma lmann nn and B. Pa Pahla hlavan vanpou pour, r, “Ne “New w gui guideli delines nes for interpretation of dissolved gas analysis in oil-filled transformers,” Electra, no. 186, pp. 30-51, 1999. [91] D. Kopa Kopaczyns czynski, ki, “AC dielectr dielectric-los ic-loss, s, power power-facto -factorr and capaci capacitance tance measurements as applied to insulation systems of high-voltage power apparatus in the field. Part 1: Dielectric theory. Part 2: Practical applications of Doble testing,” Doble Client Conference Proceedings, Apr. 1993. [92] IEEE Standard Test Code for Liquid Immersed Distribution, Power Po wer,, and Reg Regula ulating ting Tran ransfo sforme rmers rs and IEE IEEE E Gui Guide de for Short-circuit Testing of Distribution and Power Transformer, Transformer, IEEE Std. C57.12.90-1999, 1999. [93]] A. Zar [93 Zargar garii and T.R. Bla Blackbu ckburn, rn, “App Applica licatio tionn of opt opticalfiber icalfiber sen sensor sor [97] A. Bognar, et al, “Diagnostic tests of high voltage oil-paper insulating systems (in particular transformer insulation) using dc dielectrometrics,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1990. [98] E. Nemeth, “Measuring voltage response: A non-destructive diagnostic diagno stic test method of HV insula insulation,” tion,” IEE Proc. Sci., Meas. Technol., vol. 146, no. 5, pp. 249-252, 1999. [99] [9 9] T. Le Leib ibfr frie ied, d, et al al., ., “O “Onn-li line ne mo moni nito tori ring ng of po powe werr transformers—Trends, new developments and first experiences,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1998. [100] T. Leibfried, “Online monitors keep transformers in service,” IEEE Computer Applications in Power, vol. 11, no. 3, pp. 36-42, 1998. [101] M.A. Sanz-Bobi, et al., “Experiences learned from the on-line internal monitoring of the behavior of a transformer,” presented at the IEEE International Electric Machines and Drives Conference, Milwaukee, WI, 1997. [102] C.K. Mechefske, “Correlating power transformer tank vibration characteristics to winding looseness,” Insight, vol. 37, no. 8, pp. 599-604, 1985. [103] T. Bengtsson, et al., “Acoustic diagnosis of tap changers,” presented at the International Council on Large Electric Systems (CIGRÉ), Paris, France, 1996. for partial discharge detection in power equipment,” in Proc. CEIDP , San Francisco, CA,high-voltage 1996, pp. 541-544. [94] R. Dorr Dorr,, et al., “On-line transformer transformer monito monitoring: ring: Detection of partial discharges from HF measurements using FFT and time domain filters,” in Proc. 12th Int. Symp. High Volt. Eng. , vol. 5, Bangalore, India, 2001, pp. 1230-1233. [95] R.S. Brooks and G.S. Urbani, “Using the recovery voltage method to evaluate aging in oil-paper insulation,” in Proc. IEEE Int. Conf. Conduc. and Breakdown in Solid Dielec. , Vasteras, Sweden, 1998, pp. 93-97. [96]] G. Cse [96 Csepés pés,, et al. al.,, “Re “Recov covery ery vol voltag tagee met method hod for oil/paper oil/paper insulation insula tion diagno diagnosis,” sis,” in Proc. CEIDP, At Atla lant nta,GA, a,GA, 19 1998 98,, vo vol. l. 1,pp. 345-355. [104] A. Golubev, et se al., “On-line alternative to the frequen fre quency cy res respon ponse analys ana lysis is vibro-acoustic and onon-lin linee par partia tiall dis discha charge rge measurements on large power transformers,” presented at the Tech Con’99 Annual Conference of TJ/H2b, New Orleans, LA, 1999. [105] R.J. Denis, S.K. An, J. Vandermaar, and M. Wang, “Comparison of two FRA methods to detect transformer winding movement,” pres pr esen ented ted at th thee EP EPRI RI Sub Subst stat atio ionn Eq Equip uipme ment nt Di Diag agno nost stic icss Conference VIII, New Orleans, LA, 2000. [106] A.J. Vandermaar, “On-line transformer winding and dielectric monitoring-laboratory and field test results,” EPRI, TR-113650, Sept. 1999. [107] M. Waters, The Short-Circuit Strength of Power Transformers . London, U.K.: Macdonald & Co. (Publishers) Ltd., 1996. November/December 2002 — Vol. 18, No. 6 25
