Asset Management for Power Transformer in High Voltage Substation 9

Asset Management for Power Transformer inHigh Voltage Substation

Thanapong Suwanasri1 ,

Rattanakorn Phadungthin2 , and Cattareeya Suwanasri3 , Non-members

ABSTRACTIn this paper, an innovative asset management in

inventory control of power transformer in high volt-age substation is proposed. Using Pareto analysis,the components of power transformer are classifiedinto ABC Classes. Class A means the few most ex-pensive ones that need special care. Class B meansordinary ones that need standard care. Class C meansthe large number of cheap items that need little care.The inventory management strategies as statisticaldistribution methods, economic ordering policy, andtwo-bin policy are initiatively applied to the trans-former components in the ABC Classes. Finally, theoptimum number of spare parts, the optimum order-ing quantity, the suitable time for reordering, and thesaving inventory cost are determined in order to mini-mize the total inventory cost. The power transformerfleet in Thailand is presented as example of inventorycontrol of transmission asset.

Keywords: Asset Management, Inventory Control,Pareto Diagram, Economic Ordering Quantity, Pois-son Distribution, Normal Distribution, Two-bin Pol-icy

1. INTRODUCTION

High voltage equipment such as power trans-former, circuit breaker, disconnecting switch, surgearrester, etc. are important components in powersystem. Especially, power transformer is graduallydeteriorated every day such as shortened lifetime dueto oxidation, moisture and temperature, acidity andcontamination of insulating oil, seal and gasket dete-rioration. If the maintenance cannot detect and fixthe problem in time, the catastrophic failure will oc-cur, which leads to tremendous damage and outagecost to electric utility [1]. Therefore, high voltageequipment must be maintained its satisfactory oper-ating condition by applying an effective maintenancestrategy. Maintenance, availability and reliability are

Manuscript received on June 6, 2012 ; revised on November26, 2012.1,2 The authors are with The Sirindhorn Interna-

tional Thai - German Graduate School of Engineering,Bangkok, Thailand., E-mail: thanapongs@kmutnb.ac.th andrpt.kmutnb@gmail.com3 The author is with Faculty of Engineering King Mongkut’s

University of Technology North Bangkok, Bangkok, Thailand.,E-mail: cattareeyas@kmutnb.ac.th

closely related so that a level of maintenance shouldbe specified to ensure an acceptable level of trans-former reliability. Maintenance strategies are mainlyclassified into corrective, preventive, condition-basedand reliability-centered.

The corrective maintenance for replacement or re-pair is needed when a failure occurred. It is mainlyused in systems with lower voltages. Advantages ofthe corrective maintenance are low cost and man-power saving, whereas one of disadvantages is thatit might be too late to be repaired if failures are notdetected early [2]. The preventive maintenance is rou-tine and basic maintenance in fixed intervals for in-spections and maintenance. Advantages are lifecycleincreasing and fault inception detecting. However,disadvantages are that if some parts of the equip-ment might be timely maintained, damages mightoccur before the maintenance. If the parts are tooearly maintained, it will be expensive due to suchmaintenance and unnecessary shutdowns [3]. Thecondition-based maintenance is followed by its condi-tion or the condition is taken into account. All majorparameters such as diagnostic methods are consideredin order to determine the condition with maximizedaccuracy. Advantages are that the maintenance isdone when needed; and costs as well as manpowerare saved. Disadvantages are that experienced peo-ple and suitable data are required [4]. The reliability-centered maintenance (RCM) considers the conditionas well as the importance of equipment. It evalu-ates priority for maintenance actions and ranks thereplacement and refurbishment activities. If the eco-nomical consequences from different actions are con-cerned, the RCM is extended to risk-based mainte-nance [5]. However, even if the RCM is cost opti-mizing method based on risk and unnecessary shut-down, disadvantages are that it is a complex modeland lots of data are required. The most appropri-ate maintenance strategy to maintain the equipmentfor the electric utilities is preventive maintenance be-cause of simple, noncomplex, and minimum data re-quired method. However, its disadvantages in spareparts and costs of inventory should be mitigated.

In this paper, the effective inventory managementstrategy for high voltage equipment in a Thai utilityis proposed. The methods to determine the optimumnumber of spare parts, the optimum ordering quan-tity, the suitable time for reordering, and the saving

10 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.11, NO.1 February 2013

inventory cost are suggested in order to minimize thetotal inventory cost of high voltage equipment. Sincethe major components of power transformer as shownin Fig. 1 are usually expensive and have low turnoverdemand of inventory. Thus, a number of such com-ponents stored in stock are limited by capital invest-ment cost. Power transformer components are classi-fied in ABC Classes using Pareto analysis. For ClassA, bushing and arrester are proposed. The failurerecords of power transformer are analyzed by statis-tical techniques as Normal and Poisson distributionsin order to determine failure rate and replacementrate. The optimum number of spare parts can be de-termined according to the required stock reliability.For Class B, the insulating oil is given as an exam-ple. The amount and time for reordering insulatingoil is proposed. For Class C, the spare parts of cheapitems of power transformer as nuts, bolts and gasketsare presented by using two-bin policy. This proposedstrategy performs the revolution to the new era of ef-fective preventive maintenance in order to optimallycontrol the spare parts of HV equipment by analyz-ing the statistical failure records and inventory costs.Consequently, the overall cost of maintenance can bereduced.

Fig.1: Power Transformer’s Components.

2. REVIEW OF SPARE PART MANAGE-MENT

The availability of spare parts is a significant sub-ject of effective maintenance strategy because it canreduce downtime or supply interruption due to equip-ment failure. A large amount of spare parts stored inthe stock can provide the high reliability of equip-ment utilization and reduction of downtime and re-pair time, but it is not cost-effective method due tohigh initial investment cost. Therefore, the optimumnumber of spare parts in stock must be determinedin order to minimize the cost of inventory. Severalmethods were applied for the effective inventory man-agement. In [6], aircraft spare parts were performedby using an organizational perspective on inventory

control to ensure maximum availability of aircraft foroperational use. However, the proposed method wasappropriate only for small inventory system.

In [7], three probabilistic models as Poisson dis-tribution, Markov process, and chronological MonteCarlo simulation were compared to assess the reli-ability and inventory of spare transformers in dis-tribution substations. The reliability of the systemwas assessed in the aspect of when inventory can-not be timely replenished by using Poisson method.However, the Markov and chronological Monte Carlomodels were proposed to extend the capability of re-liability assessment. Only spare parts for distributiontransformers were determined.

In [8], the Homogeneous Poisson Process (HPP)method was applied as a mathematic model to thefailure data of distribution transformers in order topredict proper number of spare distribution trans-formers. This method did not require a complete ho-mogeneity of the failure data; but it was based on thechi-square goodness of fit test.

In [9], the probabilistic approach was proposed todetermine the number and timing of spare transform-ers in medium, voltage substation. The aging failureof transformers, overall reliability, probabilistic dam-age cost, and capital cost of spare parts were ana-lyzed. However, only non-repairable failure compo-nents were considered in the inventory. In addition,the Poisson probability distribution was mentioned todetermine the optimal number of transformer’s sparepart for distribution transformer [10]. The numberof spare requirements for a system depended on thereliability level demanded from the system. The eco-nomic model was involved to minimize the total sys-tem cost.

In [11], the economic order quantity was proposedfor the practical inventory control. The stock onhand, net stock, optimum spare part, order size, andorder point were analyzed. This method was simpleand applied suitably to the complex task with var-ious control variables. Furthermore, the inventorycost optimization was investigated.

Therefore, this paper discusses two probabilisticdistribution methods, which are Poisson and Normaldistributions, to determine the number of spare partsin Class A of high voltage power transformer becausethese two techniques are suitable for the repairablecomponents with sufficiently available failure data.Regarding to aforementioned advantages of economicorder quantity, this method is proposed for Class Bspare parts of power transformer. In Class C, due toinexpensive items, a basic inventory control method,so-called the two-bin policy is introduced in this pa-per. With this effective inventory management, elec-tric utility or company can not only reduce the directcost of inventory, but also can significantly reducehidden capital costs such as supply interruption andincrease reliability of electricity supply.

Asset Management for Power Transformer in High Voltage Substation 11

3. SPARE PARTS CLASSIFICATION

Pareto diagram [12-13] is a general method to clas-sify spare parts or item listed in the inventory intoA, B, C Classes according to the cost of items andamount of usage. Thus, spare parts of high voltageequipment in power system can be classified by us-ing the aforementioned Pareto diagram. Usually ininventory system, the higher cost items account fora smaller portion of a total number of items. Theanalysis starts from; firstly the cost of each item iscalculated as a percentage of the total cost of inven-tory items. Then, the items, which are in descendingorder of the percentage of the item’s cost to the to-tal inventory cost, are ranked starting from the itemsthat contribute the highest cost. Finally, the graphis plotted with percentage of item used on the X axisand percentage of its cost on the Y axis as shown inFig. 2.

Fig.2: ABC Class Analysis in Pareto Diagram.

Using Pareto analysis in Fig. 2, the componentsof high voltage equipment can be classified into ABCClasses.• Class A items are account for 70% to 80% of the

total inventory cost but about 20% of total inven-tory items. This class needs a high capital invest-ment and requires a close control. Then it will beordered based on calculations of the most economicorder quantities. Due to high cost of these items,a minimum safety stock is maintained.

• Class B items are account for 20% to 30% of thetotal cost but about 20% to 30% of total items.Compared to Class A, the larger safety stocks maybe maintained and do not need a special control.

• Class C items are account for 20% of thetotal cost but about 70% to 80% of total items.

Fig. 2 shows the relationship between yearly per-centage costs of inventory versus yearly percentage ofinventory items of ABC Classes. Thus, the cost ofitem usage in each class can be calculated from thecumulative number of item used multiply to the priceper item.

After the items are classified into three different

ABC Classes, different inventory policy has been ap-plied for each class. For Class A, it needs a closecontrol and number of spare parts is based on mini-mum safety stock with acceptable stock availability,so that the statistical distribution techniques are ap-plied to this class. Economic order quantity is usuallyemployed to Class B item to minimize inventory cost,while Class C item is managed by a simple two-binpolicy. The procedure is summarized in Fig. 3.

Fig.3: Spare Part Classification and Inventory Con-trol.

4. SPARE PARTS CALCULATION

4.1 Class A Item of Power Transformer

Since items in Class A are the most expensive butused only few, e.g. bushing, surge arrester, on-loadtap changer of the power transformer; the amountof spare part in this class is also accounted onlyfew items. To find the optimum amount of spareparts, the statistical distribution techniques are ap-plied. The failure records are analyzed in order to de-termine failure rate or replacement rate of each com-ponent. Subsequently, the known failure or replace-ment rate is calculated by using statistical distribu-tion techniques as the Normal and Poisson distribu-tion techniques to determine the optimum number ofspare parts.A. Normal Distribution Technique [14]

Normal distribution technique is a highly accuratetechnique, especially when the number of recordedfailure data is sufficient with adequate informationsuitable for the analysis and the first energization orstartup date or manufacturing date must be known.First, the number of service years is calculated fromthe sum of operating time that every piece of compo-nent is in service or until failed date of prematurelyfailed components. Therefore, the service year time isa time span from the first startup date until presentfor healthy components in service. Then the failurerate is calculated; and finally, the optimum stock forsmooth operation within the equipment lifetime canbe determined from:

12 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.11, NO.1 February 2013

N = M(λ · T +√λ · T · Z) (1)

λ = failure rate =number of failures

number of service years(2)

where: M = number of the items used in the system,T = lifetime of considered item. Z can be obtainedfrom the pre-calculated standardized normal tables.Thus from Table in [15], if availability in stock is re-quired as 99% of service level, then Z is equal to 2.33.B. Poisson Distribution Technique [16]

This technique is an effective tool to analyze a fewrecorded failure data within a specific time intervals(Ti). The Mean Time between Failure (MTBF ) orMean Time to Replacement (MTTR) can be calcu-lated from:

MTTR(µ) =MxTi

Nf(3)

where M is number of the items used in the sys-tem. Ti is time intervals for replacement rate con-sideration. It is usually counted from present to aspecific time in the past in order to count the num-ber of failure occurring with each component. Nf isnumber of failures occurring within time interval Ti.The replacement rate (λ) can be written as:

λ =1

µ(4)

and the expected number of demand (A) is

A = MλT (5)

when T is lead time of stock ordering. The probabil-ity of item (P ) will be kept in time is

P =AI × e−A

I!× 100 (6)

where I is number of the items kept in the stock. Fi-nally, the percentage of stock reliability or availabilityof items in the stock within the time interval is

%RI =I∑

i=0

Pi (7)

The statistical method proposed in this paper canalso be applied to other components of high voltageequipment.

4.2 Class B Item of Power Transformer

B items are ordinary ones that need standard care.For Class B item, three categories as economic orderquantity, safety stock, and reorder level based on de-mand rate can be applied for inventory control of thisclass.

Economic Order Quantity: The main objective ofeconomic order quantity [17-18] is to minimize costof operating inventory system and overall cost. The

relevant costs considered to be significant are cost ofitem, procurement costs and carrying cost. Thesecosts are related as the following relationship.

TC = CD + SD

Q+ Ic

Q

2(8)

where TC is total annual cost, C is item cost, Dis annual demand, S is ordering cost, Q is quantityordered, and Ic is inventory carrying cost.

By taking partial derivatives of TC with respectto Q in (9), the order quantity Q can be calculatedin (10).

∂TC

∂Q= 0 + (−SDQ−2) +

Ic2

(9)

Q =

√2DS

Ic(10)

For example, an item has following data. Theannual demand (D) is 200 units, Cost/item is $10,Cost of placing order (S) is $15 and Annual carry-ing charges is 15% of the item cost. When all itemsare substituted into (10), thus Q is approximatelyequal to 64 units. By applying this category, the partshould be ordered as 64 units.

Safety Stock: Safety stock is established to preventstock outage during lead time period. It is usuallyapplied when the lead time is fluctuated [19]. Thesafety stock can be determined as:

Safety stock = R− µ = σZ (11)

where R is reorder level, µ is expected demand dur-ing lead time, σ is standard deviation, and Z can beobtained from standardized normal table.

For example to compute safety stock for HV bush-ing, the lead-time demand of a certain type of HVbushing is normally with a mean (µ) of 150 units anda standard deviation (σ) of 5 units. The managerwants to pursue a policy whereby the HV bushing isnot available only 1% of the time when demanded.Thus from Table in [14], for 99% service level, Z canbe obtained from standardized normal tables. ThusZ is equal to 2.33. So, safety stock (R − µ) is equalto 11.65 ≈ 12 units.

Reorder Level: The reorder level of inventory con-trol system shown in Fig. 4 refers to amount of itemin inventory, at which level order should be triggered[15].

It is calculated by the known demand during leadtime to make sure that the new order quantity will bereceived before the item in inventory reaches 0. Thereorder level can be calculated as:

R = µ+ σZ (12)

From the previous example, the replenishment or-der should be placed when unit is R = 150+12 = 162units.

Asset Management for Power Transformer in High Voltage Substation 13

Fig.4: Reorder Level Management.

4.3 Class C Item of Power Transformer

C items are the large number of cheap items thatneed little care. Two-bin policy [14] is applied for thisclass as presented in Fig. 5.

Fig.5: Two-bin Inventory Control.

Items are kept in the maintenance shop in two-binof equal sizes. Items are withdrawn from the firstbin. When its contents are exhausted, it is time toreorder. The second bin size is sufficient to satisfyexpected demand during the lead time period and asafety stock in order to avoid a possible stock outage.

5. RESULT AND ANALYSIS

The power transformers in the Thai power systemare presented as examples. Power transformer’s com-ponents are classified into ABC Classes according toPareto diagram as given in Table 1. The componentsof power transformer in Class A are busing, arresterand on-load tap changer, etc. The Class B item ofpower transformer is insulating oil. The item in ClassC has the cheapest price and needs a minimal con-trol, e.g. seal, gasket, bolt and nut. In this paper,the calculation for the optimum spare part numberof items in Classes A, B and C are presented.

5.1 Class A Item of Power Transformer

A. Transformer BushingThe 115 kV bushings will be investigated as an

example for spare part management for a fleet ofpower transformers in transmission network. Thedata of 120 bushings from 40 power transformers rat-ing 115/22 kV, 25 MVA show that the total serviceyears is equal to 1,890 unit-years. The design life-time of bushing is usually specified as 25 years. From

Table 1: Power Transformer Component Classifi-cation.

Class Power Transformer’s Components

A

1. Active parts: core, winding (HV/LV/TV)2. Bushing (HV/LV/TV)3. Cooling system: fan, pump4. Protective device: pressure relief, Buchholz relay,

OLTC pressure relief, OLTC protective relay,OLTC pressure relay, OLTC oil flow relay, oiltemperature, gas detector, sudden pressure relay,winding temperature, percentage differentialrelay, over current relay

5. OLTC: contact of diverter switch, contact ofselector switch

6. Surge arrester (HV/LV/TV)7. Tank: diaphragm/rubber bag

B Insulating oil in main tank and in OLTCC Others as nuts, bolts, o-rings, gaskets, etc.

the recorded failure, there are 5 failure records within10 years time interval. If the need for availability ofitems in stock is 99% of service level, the number of115 kV bushing, which should be kept in the stock,can be calculated as follows.

1)Normal distribution technique: Failure rate isequal to 5/1,890 = 0.002646 times per year. If therequired availability is 99%; then from the standard-ized normal table, Z = 2.33. Thus, the number ofbushings in the stock is equal to

N=120×0.002646×25+

√0.002646×25×2.33

25=2.2

or 3 units per year for the total 40 units of powertransformers.

2)Poisson distribution technique: For the samegroup of transformers, Mean Time to Replacement(MTTR) for the specific time 10 years period is cal-culated as

MTTR(m) =M × Ti

Nf=

120× 10

5= 240

Thus, the replacement rate is λ = 1m = 0.00417.

Considering the lead time (T) for ordering bushingis 1 year, the expected number of demand for bushingis equal to A = MλT = 120×0.00417×1 = 0.5. Thus,the probability of having 0.5 bushing available in thestock in one year time interval can be calculated from(6) and reliability from (7). The results are given inTable 2.

From the probability and reliability of bushing instock, it is seen that the 115 kV bushing should bekept in stock 3 units per year in order to obtain the99% reliability of stock. This number is similar to theresult obtained from the Normal distribution tech-nique.

B. Surge ArresterWith the same method, the 115 kV surge arrester

will be further investigated. The data from a similar

14 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.11, NO.1 February 2013

Table 2: Probability and Reliability of BushingStock.

group of 40 power transformers rating 115/22 kV, 25MVA shows that the total service year of 120 arrestersis equal to 2,350 unit-years. From the recorded fail-ure, there are only 2 failure records occurring within10 years time interval. The design lifetime of arresteris also specified as 25 years. If the need for avail-ability of items in stock is 99% of service level, thenumber of 115 kV arresters, which should be kept inthe stock, can be calculated as follows

1) Normal distribution technique: Failure rate isequal to 2/2,350 = 0.000851 times per year. If therequired availability is 99%; then from the standard-ized normal table, Z = 2.33. Thus, the number ofarresters is

N=120×0.000851×25+

√0.000851×25×2.33

25=1.2

or 2 units per year for the total 40 units of powertransformers.

2)Poisson distribution technique: For the samegroup of transformers, the Mean Time to Replace-ment (MTTR) is calculated as:

MTTR(m) =M × Ti

Nf=

120× 10

2= 600

Thus, the replacement rate is λ = 1m = 0.00167.

Considering the lead time (T) for ordering arresteris also 1 year, the expected number of demand for ar-rester is equal to A = MλT = 120×0.00167×1 = 0.2.Thus, the probability of having 0.2 arrester availablein the stock in one year time interval can be calcu-lated from (6) and reliability from (7). The resultsare given in Table 3. From the probability and reli-ability of arrester in stock, it is seen that two unitsof 115 kV arresters should be kept in stock per yearin order to obtain the 99% reliability of stock. Thisnumber agrees with the result obtained from the Nor-mal distribution technique.

Table 4 and Table 5 summarize the number ofbushings and arresters per year for various powertransformer ratings in the system. For instance by us-ing both Normal and Poisson distribution techniques,the number of 500 kV bushing for 333 MVA powertransformers is 2, while that of 115 bushing for 100

Table 3: Probability and Reliability of ArresterAvailable in the Stock.

Table 4: Number of Bushing per Year for VariousTransformer Rating.

Table 5: Number of 115 kV Arrester Spare Partsfor 50 MVA Transformer.

MVA power transformers is 1. The number of 115 kVarresters for 50 MVA power transformers is 5 and 7 byusing Normal and Poisson distributions, respectively.

5.2 Class B Item of Power Transformer

For Class B, the optimum number of ordering insu-lating oil is presented. The economic order quantityis applied to determine the optimum ordering quan-tity of insulating oil. By using the data in Table 6,

Asset Management for Power Transformer in High Voltage Substation 15

substituting in (10), the optimum order quantity ofinsulating oil is 175 tanks/times. Note that 1 tank =200 liters.

Table 6: Insulating Oil Ordering Information.Items Data

Annual demand (D) 540 tanks/yearCost of item (C) 13,402 Baht/tankCost of placing order (S) 60,000 Baht/timeAnnual carrying cost (Ic) 2,116 Baht/tankLead time 90 daysDemand during lead time (µ) 133 tanksStandard deviation (σ) 5 tanks/90daysProbability of unit not available 1%

Fig.6: Ordering Schedule of Insulating Oil.

For the 99 percents of confidence which is surelythat insulating oils are available when demand oc-curred, Z can be obtained from standardized normaltables as 2.33. By substituting data in (11), σZ =12 tanks. So, the safety stock for insulating oil is12 tanks. It can make sure that 99 percent duringlead time, stock is not outage when the demand oc-curred. Then, to determine the appropriate time forplacing an order, substituting safety stock and meandemand during lead time into (12), R is 133+12 whichis equal to 145 tanks. It means that when insulatingoils in stock decrease to 145 tanks, the order of 175tanks should be placed. If daily demand is constant,it will take around 4 months to place an order again.Therefore, it is around 3 times per year for orderinginsulating oil, as shown in Fig. 6.

The total annual cost is calculated by using theresults from economic order quantity method. Bysubstituting the calculated data into (8), the totalannual cost is accounted around 7,602,230 THB/year.However, if placing an order once a year, the total an-nual cost will increase to 7,868,400 THB/year. Thus,by following the economic order quantity, the utilitycan save the budget for ordering insulating oil around266,170 THB/year.

5.3 Class C Item of Power Transformer

The components in Class C for transformer areusually small or low value items such as nuts, bolts,o-rings, gaskets, etc. Therefore, the items are ordered

as a large quantity but low budget. As shown in Fig.5, when the first bin of such components is used up,an order is made out for replenishment. The secondbin contains enough quantity of the item to last untilthe ordered quantity arrives.

6. APPLICATION TO HV EQUIPMENT INPOWER SUBSTATION

The proposed inventory management strategy inthis paper can be further applied to other high voltageequipment in power system. Example on power cir-cuit breaker is explained. The components of powercircuit breakers can be similarly classified into ABCClasses based on Pareto diagram and according tothe IEEE standard [20], as written in Table 7.

Table 7: HV Circuit Breaker Component Classifi-cation.

Class HV Circuit Breaker’s Components

A

1. High-voltage components: interrupter, auxiliaryinterrupter, insulation to ground2. Mechanical parts: energy storage, controlelements, actuator3. Operating mechanisms: compressors, pump,mechanical transmission

B1. Electrical control and auxiliary circuits: trip andclose circuits, auxiliary switches, contactors, etc.2. Insulating medium such as oil, SF6

C Others, nuts, bolts, o-rings, gaskets, etc.

The high voltage components of power circuitbreaker such as interrupter, auxiliary interrupter arecategorized into Class A because their high capitalinvestment and close control are required. The in-sulating medium such as oil and SF6 is categorizedinto Class B that needs standard care, whereas in-expensive items such as nuts, bolts, and o-rings arecategorized into Class C that needs little care.

Similar to the strategies applied to power trans-former components, to optimize the total inventorycost of power circuit breaker, the statistical distribu-tion techniques with the Normal and Poisson distri-butions are applied to Class A items of power circuitbreakers. The economic order quantity and the two-bin policy are applied to Class B and Class C, respec-tively. The calculation and analysis can be performedaccordingly.

7. CONCLUSION

This paper presents the cost-effective inventorymanagement strategy for high voltage equipment inpower system. The spare parts of power transformerare categorized into ABC Classes on a basis of thePareto analysis. Class A items are spare parts, whichare ordered by considering failure rate of componentsand using the statistical distribution techniques thatare the Normal and Poisson distributions. EconomicOrder Quantity model is applied to estimate the op-timum number of spare parts in Class B, whereas

16 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.11, NO.1 February 2013

two-bin policy is handled to place an order of spareparts in Class C.

The 115 kV bushings and 115 kV arresters of powertransformers rating 115/22 kV 25 MVA are investi-gated thoroughly for spare part management as ClassA. The result shows that Normal distribution tech-nique provides the optimum number of spare partssimilar to Poisson distribution technique for bothbushings and arresters. Besides, the optimum num-ber of bushings and arresters for various power trans-former ratings in the system are determined. Only afew numbers of bushings and arresters should be keptin the stock each year. Insulating oil of power trans-former is examined as Class B. The result indicatesthat the order of 175 tanks should be placed, whilethe safety stock and the reorder level are 12 tanksand 145 tanks, respectively. The cheapest price itemssuch as nuts, bolts, o-rings and gaskets are mentionedfor inventory management as Class C.

This innovative inventory management is now im-plemented in an electricity utility and can effectivelyreduce the total inventory cost with acceptable relia-bility in operation of power transformer.

8. ACKNOWLEDGEMENT

The authors gratefully acknowledge the Transmis-sion System Maintenance Division at Electricity Gen-erating Authority of Thailand for data support.

References

[1] Cigre Working Group A2.34, Guide for Trans-former Maintenance, February 2011.

[2] J. Schneider, A. Gaul, C. Neumann, J. Hograe-fer, W. Well ow, M. Schwan, A. Schnettler, AssetManagement Techniques, Power Systems Com-putation Conference, Liege, August 2005.

[3] Cigre Working Group C1.1, Asset Managementof Transmission Systems and Associated CigreActivities, December 2006.

[4] G. Balzer, F. Heil, P. Kirchesch, R. Meister,C. Neumann, Evaluation of Failure Data of HVCircuit-Breakers for Condition Based Mainte-nance, Cigre Session 2004.

[5] M. Schwan, W. H. Well ow, A. Schnettler, U.Zickler, M. Roth, J. Schneider, Risk-based As-set Maintenance for Substations in Distribu-tion Networks Considering Component Reliabil-ity, Cigre 2006.

[6] L. G. Zomerdijk, J. De Vries, An OrganizationalPerspective on Inventory Control: Theory and aCase Study, International Journal of ProductionEconomics, 2003.

[7] A. M. Leite da Silva, J. G. De Carvalho Costa,A. A. Chowdhury, “Probabilistic Methodologiesfor Determining the Optimal Number of Substa-tion Spare Transformers," IEEE Transactions onPower Systems, Vol. 25, No. 1, February 2010.

[8] V. I. Kogan, C. J. Roeger, D.E. Tipton, “Sub-station Distribution Transformers Failures andSpares," IEEE Transactions on Power Systems,Vol. 11, No. 4, November 1996.

[9] W. Li, E. Vaahedi, Y. Mansour, “DeterminingNumber and Timing of Substation Spare Trans-formers Using a Probabilistic Cost Analysis Ap-proach," IEEE Transactions on Power Delivery,Vol. 14, No. 3, July 1999.

[10] A. A. Crowdhury, D. O. Koval, “Developmentof Probabilistic Models for Computing Opti-mal Distribution Substation Spare Transform-ers," IEEE Transactions on Industry Applica-tions, Vol. 41, No. 6, November 2005.

[11] P. Alstrom, “Numerical Computation of Inven-tory Policies, Based on the EOQ/σx value forOrder-Point Systems," International Journal ofProduction Economics, 2001.

[12] A. Tanwari, A. Qayoomlakhiar, G. Y. Shaikh,“ABC Analysis as a inventory Control Tech-nique," Quaid-E-Awam University ResearchJournal of Engineering, Science and Technology,Vol. 1, No. 1, January-June 2000.

[13] Y. Chen, K. W. Li, J. Levy, K. W. Hipel, D.M. Kilgour, A Rough Set Approach to MultipleCriteria ABC Analysis, Springer LNCS Trans-actions on Rough Sets, in press, 2008.

[14] D. Waters, Inventory Control and Management,John Wiley & Sons Inc., 2003.

[15] O. Salih Duffuaa, A. Raouf, J. Dixon Campbell,Planning and Control of Maintenance Systems,John Wiley & Sons Inc., 1999.

[16] D. Louit, R. Pascual, D. Banjevic, “OptimizationModels for Critical Spare Parts Inventories - Re-liability Approach," Journal of the OperationalResearch Society, 2010.

[17] J. Viale, David, Basics of Inventory Manage-ment: from Warehouse to Distribution Center,Thomson Course Technology, 1996.

[18] H. Kai Chan, F. T. S. Chan, “Early Order Com-pletion Contract Approach to Minimize the Im-pact of demand Uncertainty on Supply Chains,"IEEE Transactions on Industrial Informatics,Vol. 2, No. 1, February 2006.

[19] Z. Lianfu, Z. Shuzhi, W. Min, Z. Zhihui, Z. Yong-gang, “Analyzing on Impact Factors of Safetystock under Random Requirement," Interna-tional Conference on Networks Security, Wire-less Communications and Trusted Computing,2009.

[20] IEEE Std. C37.10TM-1995 (R2002), IEEE Guidefor Diagnostics and Failure Investigation ofPower Circuit Breakers, 1995.

Asset Management for Power Transformer in High Voltage Substation 17

Thanapong Suwanasri received hisB.Eng. from King Mongkut’s Instituteof Technology North Bangkok, Thailandin 1993, M.Sc. from Rensselaer Poly-technic Institute, NY, USA in 1995 andDr.-Ing. in High Voltage Technologyfrom RWTH Aachen University, Ger-many in 2006, all degrees in electricalengineering. Currently he is an Assis-tant Professor and Head of Electricaland Software System Engineering de-

partment at the Sirindhorn International Thai-German Grad-uate School of Engineering (TGGS), King Mongkut’s Univer-sity of Technology North Bangkok, Bangkok, Thailand. Hisresearch interest includes power transformer, power circuitbreaker, asset management, condition based maintenance andmaintenance strategy.

Rattanakorn Phadungthin receivedher B.Eng. from Chiang Mai Univer-sity, Thailand in 1993, M.BF. from Uni-versity of Technology, Sydney, Australiain 1997 and M.Sc. from the Sirind-horn International Thai-German Grad-uate School of Engineering (TGGS),King Mongkut’s University of Technol-ogy North Bangkok, Thailand in 2005.Currently, she is a Ph.D. candidate atKing Mongkut’s University of Technol-

ogy North Bangkok. Her research interests are diagnostic tech-nique and economic analysis in power transformer asset man-agement.

Cattareeya Suwanasri received herB.Eng. from Khon Kaen University,Thailand, M.Eng. and D.Eng. fromAsian Institute of Technology, Thailand,with the Sandwich Program at Insti-tute of Power System and Power Eco-nomics (IAEW), RWTH Aachen Univer-sity, Germany, in 1998, 2002 and 2007,respectively. Currently she is a lecturerat Department of Electrical and Com-puter Engineering, Faculty of Engineer-

ing, King Mongkut’s University of Technology North Bangkok,Thailand. Her research interest includes electric power systemmanagement, power system analysis, power economics, assetmanagement, and high voltage engineering.