Asset Management Techniques

7/28/2019 Asset Management Techniques

1/12

Asset management techniques

Joachim Schneider a, Armin J. Gaul a,*, Claus Neumann b, Jurgen Hografer c,Wolfram Wellow d, Michael Schwan d, Armin Schnettler e

a RWE Energy AG, Assetmanagement, Rheinlanddamm 24, 44139 Dortmund, Germanyb RWE Transportnetz Strom GmbH Dortmund, Germany

c SAG Energieversorgungslosungen GmbH, Langen, Germanyd Siemens AG, Power Transmission and Distribution, Erlangen, Germanye RWTH-Aachen, Institut fur Hochspannungstechnik, Aachen, Germany

Received 31 March 2006; accepted 31 March 2006

Abstract

Deregulation and an increasing competition in electricity markets urge energy suppliers to optimize the utilization of their equipment,focusing on technical and cost-effective aspects.

As a respond to these requirements utilities introduce methods formerly used by investment managers or insurance companies. Thearticle describes the usage of these methods, particularly with regard to asset management and risk management within electrical grids.The essential information needed to set up an appropriate asset management system and differences between asset management systemsin transmission and distribution systems are discussed.

The bulk of costs in electrical grids can be found in costs for maintenance and capital depreciation. A comprehensive approach for anasset management in transmission systems thus focuses on the life-cycle costs of the individual equipment. The objective of the life

management process is the optimal utilisation of the remaining life time regarding a given reliability of service and a constant distributionof costs for reinvestment and maintenance ensuring a suitable return.In distribution systems the high number of components would require an enormous effort for the consideration of single individuals.

Therefore statistical approaches have been used successfully in practical applications. Newest insights gained by a German research pro-ject on asset management systems in distribution grids give an outlook to future developments. 2006 Elsevier Ltd. All rights reserved.

Keywords: Asset management; Replacement; Maintenance; Component condition

1. Introduction: fundamental asset management tasks

Asset management means operating a group of assetsover the whole technical life-cycle guaranteeing a suitablereturn and ensuring defined service and security standards.

Distribution and transmission network operators arefacing many different and partly even competing targets.

It is their task to find a balance between the requirementsof the customers concerning product and service qualityat affordable prices as well as the shareholder demandsfor suitable returns on the capital they invest. Alsopotential regulatory impacts on revenues and changes inthe political perception of e.g. renewable energy have tobe focussed. To optimize between these demands net-work operators have to develop and extend best prac-tices in asset management. The main question is notWhich network design will provide the best servicequality? but instead, Which network design will provide

0142-0615/$ - see front matter 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ijepes.2006.03.007

* Corresponding author. Tel.: +49 231 438 5532; fax: +49 231 438 385532.

E-mail addresses: [email protected] (J. Schneider), [email protected] (A.J. Gaul), [email protected] (C. Neumann),[email protected] (J. Hografer), [email protected] (W. Wellow), [email protected] (M. Schwan), [email protected] (A. Schnettler).

www.elsevier.com/locate/ijepes

Electrical Power and Energy Systems 28 (2006) 643654
mailto:[email protected]:armin.%[email protected]:armin.%[email protected]:[email protected]:[email protected]:wolfram.wellssow@siemens.%20commailto:wolfram.wellssow@siemens.%20commailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:wolfram.wellssow@siemens.%20commailto:wolfram.wellssow@siemens.%20commailto:[email protected]:[email protected]:armin.%[email protected]:armin.%[email protected]:[email protected]


2/12

better-than-required service quality while maximizingfinancial performance? [1].

Asset management in electrical grid companies plays akey role in the detection and evaluation of decisions leadingto long-term economical success and best possible earnings.For asset management to live up to these expectations, it

has to meet a number of challenges. The four key challengesare (a) alignment of strategy and operations with stake-holder values and objectives; (b) balancing of reliability,safety, and financial considerations; (c) benefiting from per-formance-based rates and (d) living with the output-basedpenalty regime [1].

For this reason fundamental asset management taskscover aspects from technical issues like network planningor the definition of operational fundamentals to more eco-nomical themes like planning of investments and budget-ing, and end up in strategic planning issues.

Causal loop diagrams can be used to visualize the rela-tionship between the elements of the system. Starting on

the left side, Fig. 1 shows that the network conditiondegrades over time due to fundamental ageing processes.This has an impact on network performance which leadsto investments if internal performance targets are not metany more. Investments depend on availability of moneywhich directly results from the demands customers are will-ing to pay for. Finally investments (Capital expenditureCAPEX) or operational expenditure (OPEX) improvethe network condition.

In the following chapters some of the most importantstrategies and management techniques actually in use bygrid operators are described. These are:

maintenance strategies determination of component condition asset simulation statistical fault analysis and statistical asset management

approach (distribution) life management (transmission)

There is growing concern that the pressure for short-term efficiency targets and customer satisfaction has placedasset engineering in a background role. The time constantsinherent in asset degradation and the delays between assetinvestment and realised benefits are sufficiently long toreduce asset investment without unduly affecting short-term performance. Additionally they are much longer thanthe tenure of the regulator and his political masters. There-fore, such strategies could lead to an irreversible time bombsituation with a degree of asset condition degeneration thatmight shatter the benefits current regulator policy seeks toachieve.

In order to make the right decisions it is therefore veryimportant to develop the ability to analyse the complexdependencies between maintenance and renewal actionsand the costs and the quality of service. The ability toassess different scenarios for the whole grid or parts ofthe grid having the same structure or technology is a majorcompetence in asset management. These assessmentsprovide extensive knowledge about the effects of alternativestrategies on the asset base. By using this knowledge assetmanagers can actively develop the grid and spend themoney in a way that the long-term goals as well as theshort-term budgets are met.

Fig. 1. Causal loop diagram.

644 J. Schneider et al. / Electrical Power and Energy Systems 28 (2006) 643654


3/12

2. Maintenance strategies

Maintenance strategies can be divided into differentapproaches which lead to varying maintenance costs andasset availability.

One common way to classify maintenance strategies is

to distinguish whether the condition of the component isconsidered on the one side and whether the importanceof the component is considered on the other side. Fig. 2gives an overview of this classification. Both the conditionand the importance of components can be defined in manyways, depending on the desired level of detail and the avail-ability of appropriate data. For example, the definition ofimportance may be oriented at quite simple aspects likee.g. the number of feeders of a substation, or at moresophisticated indices like e.g. the share of the energy notsupplied in time caused by failure of each respectivecomponent.

The most simple maintenance strategy according to this

classification is the so-called corrective maintenance. Infact, there is no preventive maintenance at all in this strat-egy the component is operated until it fails. Then it isdecided whether the component can be repaired or whetherit must be replaced. In general, this strategy is not themaintenance strategy causing the lowest total cost, as thedamages caused by the failures of the components may cre-ate more cost in the end than a different and more appro-priate maintenance strategy. Besides, this strategy of coursesignificantly derogates supply reliability, which may causeadditional economic consequences for the network opera-tor. It is suitable if the regarded equipment is non-critical

and consequences of the failure are not serious. In distribu-tion networks this method is widely used for MV-XLPEcables, where no preventive maintenance measures areavailable, and usually at the low-voltage level.

So, preventive maintenance is required in order to pre-vent failures and significant damage or even destructionof equipment. The easiest way, and the strategy mostwidely used today, is the so-called time-based maintenance.

There are fixed time intervals for inspections and for cer-tain maintenance works. These time intervals are eithergiven by the manufacturers of the equipment or they arebased on the experience of the network operator. However,it seems that the time intervals chosen in the past were faron the safe side, as there are many inspections revealing no

problems at all. So, the time intervals obviously could beextended the question is from which point on the occur-rence of failures will increase significantly. This method isoften appropriate for those instances where an abrasive,erosive or corrosive wear out takes place and/or materialproperties change due to fatigue, etc. This strategy waswidely used in maintenance of HV- and MV-grids in thepast. Depending on the length of the intervals it combinesan acceptable availability with comparably high mainte-nance costs.

In order to be able to determine the condition of theequipment, additional information on the current condi-tion of the component is required. This current condition

is described by certain indices, and in the concept of condi-tion-based maintenance, maintenance activity is triggeredby the estimated condition reaching certain thresholds.Of course, the determination of the component conditionis not trivial, and several methods are described in the fol-lowing chapters. Condition-based maintenance leads tohigh availability with moderate maintenance costs and ismainly in use within EHV- and HV-grids [3]. Nowadaysa lot of utilities try to adopt this approach also for the med-ium voltage level.

In practice, the limitation of both financial and logisticalmeans requires the additional definition of priorities. And

especially in competitive markets, the effectiveness of anyactivity is watched with special interest. These aspectsrelate to the importance of the network components. Ofcourse, the importance should be defined as objective aspossible. As supply reliability in distribution is a technicalkey characteristic and as it can be calculated and quantifiedby a set of well-defined indices [4] the usage of probabilisticreliability calculation as presented in Chapter 5, it is afavorable solution to relate the importance to supply reli-ability. In EHV-grids, focus in importance evaluation ison system integrity, scheduling of power stations and theprevention of bottlenecks in transmission capacity. So,the reliability centered maintenance does not only considerthe condition of the system components, but also takes intoaccount the impact on the performance of the system[3,57].

RCM is not only evaluating the priorities for mainte-nance actions but is also a powerful tool for the rankingof replacement and refurbishment activities, because badequipment conditions lead immediately to the questionwhether it is more economical to do further maintenanceor replace the equipment. This analysis can be done forthe single equipment as well as for complete substationsin transmission where a combination of the conditions ofthe different parts (concrete and steel, primary and second-

ary equipment) is regarded [8,9].

Importance

CBMCondition Based

Maintenance

Not considered Considered

Continuous oroccasional monitoring

Maintenance whenrequired

CMCorrective Maintenance

No inspection ormaintenance untilbreakdown

TBMTime Based Maintenance

Fixed time intervals forinspections andmaintenance

RCMReliability Centered

Maintenance

Priority list

Connection of conditionand failure effect

Risk management

Considered

NotconsideredC

ondition

Fig. 2. Classification of maintenance strategies [2].

J. Schneider et al. / Electrical Power and Energy Systems 28 (2006) 643654 645


4/12

If the economical consequences (penalties, kW h notsupplied, change of merit order) of different actions aretaken also into consideration the RCM approach isextended to the so-called risk-based maintenance.

3. Determination of component condition

There are several basic possibilities on how to acquireinformation on the actual condition of network compo-nents. These possibilities differ significantly in the amountand type of information that is delivered, and of coursein the effort required to gather the information. It is com-mon to all methods that it is not possible to measure anindex directly expressing the remaining time to failure.So, the derived statement on the component condition issubject to certain approximations and statistical aspects,which imply a certain risk.

The best method from the technical point of view is theon-line monitoring of condition-related indices of the com-

ponents, e.g. coil temperatures or parameters describingthe oil quality of transformers, or the SF6-pressure in agas-insulated circuit breaker. The evident drawback of thismethod is that it requires a very high effort, both relating tofinancial aspects as well as relating to the handling of thegathered information. Due to their high costs most of thepower or generator transformers are equipped with moni-toring systems. Nowadays also more and more new substa-tions (mainly GIS) are equipped with condition monitoringdevices based on modern digital technologies like sensorsand actuators in the drives.

Another possibility is to perform off-line measurements

delivering statements on condition-related aspects, e.g.partial-discharge-measurements on cables in order to deter-mine the quality of the insulation, or special measurementsin order to determine the loss of contact material of circuitbreaker contact elements. Such measurements also requirea high effort, additionally it is required that the compo-nents under consideration are taken out of service for theactual measurement. Comprehensive and practical exam-ples on applications for power and generator transformersare given in Chapter 6.

As these two methods are comparatively expensive,more or less broad application is practicable only inEHV- or HV-networks. Here, the limited number of com-ponents compared to MV-networks the high value ofan individual component and the possible severe conse-quences of system failures justify the high effort [1012].

One way to reduce the effort in condition evaluation isthe usage of operational stresses, like e.g. voltages, currentsand power flows, as these values are measured on severalcomponents throughout the system. In order to link theseparameters to maintenance specific aspects, physical mod-els on the impact of operational stresses and ageing arerequired. For example, the life-time-usage of a transformercan be modeled using the actual power flows, and for a cir-cuit breaker it may be oriented at the sum of the actually

switched-off short-circuit currents I00

k. Of course, the mean-

ing and the accuracy of these derived maintenance-relatedindices strongly depend on the physical models available.At present, such models typically are very rough, if avail-able at all. One reason for this weakness is, that the modelsstrongly depend on the construction details of the varioustypes of equipment.

Finally, statistical methods can also provide informationon the condition of system components. The basis of thisapproach is the systematic collection of any data relatedto component condition, e.g. maintenance or failurereports. The information is categorized and assigned to cer-tain component types [13]. The level of detail in the catego-rization and in the definition of component types is acompromise between the resulting level of detail of theresults on the one side and the availability of data andeffort for the data collection on the other side. Also, thequality of results provided by statistical methods improveswith the number of incidents taken into consideration, sothere should be a certain number of devices in each compo-

nent type. Due to the high number of components installedin distribution networks, the statistical method is the onlysolution practicable for a widespread application in thesenetworks.

Regarding single equipments in MV-networks assump-tions of the equipment condition can also be derived frominspection and maintenance protocols. Very important isalso the operational experience of the utility personnel.There are methods available [3,5] based on only very littleinformation to estimate the condition of accessible equip-ment in MV networks.

Nationwide fault statistics are mandatory and do exist

in a lot of countries to survey the component availabilityand the reliability of the electrical grid [14]. What couldbe found is data based on the maintenance practices ofthe past and very often not detailed enough to analysethe real causes of equipment failures. Chapter 5 explainsin detail a comprehensive approach to overcome theseweaknesses in a German research project.

4. Asset simulation

4.1. Basic theory

Asset simulation is an approach to predict the long-termmonetary consequences of maintenance and renewal strat-egies for electrical grids [15]. It can me modelled by the useof system dynamics [16,17]. The system dynamics approachhas also shown its strength in other fields, where cause andeffect can be described well while the availability of data israther poor [1820]. It is a modelling approach that enablesbuilding of a formal representation of a dynamic behaviourof a business system, where the behaviour of the system is adirect result of casual relationships between different ele-ments of the system. The effects of causal relationshipsare based on assumptions and decision rules, which arethen formalized by using mathematical equations. Funda-

mental to system dynamics is the idea that all dynamic



5/12

behaviour is a consequence of the structure of the system,where the structure refers to how the elements of the systemare put together. Unlike the linear flow of spreadsheetmodels, system dynamics focuses on interrelationshipsrather than linear cause and effect and sees change as a pro-cess rather than a series of snapshots. It operates with feed-

back loops. The system is made up of these feedbackstructures, which reflect the actions of one factor on theother. This allows the user to model highly complex sys-tems with relative ease [1,21].

4.2. Simulation approach

Since the grid is basically a number of ageing assetswhich are being inspected, maintained, refurbished andrenewed, the model has to describe the asset ageing process(Fig. 3). Additionally the strategies of doing something tothe asset have to be described as a set of rules like if the

asset is >x years old and not renewed, then y will happen[22,23]. Also the costs and resources needed for theseactions have to be calculated. Last but not least, the agedistribution and a failure distribution of the assets areneeded.

4.3. Data and models

The most crucial part of the simulation is the assetmodel. This model describes the ageing of the assets andthe actions which can be done to the assets to prevent themfrom reliability degradation. A deep discussion on probabi-

listic and deterministic ageing models can be found in [23].Fig. 4 shows a very simple ageing model which can be usedto get first suitable results.

This model describes an asset with three different states:reliable, degenerated and unpredictable. For each of thestates different calculations for e.g. strategic maintenance,renewal decisions and failure rates are made. The principleis that, during its life time, an asset will pass through eachstate and spend a certain amount of time in each state.Work done by maintenance and refurbishment will slowdown the rate at which an asset moves from one state tothe next. Eventually an asset moves out of the system by

being replaced. In each state an asset will have a differentlevel of performance.

Such models are being implemented in dynamic system

simulation methods. Tools implementing such methodsare available in the market (e.g. Powersim, Anylogic,iThink, SD-Library). A common visualization of the modelin those tools is shown in Fig. 5. The states an equipment isin like reliable, degenerated and unpredictable are mod-elled either as conveyers or stocks (conveyers preserve afirst in first out order, stocks do not). The transfers betweenthese conveyers or stocks are modelled as flows which arecontrolled by the valves (e.g. ageing, replacement after fail-ure, . . .). As shown in the example conveyers know two dif-ferent kinds of outflows. One of them represents the timean asset needs to go through the conveyer (used for age-

ing). The other one represents a leakage which occurs inde-pendent from the asset age (used e.g. to model replacementafter failure or refurbishment during the degeneratedstate).

In the example an asset starts its life at the beginning ofthe reliable conveyer pass through the state degeneratedand will end up in the stock of unpredictable assets. Theoutflows of the conveyers and stocks are parameterisedby use of statistical data for asset life time and failurestatistics.

Output:CAPEX / OPEX

reliabilityriskSystem model

Asset

models forequipment and

measures

Input:planned grid changes

renewal - strategymaintenance - strategy

Interpretation of the resultsand feedback from engineers

Asset Segment

Fig. 3. The link between strategy and costs.

Reliability

Initial operation

100 %

0 %

0

schematic

reliable

degeneratedunpredictable

refurbishment

Remaining life

replacement after failure

repairafter failure

itial operation

100 %

0 %

0

schematic

reliable

degeneratedunpredictable

refurbishmentreplacement after failure

repairafter failure

replacement

Fig. 4. Simple ageing model for asset simulation.

Fig. 5. Simple ageing model for asset simulation.



6/12

4.4. Uncertainties due to the stochastic process

Since the flows in the model which represent the age-ing process are parameterised by use of statistical data theapproach is suitable for modelling the system behaviourbut inaccurate with respect to individual equipments. Also

only those effects are parameterised with highly accuratefigures which rely on incidents or actions which happenquite often. Parameters for incidents which are very rarewill be of poor quality and results of simulations have tobe verified to see if they are more or less independent fromthose parameters or not. In practise this theoretical prob-lem should not influence the quality of the results muchsince the actual processes are dominated by the more fre-quent procedures.

4.5. Inaccuracies due to model and data deficiencies

Obviously the approach can only be as good as the

underlying models are. The task here is to find a balancebetween theoretical model accuracy and data availabilityto parameterise the model. Especially the knowledge ofthe state where the assets are in the asset chain of the modelmight not be available, if no maintenance strategies whichrely on the knowledge of the condition of the assets are cur-rently used in the utility. Using the age of the assets insteadmight lead to the problem, that the asset age databasesonly provide the time when there were financial invest-ments but not the information what exactly was replaced.Additionally e.g. failure rates have to be provided for allthe states within the asset model. So here it is also necessary

to verify if an additional state which might be necessary toexplain a special effect can really be parameterised with theavailable data. But since the decisions for e.g. renewal arebeing taken in the utilities in the processes today, for a startit is definitely good enough to find out how those decisionsare taken today and then to model them exactly like that.Doing this the model will reproduce the decisions like theywould be taken in practice and that is what asset simula-tion is all about.

4.6. Sensitivity analysis

For scenario analysis and assessment detailed and sen-sitive data regarding the current asset portfolio are neces-sary. In addition to asset data specific informationregarding expenditures, quality of supply, risk, etc. haveto be provided. The availability of such data has alreadybeen discussed in Chapter 3. Gaps regarding data avail-ability and quality have to be systematically identifiedand closed by specific measures. Sensitivity analysis onthose specific parameters by use of the model can helpto identify where improved data quality has a greatimpact on the results and so efforts can be focused onimportant data.

Provided the ageing models and the model parameters

are accurate enough, the results can be used to compare

different asset management strategies. For example, theanalysis of different scenarios for asset investment planningprovides a transparent and reproducible basis for the qual-ity of a strategy and the costs resulting from it. It also givesinsight in the way the long-term costs of a network can beinfluenced by strategic investments in the grid. The

approach also has the capability to identify and explorethe risks associated with different strategies and how theserisks can be minimized.

5. Asset management in distribution networks

5.1. Introduction

As discussed in Chapter 3, the high number of compo-nents in distribution networks gives advantages to statisti-cal asset management approaches. Not the individualcomponents, but component classes are considered in the

corresponding methods. The level of detail in the definitionof component classes depends on different aspects, a majoraspect being the availability of suitable data.

Different statistical approaches vary in their level ofcomplexity and the focus of the analysis. The range is frompractice-proven methods aiming at preventing the deterio-ration of components beyond certain levels to comprehen-sive approaches considering failure and ageing models ofthe components, which are currently developed by manydifferent institutions worldwide. The following sectionsgive short examples on asset management processes cur-rently in operation or under development, respectively, in

Germany.

5.2. Statistical fault analysis

A practice-proven statistical asset management methodis the so-called surveyed break-down strategy [24]. Itconsists of a comprehensive fault and damage analysiswhich is realised by the use of a fault and damage databasein conjunction with SCADA, ERP-software and a geo-graphical information system (GIS). Fig. 6 schematicallyshows the surveillance of disconnectors.

In this example the number of failures of all disconnec-tors is compared with the number of switching cycles andthe number of all disconnectors. Thus, the availability fora given time period in the near past can be calculated. Itis compared against threshold Z, in this example itshould be better than, e.g., 95% which is the probabilityfor a successful disconnector operation.

Furthermore, the maintenance effort is calculated andcompared with a second parameter K which in thisexample should be below 60% of the initial maintenancecosts (based on time-based overhaul).

If one of the limits is exceeded, a trigger informs theasset manager about a worsening of the equipment condi-tion. The detailed fault and maintenance histories enable

the asset manager now to detect if the failures are caused



7/12

by material or other problems or are due to a lack of main-tenance. This can lead to a variation of the maintenancestrategy as well as to selective maintenance actions for anindividual equipment type.

5.3. Comprehensive statistical asset management approach

A comprehensive asset management approach of coursehas to consider the life-cycle costs of the equipment and ofthe system in total. But it should also consider the qualityof supply that is delivered by the system, as the depen-

dency between costs and quality is obvious and of highrelevance in recent discussions. In the end, asset manage-ment has to support the delicate balancing of costs versusquality according to the given requirements andregulations.

The principle of a comprehensive, risk-based asset man-agement approach for distribution networks using a statis-tical description of the network equipment is shown in theschematic diagram in Fig. 7 [25].

An important part of this approach is the calculation ofrelevant costs which is typically much easier than thequantification of supply quality. Here, deterministic andstochastic costs are differentiated.

Deterministic costs are determined by the chosen strate-gies for inspections, servicing, reinvestment and fault elim-ination. Also these deterministic costs cannot be predictedexactly, but compared to the cost aspects described as sto-chastic costs, the uncertainties are on a far lower level.Other fixed costs, e.g. capital costs, are included in thisgroup, too.

Stochastic costs are depending on the occurrences ofdamage and disturbance events on the components in thenetwork which are governed by chance. Thus, these costs including costs for fault clearance, costs for repair orreplacement of damaged equipment, and costs for penalties

or compensations (if applicable) can only be calculated

stochastically. Typically, these costs are subject to distribu-tions with large dispersions.

While the assessment of deterministic costs is a straight-forward task based on the chosen strategies for mainte-nance, reinvestment and fault elimination, the stochasticcosts are based on the results of a probabilistic reliabilitycalculation of the network. This calculation requires nextto usual network and component data a description ofthe failure occurrences of the components. This descriptionis given by so-called component reliability data.

The calculation methods required for this asset manage-

ment approach are all readily available today including

Number of

failures

Number of

switch. cycles

Number of

Disconnectors

Availability

2 years

3 years

4 years

5 years

6 years

Maintenance

costs

Loss due to

failure

Maintenanceeffort

Inspection intervals

Parameter

Z > 95 %

Parameter

K < 60 %

Initial costs

DatabaseGIS ERP System

DatabaseSCADA

Fig. 6. Surveyed break-down strategy.

Component -

reliability data

Network andcomponent data

e.g. component age,maintenace history

CalculationReliabil ity calculation.

Risk analysis

Stochastic costs

Fault eliminationRepair / replacement

Penalties / compensations

Deterministic costs

Servicing

Reinvestment

Other fixed costs

Analysis

and

optimisation

Strategies

Servicing

Reinvestment

Fault elimination

PrognosisPrognosis

Inspections

Inspections

Fig. 7. Principle of a comprehensive asset management approach [25].



8/12

probabilistic reliability calculations [26]. However, certainaspects of this approach typically cause problems in practi-cal application:

Meaningful statistics about the distribution of the costscaused by damages of network components, e.g. repair

costs, are not available. Up to now, only rough esti-mated values can be assumed. Component reliability data can be determined for net-

works in their current state by suitable statistics. Butchanges in the asset management strategies e.g.reduced maintenance, reduced reinvestment leading toincreased component age, or staff reductions increasingthe duration for fault elimination of course will haveeffects on component reliability. As the parameters ofthese asset management strategies are to be optimizedin this process, such effects have to be considered appro-priately by models for the prognosis of componentreliability.

The level of detail of the implemented models for theprognosis of the component reliability is a key characteris-tic of the different methods following this basic approach.Especially concerning XLPE cables affected by water treeproblems, sophisticated models and suitable input datawere developed already (e.g. [9,27]).

However, for the prognosis of component reliability independency of these parameters which is a crucial stepfor asset management methods typically only very rudi-mental models and scarce data are available.

As an example, one research project trying to contribute

to this field is presented in the next section.

5.4. Research project on asset management in distribution

systems

This knowledge gap in the modelling of component reli-ability in dependency of characteristic influence factors hasbeen the motivation to initiate a research project carriedout by more than 20 network operators, academic institu-tions and service providers in Germany. The project issponsored by the German Federation of Industrial Coop-erative Research Associations Otto von Guericke(Arbeitsgemeinschaft industrieller Forschungsvereinigun-gen AiF) with funds of the Ministry of Economics andLabour (Bundesministerium fur Wirtschaft und ArbeitBMWA) as well as by the German Research Foundation(Deutsche Forschungsgemeinschaft DFG).

The main goal of this project is to collect information oncomponent damages and outages in a specially designed sta-tistic. This database then allows not only the determinationof component reliability indices, but in particular the anal-ysis of the influences of component age and maintenancehistory. These results allow the definition and even moreimportant for practical application also the parameteriza-tion of suitable models to predict component reliability

depending on the chosen asset management strategies.

Additionally, also the costs that are caused by damageoccurrences on components are not available in systematicstatistics up to now. The specially designed statistic in thisresearch project also includes information on such costs, sothat damage costs and their statistical dispersion can bedelivered.

As an example, Fig. 8 shows the age-related outage ratefor low-oil-content circuit breakers in medium voltage sub-stations. Note that this diagram is based on a preliminarydata set, as the data collection was not completed yet. Inthe diagram, a very clear dependency of the outage rateon the component age can be seen.

6. Asset management in transmission systems example:

transformer life management

6.1. Objectives of life management processes

As explained in the precedent chapter distribution assetmanagement can be approached from a statistical or stoch-astical point of view. In transmission grids, where the singleequipment is much more expensive, an individual observa-tion is required.

In the following life management for transformers willbe discussed as an example. In principle, similar modelscan be developed also for other equipment types. Forexample, for HV circuit breakers, several failure modesfor subsystems of the circuit breaker were evaluated in aresearch project [28].

A reasonable asset management of transformers has toinclude life management ensuring long-term utilisationand exploitation of the assets. Regarding the high reinvest-ment costs, the long manufacturing time and the capacitiesof the transformer manufacturers a long-term planning isnecessary.

The objective of the life management process is the

optimal utilisation of the remaining life time regarding a

0 5 10 15 20 25 30 35 a 450

0.2

0.4

1/a

0.8

Age

hS hS= 0.3816

Fig. 8. Medium voltage stations, low-oil-content circuit breaker, outagerate fO depending on component age.



9/12

definite reliability and a constant distribution of costs forreinvestment and maintenance.

6.2. Ageing behaviour of transformers

When discussing the ageing behaviour of transformers,

it is useful to subdivide the transformer in its main compo-nents. In this contribution only the ageing behaviour of theactive part, the components bushings and the tap changerwill be described. Presuming a good maintenance the age-ing of the other components like vessel, cooling system,etc. can be neglected, because the ageing of these parts isnot dominant or they can be exchanged or repaired inthe substation if necessary.

6.2.1. Active part

The ageing of the active part mainly depends on theloading of the transformer. The thermal stress reduces

the mechanical strength of the cellulose insulation, a weakcellulose insulation can cause an internal fault due to tran-sient current stress. The strength of the cellulose insulationcan be determined with DP measurement (degree of de-polymerisation). A new cellulose insulation has DP >1000, cellulose insulation with insufficient mechanicalstrength has DP < 200. Unfortunately DP measurementcan only be performed during a workshop investigationor when the transformer is scrapped. The design of theinsulation system, the cooling method and the specification(power) of the cooling system affect the ageing gradient.The distribution of measured DP values in Fig. 9 demon-strates that the technical life time of generator transformers

(GSU) is limited to 25 years whereas transmission trans-formers can reach a technical life time up to 50 years.

The technical life time of insulating oil is influenced bythe loading condition of the transformer and the ageingstability. Using high quality mineral oil with extended age-ing stability the technical life time of the insulating oil issimilar to technical lifetime of the active part.

6.2.2. Condenser bushings

The condition of condenser bushings can be determinedby measurement of capacitance and dissipation factor tan d.An evaluation of the measurement results shows that ageing

is only observed at resin bonded paper bushings (Fig. 10).The ageing of these bushings is caused by cracking of theresin bonded paper and inhomogeneous impregnation withinsulating oil. The technical life time has a great scatter andshould not exceed more than 30 years.

6.2.3. On load tap changer

The ageing of the on load tap changer (OLTC) dependson the number of switching operations and the cumulatedswitching current. Expecting a good maintenance the tech-nical life time of an OLTC is nearly unlimited, because thecondition of OLTC is determined during a diverter switchinspection and wearing parts such as switching contactsand insulating oil are exchanged if necessary. Howeverthere is a risk of oil cooking if the tap selector contactsare not silver-plated.

6.3. Diagnosis strategy

A reasonable life management has to be supported by astepwise diagnosis strategy (Fig. 11). There are some rou-tine tests for condition assessment of the active part, theinsulating oil, the bushings and the tap changer. In caseof an indication or for extended condition assessment sev-eral additional diagnosis methods are used. During repairin a workshop or when the transformer is scrapped a visualinspection is done and paper samples from the windings aretaken to perform DP measurements.

The dissolved gas analysis (DGA) is the most importantdiagnosis method to assess the condition of the active part

0

100

200

300

400

500

600

700

800

900

1000

0 10 20 30 40 50

Age/years

DP

transmission transformers

GSU transformers

Fig. 9. Ageing behaviour of cellulose insulation of power transformers.

0.0

0.5

1.0

1.5

2.0

0 5 10 15 20 25 30 35 40

Age/years

Dissipationfactor[%]

Fig. 10. Ageing behaviour of resin bonded paper bushings.



10/12

of a transformer. For an optimal interpretation of DGAthe consideration of gas development rate and the compar-ison of the measured gas values with normal values is nec-essary [29]. By an additional expert system the DGA istransformed into a three stage warning level (normal, firstwarning, urgent warning). The expert system also providesa condition based oil sampling.

The measurement of furanic components in the insulat-ing oil is used to get information about the degradation ofthe cellulose insulation [30]. The interpretation of measure-ment results depends on oil treatment and cooling method.

The capacitance and tan d measurement of bushings andthe diverter switch inspection are fundamental to ensurethe integrity of bushings and tap changer. A failure of abushing can cause a collateral damage, the break-downof the diverter switch can destroy the regulating windingof the transformer.

Visual inspection and analysis of paper samples are theonly tools to get a feedback of the true condition of thetransformer compared to the results of condition assess-ment gained by the different diagnosis methods. The anal-ysis of the paper samples shows that most of the scrappedgenerator transformers had a cellulose insulation withinsufficient mechanical strength. The visual inspection con-firms that the failures of some transformers are caused byinsufficient mechanical strength of the cellulose insulation.The typical break-down mechanism starts with aninterturn fault in the low-voltage winding near the locationof the hotspot. The interturn fault causes a fault currentthat deforms the low-voltage winding. From the deformedwinding arcing causes tripping of the Buchholz relay.

6.4. Life-time extension methods

Life-time extension procedures without any conditionassessment cannot technically be justified. In case of an

indication a refurbishment in the substation may be

arranged as a first step in order to avoid cost for transpor-tation and to minimise the outage time. The following pro-cedures are established:

oil treatment/oil exchange replacement of bushings exchange of tap changer contacts drying of active part

On-site refurbishment procedures need an exact inspec-tion. The fault has to be identified and located. Further-

more the fault location must be accessible. At least thesuccess of the refurbishment procedure has to be proved[31]. Before carrying out refurbishment procedures chancesand risks have to be judged. In case of any doubt a work-shop refurbishment or repair has to be preferred.

6.5. Strategic actions

A continuous assessment and evaluation of transformercondition is the most important action for life manage-ment. A special assessment scheme and ranking tool basedon condition and importance can be implemented. Bymeans of the ranking tool forecasts of the replacement oftransformers can be done [32].

Another strategic aspect is the availability of a sufficientamount of spare transformers and bushings. Sparetransformers and bushings minimise the consequences ofa failure and can reduce the outage time considerably.Especially spare generator transformers are very valuable.However the optimal use of spare transformers and bush-ings requires a wide standardisation. Moreover, condi-tion-based diagnostic and maintenance can minimise therisk of a sudden failure of a transformer.

In case of failure, detected during refurbishment proce-dures or when a transformer is scrapped, a consequent

inspection is necessary. The inspection of a transformer is

component diagnosis method online offline offsite remark

active part oil ageing X routine

DGA X routine

resistance measurement X after indication (DGA)

PD measurement X after indication (DGA)

imdepance measurement XX

after short circuit

FRA after short circuit

PDC measurement X condition assessment

furfurol measurement X condition assessment

DP measurement X condition assessment

bushing capacitance / tan measurement XX

routine

DGA condition assessment

PD measurement X condition assessment

tap changer diverter switch inspection routine

torque measurement X

X

condition assessment

Fig. 11. Stepwise diagnosis strategy.



11/12

very valuable to collect experience and can show hiddenweak points of a transformer design and construction. Fur-thermore, it is the only method to verify the empiricalassumptions of the condition assessment scheme.

At least, a main issue of transformer life management isto elaborate a renovation strategy and corresponding pro-

grams for replacement as far as necessary. Furthermore,the final demand analysis has to include the future demandof transformers due to further network development aswell as possible unused transformers which have becomeavailable due to reconstruction measures.

7. Outlook

The actual challenge of asset managers is very often noton the methodical side but lies on the IT side to support assetmanagement decisions. Commercial data can be found inERP systems, but very often it is just subdivided in a few

groups, e.g. voltage levels and not related to the single equip-ment or substation. Also internal costs for operation andmaintenance are collected and are not related to the equip-ments in the system. Geographical data can be found inGIS system, whereas maintenance-related data and reportsare often stored in separate technical data bases. There areadditional tools for the evaluation of maintenance strate-gies, network calculation, on-line condition data analysis,etc. There is definitely a need for integrated IT systemsand decision tools and a lot of utilities are working on theirIT landscape mostly in combination with work managementsystems to execute the task of the asset management.

Besides the IT support a second challenge is the need forbetter knowledge on equipment deterioration and the corre-sponding models, especially in distribution systems. Ideally,these models should provide estimates on equipment failureprobabilities based on input data of equipment age, opera-tional stresses and maintenance activities in the past. Thesemodels are needed urgently in order to make the results ofasset simulation approaches reliable. The shown Germanresearch project is an example for a first step in this direc-tion, however extensive further research is necessary.

Very often asset mangers face the question of budgets incorrelation with supply quality. To answer this and to get abetter feeling about reserves in the networks new tools like

asset simulation and RCM for the prioritisation ofmeasures and the definition of budgets are an importantstep. This is mandatory in the utilities to get the neededresources and budgets for the networks in a competitiveenvironment.

References

[1] Wenzler Ivo. Development of an asset management strategy for anetwork utility company: lessons from a dynamic business simulationapproach. Simulat Gaming 2005;36(1):7590.

[2] Balzer G, Brandl M, Schmitt O, Strnad A. ETG-Tage 1999,Rechnergestutzte Instandhaltungsplanung von Betriebsmitteln der

Elektrischen Energieversorgung. p. 41120.

[3] Vetter F, Muhle J. Das anlagenbezogene Risikomanagement in derInstandhaltungsplanung der E.DIS AG. EVU Betriebspraxis 10/2002, p. 24 ff.

[4] International Electrotechnical Commission, IEC 60050-191 Inter-national electrotechnical vocabulary. Chapter 191: Dependability andquality of service, Geneva, 1990.

[5] Schmitt O, Schneider A. Moderne Werkzeuge zum integriertenAsset- und Instandhaltungsmanagement. International ETG- Con-gress Energietechnik fur die Zukunft, 2324 October 2001,Nurnberg.

[6] Balzer G, Schmitt O, Schneider A. The reliability centered mainte-nance strategy leads to an optimized asset management CIRED 2002,paper no. 33.

[7] Hografer J. Von der Erfassung zur Entscheidung: Einsatz vonExpertensystemen zur Bewertung der Ergebnisse aus der Betriebsmit-teldiagnostik. ETG-Fachbericht 87, p. 15760.

[8] Breitling Lina, Allan Ron, Eriksson Roland. A reliability-centredasset maintenance method for assessing the impact of mainte-nance in power distribution systems, IEEE Transactions papersubmitted 030511 revised 031210, 040324 and 040629, accepted040716.

[9] Bertling Lina. Reliability centred maintenance for electric powerdistribution systems, Doctoral Dissertation, Stockholm 2002.

[10] Neumann C, Sundermann U. Assetmanagementim Hochspannungs-netz Transformatoren und Schaltgerate. FGH Symposium Vers-orgungs- & Anlagenqualitat, Heidelberg, September/October 2004.

[11] Balzer G, Drescher D, Heil F, Kirchesch P, Meister R, Neumann C.Selection of maintenance strategy by analysis of service experience;Cigre_Symposium: Cigre SC A3/B3 Joint Colloquium, Tokyo 2005,report 113.

[12] McIlroy C, Cunningham M. Transfix ein neues Online DGA-System zur Transformatoruberwachung, ew 103/26, 2004, p.305.

[13] Stotzel M, Wellssow WH, Zdrallek M. Reliability calculation of MV-distribution networks with regard to ageing in XLPE-insulatedcables. IEE Trans Generat Transmission Distribut 2001;148(6):597602.

[14] Obergu

nner M, Schwan M, Krane C, Pietsch K, von Sengbusch K,Bock C, et al. Using the VDN statistic on incidents to derivecomponent reliability data for probabilistic reliability analyses.Elektrizitatswirtschaft 2004;15(Juli).

[15] Nilges J, Gaul AJ, Spitzer H. Strategische Investitionsplanung Praxisnahe Wege fur eine nachhaltige Bewirtschaftung der Assets.Energiewirtschaftliche Tagesfragen et 54. Jg. Heft 2004;10:6556(German language).

[16] Forrester JW. Industrial dynamics. Cambridge, MA, 1961.[17] Thurlby R, MacQueen C. Managing the asset time bomb. In:

Proceedings of the DADSM Conference in Spain 1999.[18] Vannan Mani, Glass-Husain Will. Managing gas assets through

simulation and scenario planning; international system dynamicsconference Wellington, New Zealand, 1999.

[19] Kenward Michael. System dynamics: a simulating idea widens

managers horizons, Financial Times, 7. 1.1998.[20] Lander Sandra. Strategische Planung von Kreislaufwirtschaftsyst-emen, Strategic planning of recycling management systems, Disser-tation TU Berlin 2004.

[21] Moeller A. Continuous simulation in material flow networks, In:International Environmental Modelling and Software Society iEMSs2004 International Conference.

[22] Sturmer J. Instandhaltungs- und Erneuerungs-strategien in Verteilei-lungsnetzen. Aachen, Germany: Shaker Verlag; 2002 (Germanlanguage).

[23] Endrenyi J. IMPACT OF MAINTENANCE STRATEGY ONRELIABILITY, Final report by the IEEE/PES task force on impactof maintenance strategy on reliability of the reliability, risk andprobability applications subcommittee, July 1999.

[24] Giese H-P, Hografer J. Einfuhrung des automatisierten Netzbetriebs

bei der enviaM. etz 2002, vol. 2324, p. 14 ff.



12/12

[25] Schwn M, Weck K-H, Roth M. Assessing the impact of maintenancestrategies on supply reliability in asset management methods. 40thCIGRE session, paper C1-108, August 2004.

[26] Singh C, Schwan M, Wellssow WH. Reliability in liberalizedelectric power markets- from analysis to risk management -surveypaper. In: 14th Power system computation conference PSCC, 2002,Sevilla.

[27] Bohme M, Krage I, Weck K-H. Instandhaltung von Mit-telspannungskabelanlagen. ETG-Fachtagung Verteilungsnetze im lib-eralisierten Markt (ETG-Fachbericht 82), 2000, p. 15767.

[28] FGH: Zustandsmodell fur die Instandhaltungsplanung fur einen123-kV-Leistungsschalter. Technischer Bericht 295, FGH e.V.,2001.

[29] Mollmann A, DePablo A. New Guidelines for furans analysis as wellas dissolved gas analysis in oil-filled transformers, CIGRE 15/21/33-19,1996.

[30] Mollmann A, Pahlavanpour B. New guidelines for interpretation ofdissolved gas analysis in oil-filled transformers, Electra No. 186,October 1999.

[31] Neumann C, Feser K, Henningsen C-G, Pein H, Rohsler H, Weck K-H. The German practice of on-site tests and measurements on GIS,transformers and cables at commissioning and in service. CIGRE 33-305, 2002, Paris.

[32] Petterson L, Fantana N, Sundermann U. Life assessment: ranking ofpower transformers using condition based evaluation. A newapproach; CIGRE 12-204, 1998, Paris.


Documents

Asset Management Techniques