Measuring lean implementationfor maintenance service companies

Stephan J. de JongDepartment of Aerospace Engineering, Delft University of Technology,

Delft, The Netherlands, and

Wouter W.A. Beelaerts van BloklandDepartment of Transport Engineering and Logistics,

Delft University of Technology, Delft, The Netherlands

AbstractPurpose – Implementation of lean manufacturing is currently performed in the production industry;however, for the airline maintenance service industry, it is still in its infancy. Indicators such as work inprocess, cycle time, on-time performance and inventory are useful indicators to measure leanimplementation; however, a financial economic perspective taking fixed assets into consideration is stillmissing. Hence, the purpose of this paper is to propose a method to measure lean implementation from a fixedasset perspective for this type of industry. With the indicators, continuous improvement scenarios can beexplored by value stream discrete event simulation.Design/methodology/approach – From literature, indicators regarding asset specificity to measurelean implementation are found. These indicators are analysed by a linear least square method to know ifvariables are interrelated to form a preliminary model. The indicators are tested by value stream-baseddiscrete event simulation regarding continuous improvement scenarios.Findings – With the new found lean transaction cost efficiency indicators, namely, turnover, gross marginand inventory pre-fixed asset (T/FA, GM/FA and I/FA, respectively), it is possible to measure operationperformance from an asset specificity perspective under the influence of lean implementation. Secondly, theresults of implementing continuous improvement scenarios are measured with the new indicators by adiscrete event simulation.Research limitations/implications – This research is limited to the airline maintenance, repair andoverhaul (MRO) service industry regarding component repair. Further research is necessary to test theindicators regarding other airline MRO service companies and other sectors of complex service industrieslike health care.Practical implications – The lean transaction cost efficiency model provides the capability for amaintenance service company to simulate the effects of process improvements on operation performance forservice-based companies prior to implementation.Social/implications – Simulation of a Greenfield process can involve employees with possible changes inprocesses. This approach supports the adoption of anticipated changes.Originality/value – The found indicators form a preliminary model, which contributesto the usage and linkage of theories on lean manufacturing and transaction cost theory – asset specificity.

Keywords Lean, Implementation, Simulation, Asset specificity, MRO aviation,Transaction cost

Paper type Research paper

IntroductionThe aviation industry is characterized by a highly dynamic and volatile businessenvironment (Doganis, 2002, 2001). Competition between airlines is intensifying and

The current issue and full text archive of this journal is available on Emerald Insight at:www.emeraldinsight.com/2040-4166.htm

Measuring leanimplementation

35

Received 8 December 2014Revised 1 June 2015

8 June 2015Accepted 29 June 2015

International Journal of Lean SixSigma

Vol. 7 No. 1, 2016pp. 35-61

© Emerald Group Publishing Limited2040-4166

DOI 10.1108/IJLSS-12-2014-0039

margins are decreasing (Schmidberger et al., 2008) under the influence of “low cost” airlinesand the rapid deployment of airlines in the Middle East. This development created anecessity to innovate and improve aircraft maintenance, repair and overhaul (MRO) serviceoperations by implementing principles of lean manufacturing (Womack and Jones, 2005;Murman et al., 2002).

The MRO service company under research (Chün, 2009) experienced leanimplementation and provides services to the airline industry such as Air France-KLM for therepair of auxiliary power units (APUs) and air cycle machines (ACMs). The company wasstarted in the year 2000 as a joint venture between Hamilton Sundstrand and KLM RoyalDutch Airlines. Due to changes in the MRO market, the company changed ownership. KLMbecame the full owner and appointed a new management in 2006 because the service levelmeasured by on-time performance (OTP) of the serviced products was 35 per cent and theaverage turnaround time (TAT) was 28 days, while 15 days was promised to the customer.To stay in competition, it was necessary to improve the OTP and TAT of the servicedproduct to increase the service level. Aircraft types involved are from Boeing and Airbus.The serviceable systems of these aircrafts are APUs and ACMs.

Preliminary research on the implementation of Lean Six Sigma (Beelaerts vanBlokland et al., 2008a) with the small and medium-sized enterprise (SME) under researchshowed an improvement of the TAT for pneumatic components first from 28 days to20.49, followed by a further improvement to 13.13 days. Given the significantimprovement on TAT of the serviced product, the management team of the companycontinued the “Lean Six Sigma” approach. A need for measuring lean implementationduring the sustainment phase was expressed.

MRO can be defined as, all actions which have the objective of retaining or restoring anitem during its life cycle in which it can perform its required function. The actions include thecombination of all technical and corresponding administrative (certification, regulation,registration) managerial and supervision actions (European Federation of NationalMaintenance Societies vzw, 2011). By this definition, the MRO industry can be characterizedby its high asset specificity due to airline–aircraft-related regulation and certification ofprocesses, assets and components for reasons of safety. The assets to be used for MROservices can only be used exclusively for this type of service and worthless in any other typeof industries (Arnold, 2000). Asset specificity belongs to the theory regarding transactioncost economics (TCE) of Williamson (1981) and introduces the financial, economic andtransaction perspective. Current research towards lean implementation in the MRO serviceindustry does not provide indicators covering this financial economic perspective. Byhaving these indicators, it would be possible to measure operation performance from anasset specificity perspective under the influence of lean implementation. The other aspectregarding TCE is transactions. Transactions are necessary elements for a value system togenerate value. As transactions take place randomly, these can be used to test performanceindicators by discrete event simulation.

This paper continues with a literature review to identify the indicators. A linear leastsquare methodology is used to know if the indicators can form a preliminary model.Data analysis is performed to measure operation performance through time under theinfluence of lean implementation. The final part of this paper is to test the indicators bya value stream mapping (VSM)-based discrete event simulation. By simulation ofcontinuous improvement scenarios, the company can prepare for strategic decisionsregarding the use of assets and resources.

IJLSS7,1

36

Literature reviewFrom the introduction, it surfaced that asset specificity characterizes MRO processeswhich need to be researched in more detail. The other aspect is lean implementation. Thecompany changed management and started the implementation in 2007. Literatureresearch is necessary to identify variables to measure lean implementation.

Transaction cost economics – asset specificityWilliamson (1979) proposed the theory on TCE by the publication “Transaction costeconomics: the governance of contractual relations” in the Journal of Law andEconomics. TCE lies at the intersection of three fields of research, namely, economics,law and organizational theory, and defined as “a transaction occurs when a good orservice is transferred across a technologically separable interface” (Williamson, 1981),with focus on external transactions. Williamson (1985) positioned asset specificity as themost critical dimension alongside opportunism and bounded rationality as the threemain principles of transaction cost theory TCE.

The TCE has been used in the aerospace MRO industry by Masten (1984). Mastenlooked at aerospace procurement, using a measure of the degree to which componentsused in production were adaptable for use by other firms. The research found supportfor the probability that a component procured internally increases with the complexityof the component being procured. Masten’s research was preceded by related studies ofMonteverde and Teece (1982a, 1982b), who referred to asset specificity and verticalintegration versus supplier switching cost in the car industry, indicating that there is arelation between complexity of the company and transaction costs.

Douglas (2000) stated that there is more to transaction costs than just putting a “T” in acost function, in line with earlier reviews on TCE by Shelanski and Klein (1995). Williamsonmentions three limitations of his work: its crude form, instrumentalism and incompleteness(Williamson, 1998). As mentioned, asset specificity is the most critical dimension of TCE.Goods and services with high asset specificity cannot be used in other transactions beyondthe field of expertise without huge additional costs (Arnold, 2000). Low asset specificitymeans that little information has to be exchanged with the transaction partner. Highspecificity is related to complex information for complex products, processes and servicesdominated by extensive regulations and certification by airline authorities for safetyreasons.

TCE needs to find actual indicators with predictive powers (Arora, 2007) andtransactions should be organized (Hardt, 2009) to assess these. This research paper followson this argument of addressing indicators related to company internal transactionsfocussing on asset specificity instead of external transactions. Like with externaltransactions, internal transactions are taking place as components are transferred internallyacross technologically separable interfaces in combination with complex processes andinformation between interfaces. Figure 1 shows the framework as by Canbäck (1998), whichis adapted to indicate how TCE will be used for this research with focus on internaltransactions within the company. This framework was further developed by Canbäck et al.(2006) for assessing company performance relative to size. The three main principles of TCEare opportunism, bounded rationality and asset specificity. Uncertainty and frequency,however, are still considered alongside the main dimensions. Furthermore, production costscomplement the transaction costs, as multiple sources state that transaction cost should notbe seen as a stand-alone: the total costs of a system need to be minimized. Bounded

37

Measuring leanimplementation

rationality and opportunism are not playing a role in this research. The motivation is thatbounded rationality refers to individual behaviour, for instance, by customers which isintendedly rational, but only limited (Williamson, 1985). In this market of aircraftmaintenance service, it is about long-term contracts based on fixed agreed pricing. As it is anoligopolistic market, the opportunity to switch to other service suppliers is limited; therefore,opportunism is considered to be a constant.

Asset specificity in MRO processes is present in site specificity, test cells, buildingand special tooling, which are all directly related to the products and the requirementson quality and safety by the airline industry. To provide for an empirical measurement,the book value of the company’s fixed assets is taken as a proxy for the asset specificity.The total fixed assets will be used further in this research and consist of:

• Buildings: Consists of the building itself and investments and alterations to it.• Machinery and equipment: Special tooling, test equipment, specific maintenance

equipment calibrated and certified for aerospace.• Component exchange pool: Assets of the company involved in an exchange pool of

spare parts necessary to reduce the risk of “aircraft on ground”, which is the casewhen an aircraft cannot be used due to unavailability of parts/components.

• Under construction: Any of the other four items being prepaid or actually underconstruction.

Furthermore, fixed assets are known as variable in relation with the financial structure ofthe firm. Hall et al. (2000) reported about the relation between long- and short-term debt, assetstructure, size, age of company and profitability for SMEs in the UK. The effect of growth onshort-term debt, however, was consistent across industries, while profitability had no effecton long-term borrowing in any industry. Jaggi et al. (2001) concluded that revaluations offixed assets are positively associated with the firms’ future operating performance,suggesting that the managers’ motivation for revaluation of fixed assets has been to signalfair value of assets to financial statements users. Jacobs and Swink (2011) modelled therelation between product portfolio complexity by the multiplicity, diversity andinterrelatedness of products within the portfolio and the impact on operational performance.The model explicitly addresses the roles of organizational learning and the character of fixedassets (utilization and flexibility). From this reasoning, organizational learning by means oflean implementation and continuous improvement is related with utilization and flexibility

Figure 1.TCE preliminaryframework, adaptedfrom Canbäck (1998)

IJLSS7,1

38

of the fixed assets. The fixed assets appear to be an important variable in relation tooperation performance and lean implementation.

Lean and Six SigmaThe lean philosophy (Womack et al., 1990; Womack and Jones, 2003, 2005) iswell-established in the car manufacturing industry. Murman et al. (2002) researched leanmanufacturing in the aerospace industry and described lean as follows: “Becoming leanis a process of eliminating waste with the goal of creating value”.

Six Sigma originates from the mid-1980, when the engineers in Motorola Inc. in the USAwere inspired by quality improvement methodologies like quality control and total qualitymanagement. George (2002) introduced the combination of lean and Six Sigma. The goals ofLean Six Sigma are to maximize performance by achieving the fastest rate of improvementin customer satisfaction, cost, quality, process speed and flexibility. VSM introduced byRother and Shook (2003) and Hines et al. (1998) allows for a clear visualization of designactivities’ characteristics and interdependencies, the in-between inventory or transport dataand the overall planning requirements and static throughput data. They introduced thetakt-time of the process to relate actual customer demand with capacity. According toCutcher-Gershenfeld (2004), the following effects of lean implementation can be expected:elimination of defects, reduction of production and development costs, reduction of cycletime and inventory levels, increased profit margin and improved customer satisfaction.

Lean implementation in the aviation service industryIn the airline industry, the so-called low-cost carriers like Ryanair and EasyJet arecompeting on TATs and punctuality or OTP to increase the operation performance ofthe aircraft. Research by Beelaerts van Blokland et al. (2008b) regarding turnaroundprocesses of aircraft at the airport showed how KLM could improve the TAT for theBoeing 737-800 with 42 per cent by lean implementation through reduction of waitingtimes in “above the wing” processes. The TAT of a KLM aircraft was about twice aslong compared to a Ryanair aircraft in 2007. A short TAT has a positive effect on theutilization rate of the aircraft as a fixed asset and as such turnover and profit. Thisexample in the service industry shows the relation between the TAT, OTP and fixedassets from a lean implementation perspective. In maintenance service, the cycle time(Cutcher-Gershenfeld, 2004) or TAT indicates the level of efficiency regarding the use offixed assets. The indicator OTP reflects the effective use of the assets.

From this research towards the maintenance service industry cycle times or TATs,OTP, the use of inventory and fixed assets appear to have a relation with operationperformance under the influence of lean implementation.

Lean implementation in the industryResearch towards lean implementation for SMEs in the UK was reported by Achangaet al. (2006). They identified leadership, management, finance, organizational cultureand skills and expertise as the most pertinent issues for the successful adoption of leanmanufacturing for SMEs. Scherrer-Rathje et al. (2008) researched the implementation oflean in the food processing industry with mixed results. One implementation waspositive and the other was negative. It seems that just implementing lean is not aguarantee for success. The bottom-up approach was not successful, as there was a lackof leadership and management commitment. After the commitment was established, thelean implementation became successful. Culture and discipline as devised by Mann

39

Measuring leanimplementation

(2010) are essential aspects on how to create a lean culture. To sustain leanimplementation, the entire company needs to be involved following the basic elements oflean thinking and to implement a process of continuous improvement to reduce defects,improve customer demand and better utilize the assets and resources.

Bhasin (2011) and Singh et al. (2010) reported about lean implementation forproduction companies; however, the cost-benefit or financial aspects are still missing.Lu et al. (2011) remarked on the difficulty to compare future value chain structures oneffectiveness, indicating it would be of interest to have methods and tools to test thesefuture value chain structures. Jeong and Phillips (2011) addressed the possibility forfurther research on the combination of VSM with discrete event simulation to predicteffectiveness of the improvement. Singh et al. (2012) referred to VSM methodology forlean implementation in a production environment. Significant performanceimprovements were observed. It was found that reduction in lead time was 83.14 percent, reduction in processing time was 12.62 per cent, reduction in work-in-processinventory was 89.47 per cent and reduction in manpower requirement was 30 per cent.The rise in productivity per operator was 42.86 per cent.

Further research was advised to incorporate a cost-benefit analysis, as thefinancial perspective was still missing. Research by Beelaerts van Blokland et al.(2012) presented indicators to measure value leverage and stability of aerospacecompanies. The variables turnover (T), R&D and profit with the employee numbersas denominator were found to be related to form a model expressing the leverage andstability of value flow of a company. The variables T/E, RD/E and P/E form thevalue leverage model.

To summarize, researchers advised future research towards the cost-benefit orfinancial aspects of lean implementation. Lean implementation as an organizationaland cultural aspect was linked with performance regarding fixed assets. It seemslean implementation may have an impact on the utilization of fixed assets.Similarities are found between aircraft turn-around and preliminary researchtowards maintenance service processes (Beelaerts van Blokland et al. 2008a). Forairlines, the link between operation performance and fixed assets is clearlydemonstrated by performance regarding ATAT and OTP of aircraft. For both typesof services, the importance of TAT and OTP can be recognized.

This research will explore the combination of VSM and discrete event simulation, asvarious researchers mentioned it would be helpful to have these methods and tools fordesign of processes from a lean perspective.

From literature, the main research question is formulated as follows: how to measureoperation performance from an asset specificity perspective under the influence of leanimplementation regarding cycle time and OTP for maintenance service companies in theaviation industry? The following sub questions are proposed:

Q1. Can the year of lean implementation regarding TAT and OTP be identified?

Q2. What indicators capture lean implementation for maintenance servicecompanies from an asset specificity perspective and can these indicators form amodel?

Q3. What are the operation performance boundaries under the influence of furthercontinuous improvement scenarios?

IJLSS7,1

40

Preliminary frameworkFrom literature, the following indicators are found with the goal to measure operationperformance from an asset specificity perspective under the influence of leanimplementation regarding cycle times and OTP for maintenance service providers:

• Fixed assets (FA): The TCE – asset specificity perspective (Williamson, 1985)introduces FA as a variable which can be used for the denominator of thevariables.

• The gross margin per fixed asset (GM/FA): This variable measures the benefit inrelation to the FA resulting in the indicator; GM/FA. Jaggi et al. (2001) and Jacobsand Swink (2011) referred to the relation between operation performance,utilization, flexibility, organizational learning and FA. Sigh et al. (2012) referred tothe cost benefit in the context of lean implementation.

• Inventory per fixed asset (I/FA): Inventory is crucial in maintenance serviceprocesses to enable servicing of equipment and relevant to the principles of leanmanufacturing, according Womack and Jones (2003); George (2002) andCutcher-Gershenfeld (2004). Inventory is an essential value driver for themaintenance process and related to the FA. From this reasoning, the indicatorI/FA measures the variable inventory in relation to the FA.

• Turnover per fixed asset (T/FA): Continuous flow suggests the flow of productsthrough the system. To measure the flow of value through the system, theturnover (T) measures the flow of products through the processes from anoperation performance and financial economic perspective (Beelaerts vanBlokland et al. (2012). The turnover per fixed asset measures the flow in relation tothe FA (T/FA).

The variables T/FA, GM/FA and I/FA will be used for analysis on lean implementation.These new indicators complement the indicators turnaround time (ATAT) and OTP.

MethodThe method comprises the data sources, statistics, VSM, discrete event simulation andthe levels of detail based on the organization structure in the company.

Data sourcesThe analysis is based on data available from annual reports and data from thecompanies’ ERP database system. The data consist of the shipped APUs, the shippedline replaceable units and the shipped pneumatic components. The ERP data date backto 01-04-2006, spanning a period of five years. The lean implementation of the companyunder research started in 2007. The years before 2007 will be used as pre-lean years. Theyears 2007-2011 will be used as post-lean years. The annual reports date back to the year2001. The data concern the valuation of the FA per year, including depreciation based onaccountancy standards applicable to the company. Depreciation and investments takeplace every year, but stabilize over time. The data cover 11 years. For confidentialityreasons, values presented in graphs in the analysis and simulation are indexed withrespect to the year 2007. Other data sources were manufacturers’ manuals and the VSMsessions and inquiries. Financial data are of confidential nature. By indexing the data,the authors were allowed to use the financial data. The year the lean implementation

41

Measuring leanimplementation

started, 2007, is taken as the reference year. The research was closed in 2012; therefore,the population covers 11 years with 11 data points per variable.

Statistical analysisA linear regression analysis is applied to know whether the indicators are statisticallysignificant and if these have inter-relations through time to form a preliminary model.To analyse the processes, VSM by Rother and Shook (2003) and Hines et al. (1998) is usedin combination with statistics for making probability plots (Allen, 2010). These methodsare useful to depict all the process steps and transactions related to cycle time and assetspecificity like inventory and machinery.

Value stream mapping and discrete event simulationA simulation model of a system enhances the VSM model in multiple ways (Marvel andStandridge, 2009), (El-Haik and Al-Aomar, 2006). Both structural and randomvariability are commonly included in simulation models and the effects of variability onsystem performance can be determined. Discrete event simulation (Basem and Raid,2006) is used to explore the effects of continuous improvement scenarios. Design for SixSigma solutions and lean production (Houshmand and Jamshidnezhad, 2006) can bestudied before piloting or modification, and as such, risks can be better managed bypreventing defects and reducing costs that already emerge from the simulation model.

Agyapong-Kodua et al. (2009) apply system dynamics and discrete event simulationas tools to achieve a dynamic value stream model. These authors define a dynamicmodel as a model that incorporates interdependencies and interactions betweenvariables to estimate system performance of future state value streams. They explainedthat a system dynamics model based on causal loops and consequent differentialequations distinguishes cause– effect relations, and as such results in a set of potentialchange parameters. Discrete event systems are dynamic systems that evolve in time bythe occurrence of events and transactions at possibly irregular time intervals (Dotoliet al., 2012). Transactions like work orders (WOs) can randomly brought into the systemto simulate the performance of the system. The factors frequency and uncertainty arerepresented in this research by means of the number of WOs handled, the number oftasks performed and the kind of work scope of a WO reflecting transactions. Kelton et al.(2010) stated that from a practical viewpoint, simulation is the process of designing andcreating a computerized model of a real process or proposed system for the purpose ofconducting numerical experiments to give a better understanding of the behaviour ofthat system for a given set of conditions. It helps to define systems, to identify allrelevant processes, to quantify these processes, to investigate transactions and tovalidate possible future state systems and a powerful tool for Lean Six Sigma experts.

Levels of analysisWith the analysis, it is important to consider the larger processes as well as theprocesses that are necessary to execute single transactions. According to Horváth andMöller (2003), this could result in problematic interdependencies whenever the executionof processes of one level influences the other level. In line with this argumentation, thecompany under review has been divided into appropriate levels for closer investigation.The research considers the following levels: company (Level 1), department (Level 2),process (Level 3) and transactions (Level 4). Figure 2 shows the respective levels. Theproduct department handling the ACMs represents Level 2. ACMs have their own

IJLSS7,1

42

dedicated work area separated from the other products. As such, one specific ACMprocess represents Level 3 and the transactions of that process are Level 4.

• Level 1: Company: The company has been divided into two major productcategories, being the APUs (1a) and the pneumatic components (1b).

Level 2

Pneumatic Product Groups

Product Group 1

Product Group 2

Prduct Group 3

In OutLevel 1b

Level 3

In

Product process

Out

Interface A

Interface B

Department II

Department I

Process step 1

Process step 2

Process step 3

Process step n

Level 4

Transactions

Interfaff ce A

Interfaff ce B

Department II

Department I

PProcessstepp 1

PProcessstepp 2

PProcessstepp 3

PProcessstepp n

Product group ordepartment

Product process

1

OutIn

Level 1

CompanyIn Out

Variable x Variable yAPU Line

Pneumatic Products

Product process

2

Product process n

Figure 2.Company levels to

investigate

43

Measuring leanimplementation

• Level 2: ACM: The ACM product category has been chosen to represent Level 2 ofthe analysis. The data of this level have been collected and calculated from thequantum ERP database. The ACM product category, in turn, consists out ofmultiple types of ACMs; they are indicated with the type of aircraft they aredeployed in.

• Level 3: Aircraft type Boeing 747 ACM: Aircraft type has been chosen to represent thethird level of the firm. This level entails one specific product and the associatedprocesses. The Boeing 747 ACM has been chosen, as it is the ACM with the highestnumber of WOs at the ACM island throughout the period between 01/04/2006 and31/12/2011. Most of the other ACMs have too few WOs to enable for the intendedanalysis. The B747 ACM is also responsible for the largest part of the value flowthrough the ACM island, making it the most suitable candidate for analysis.

• Level 4: Transactions: This level concerns the actual transactions of a specificproduct-related assembly process. In this level, the changes and improvementsregarding the processes have to be implemented. The last step of this research is to runscenarios by discrete event simulations to know the effect of proposed changes andimprovements.

AnalysisThis section presents the data analysis regarding company Level 1. The known and newfound indicators in literature are analysed to know if the new found indicators can forma preliminary model to measure operation performances from an asset specificityperspective under the influence of lean implementation.

Value stream analysis and analysis by known indicatorsBy a value stream analysis (VSM) in 2007, a future state was designed. The process waschanged as follows:

• A pre-test of the components as first and primary process step was introduced whenthe machines arrived at the company, instead of putting the machines in stock.

• From this process step, spare parts could be ordered with suppliers earlier in theprocess.

• Disassembly and assembly were reorganized as one integrated process instead of aseparate processes. The mechanics performed both tasks.

• Mechanics certified for quality control performed their task in the assembly processinstead of a separate process.

• A pull mechanism was introduced, reducing waiting times and the number ofcomponents in the process.

The result of this lean implementation was the reduction of the ATAT from average 28 daysto 20.49 days in May and 17.62 days in June and further decreased to 13.13 days in July 2007,as reported by Beelaerts van Blokland et al. (2008b). From the graphs (Figure 3), it can benoticed that the average TATs improved from average 28 days to around 13 days, which isan improvement of more than100 per cent. The OTP increased from 35 per cent in 2006 up to80 per cent in 2008, which is an improvement of over 100 per cent.

To answer the first sub-research question: Can the year of lean implementationregarding TAT and OTP be identified? Both indicators ATAT and OTP show a change

IJLSS7,1

44

2006

2007

2008

2009

2010

2011

Aver

age

of A

TAT

29.7

229

.42

21.1

013

.71

14.9

016

.99

10.

00

15.

00

20.

00

25.

00

30.

00

35.

00

Average TAT [days]

Book

Yea

r

Deve

lopm

ent

of A

vera

ge T

urn

Aro

und

Tim

e01

/04/

2006

-31

/12/

2011

)

Aver

age

of A

TAT

2006

2007

2008

2009

2010

2011

OTP

35.1

2%40

.25%

61.9

9%78

.95%

78.9

0%77

.78%

30%

40%

50%

60%

70%

80%

90%

On Time Performance [%]

Book

Yea

r

Deve

lopm

ent

of O

n Ti

me

Perf

orm

ance

(01/

04/2

006

-31

/12/

2011

)

OTP

Figure 3.Turnaround (ATAT)

time and OTP inyears 2006-2011

45

Measuring leanimplementation

in performance starting in 2007, the year of implementing lean. The improvementcontinues up to 2009. From 2009 up to 2011, there is a stabilization. From the graphs, itcan be concluded that lean implementation started its influence indeed in 2007.

Analysis with new indicatorsTo know if the new found indicators (previous section) T/FA, I/FA and GM/FA can beused for measurement, first a linear least square analysis is applied to prove if theindicators are statistically significant. The statistical significance for linear trends istested through a two-tailed test at a level of significance provided in the tables (Allen,2010). A level of significance between 0.05 and 0.1 is accepted as boundary for statisticalsignificance. The calculated R-value is tested against the given significance level bycomparing with the corresponding critical value. The actual p value is calculated andpresented in the last column (Table AI), the number of observations N is given in thesecond column and the level of significance is positive; however, the number ofobservations is limited, as the sample set concerns yearly reporting by one company.

Analysis of the indicators over time (2001-2011) on company level resulted in thegraphs provided in Figure 4. The indicator T/FA increases from 2001 to 2011 with 433per cent. The company also shows a nearly quadrupling of the performance on GM/FA.

Regarding the model residuals, the years 2004, 2005 and 2006 show outliers regarding GM/FA. The gross margin was unstable and weak in these years. The reason for this can be found inthe low service level of 35 per cent and long TAT of the serviced products of 28 days, as stated inthe first section and Figure 3. In 2006, the company decided to change the management and startlean implementation. In Figure 4, the graph shows the weak performance regarding GM/FA intheyearsbefore2007.Thenewappointedmanagementstarted implementing lean in2007.From2006-2007, the GM improves and becomes more stable from 2008. For the indicator I/FA ,theperformance increases from 2001 to 2011 with 177 per cent.

The pattern shows a steady increase, which can be explained by the effect of areduction of the ATAT and increase of WOs, which makes it possible to graduallyincrease turnover using the same FA. Depreciation and investments took place everyyear, but stabilized over time. There were no significant investments in FA oremployees, which could have influenced the operation performance other than processimprovement. Development and significance of the indicators through time complementwith the data presented in Table AI.

To further analyse the operation performance from an asset specificity perspective under theinfluence of lean implementation, the question was raised if these indicators have inter-relations.

Analysis of relations between indicatorsThe relations between indicators T/FA, I/FA and GM/FA are exposed to a statisticalanalysis. Table AII presents the data concerning the significance of the relationsbetween indicators to a linear regression line. The relations between variables areshown to be significant, as these are within the critical value.

The next step is to know how the data spread over time regarding shape, directionand relative position, which is presented in the next section.

Measurement with the new indicatorsFigure 5 presents the development of the relation between T/FA versus GM/FA. Therelation between indicators T/FA and I/FA in Figure 6 shows a similar pattern. Therelationships between the indicators are still developing and not stable yet. Main

IJLSS7,1

46

R² =

0.9

454

020406080100

120

120

140

160

180 20

0020

0220

0420

0620

0820

1020

12

Turnover per Unit of Fixed Asset [€/€of Fixed Asset]

Book

Yea

r

Turn

over

per

Uni

t of

Fix

ed A

sset

(30/

11/2

001

-31

/12/

2011

)

T/FA

Line

ar (

T/FA

)

R² =

0.7

661

–50050100

150

200

250

300 20

0020

0220

0420

0620

0820

1020

12

Gross Margin Unit of Fixed Asset [€/€of Fixed Asset]

Book

Yea

r

Gro

ss M

argi

n pe

r U

nit

of F

ixed

Ass

et

(30/

11/2

001

-31

/12/

2011

)

GM

/FA

Line

ar (

GM

/FA)

R² =

0.7

686

020406080100

140

160

180 20

0020

0220

0420

0620

0820

1020

12

Inventory per Unit of Fixed Asset [€/€of Fixed Asset]

Book

Yea

r

Inve

ntor

y pe

r U

nit

of F

ixed

Ass

et(3

0/11

/200

1 -

31/1

2/20

11)

I/FA

Line

ar (

I/FA

)

(a)

(b)

(c)

Not

es: (

a) T/FA;

(b) GM/FA;

(c) I/FA.

[Val

ues i

ndex

ed w

ith re

spec

t to

2007

]

Figure 4.Development and

significance of theindicators through

time

47

Measuring leanimplementation

difference is the large increase in 2011 for I/FA with respect to the previous years. Whenrelating the indicators I/FA and T/FA, an increasing operation performance on I/FA ismeasured. In Figures 5 and 6, it can be observed that the performances are located in thelower left part of the graph for the years 2001-2006 compared to the performances for theyears 2008-2011, which are located in the upper left area of the graph. The year 2007 inthe middle marks the start of lean implementation and shows a first improvementcompared to 2006. The years in the lower left part can be marked as “pre lean”, and theyears in the upper right part can be marked as “after lean”. Interestingly, for all thegraphs in Figures 4 and 5, the performance increased over time; however, for the graphsin Figures 5 and 6, a remarkable improvement is visible. The measured performanceimprovement with the new indicators corresponds with the lean implementationmeasured by the indicators ATAT and OTP.

Preliminary model composed by indicators T/FA, GM/FA and I/FAThe indicators are shown to have relation over time (2001-2011). From Figures 5 and 6,it can be observed that the operation performance, measured with the new indicatorsT/FA, GM/FA and I/FA, can be orientated in the phase before lean implementation inthe lower left part of the graphs. After implementing lean in 2007, the indicators measureperformance increase, which is positioned in the upper right side of the graph.

The remarkable improvement is measured with the new indicators and is supportedby measurement with the known indicators ATAT and OTP. The OTP improved from35 per cent to 77 per cent from 2007 up to 2011. The average ATAT was halved, thedecrease started as of 2007. As these indicators ATAT and OTP refer to leanimplementation by George (2002) and Cutcher-Gershenfeld (2004), it can be reasonedthat the operation performance improved under the influence of lean implementation

2001 2002

20032004 2005

2006

2007 20082009

2010

2011

R² = 0.8867

30

50

70

90

110

130

150

0 50 100 150 200 250

Turn

ove

rper

Unit

of

Fix

ed

Ass

et

[€/€

of

Fix

ed

Ass

et]

Gross Margin per Unit of Fixed Asset[€/€ of Fixed Asset]

T/FA vs GM/FA(30/11/2001 - 31/12/2011)

Historical correlation T/FA vs GM/FA Linear (Historical correlation T/FA vs GM/FA)

Note: Values indexed with respect to 2007

Figure 5.Level 1 – T/FAversus GM/FA withclustering andquadrantsvisualization

IJLSS7,1

48

regarding TAT and OTP. In the discussion part, the influence of endogenous andexogenous factors is reviewed.

With the preliminary model, the second sub-research question can be answered:What indicators capture lean implementation for maintenance service companies andcan these indicators form a model? The answer is that the new found indicators T/FA,I/FA and GM/FA can form a model, as these show an interrelation. With the indicators,it was possible to measure operation performance from an asset specificity perspectiveunder influence of lean implementation regarding cycle time and OTP. The results aresupported by measurement with the known variables ATAT and OTP, and the momentlean implementation was actually implemented.

200120022003

20042005

2006

2007

2008

2009

2010

2011

R²= 0.7703

30

50

70

90

110

130

150

50 70 90 110 130 150

tess

Ad

exi

Ff

ot i

nU

re

pr

ev

onr

uT

[€/€

ofFi

xed

Ass

et]

InventoryperUnitofFixedAsset[€/€ ofFixedAsset]

T/FA vs I/FA(30/11/2001 - 31/12/2011)

Historical correlation T/FA vs I/FA Linear (Historical correlation T/FA vs I/FA)

2001

2002

2003

2004

2005

2006

2007

2008

20092010

2011

R²= 0.7110

50

100

150

200

250

50 70 90 110 130 150Gro

ssM

argin

perU

nit

ofFi

xed

Ass

et

[ €/€

ofFi

xed

Ass

et]

InventoryperUnitofFixedAsset[€/€ ofFixedAsset]

GM/FAvs I/FA(30/11/2001 - 31/12/2011)

Historical correlation GM/FA vs I/FA Linear (Historical correlation GM/FA vs I/FA)

(a)

(b)

Notes: (a) T/FA vs. I/FA; (b) GM/FA vs. I/FA [values indexedwith respect to 2007]

Figure 6.Level 1 – Quadrants

visualization ingraphs

49

Measuring leanimplementation

SimulationThe simulation is introduced, as several authors referred to the combination of VSM anddiscrete event simulation to explore results of improvement scenarios prior toimplementation. For this research, the preliminary model is tested by exposing the valuesystem to several improvement scenarios. The simulation is focussed on Level 4,transaction level (Figure 2). This level concerns the actual transactions of a specificproduct-related assembly process. In this level, the changes and improvementsregarding the processes have to be implemented. The first step is the baselinesimulation, and the second step is to run scenarios by discrete event simulations to knowthe effect of proposed changes and improvements.

Baseline simulationThe simulation starts at the first step, logistics, where ACMs come in together with otherpneumatic components. In the second step, the ACMs wait at a receiving area until theirWO is created and the routing for a specific work scope is known. Within the third step,five existing types of ACMs are routed to “Other ACM processes”, where they areprocessed through six process steps. When ready, the ACMs are routed to logisticsagain, where specific statistics are gathered, and they are shipped when a shuttle leaves.Another step shared by multiple products is customer account representatives, wherevarious hold-ups and accompanying processing are simulated. Last shared modellingblock is the WO creation block, where WOs and some entity specifics are assigned. Thefictive new ACM is simulated in detail with four work scopes:

(1) repair (spinning condition);(2) overhaul (seized condition);(3) test only; and(4) return as is.

A fifth classification could occur during the simulation, when repair of a certain unit isnot economical anymore.

The baseline simulation results have been validated by comparison with historical dataand by consultation of a process expert. One simulation run entails 100 replications of 365days each, this provides with accurate averages and half width values. The half width valueprovides with the range within which 95 per cent of the measurement averages fall and is astandard output of the used Arena software.

Changes for improvement, transactional Level 4With the baseline simulation in place, four changes of improvement on the Level 4(transactional level) are proposed. The model is rebuilt accordingly into the final model,and the changes of improvement are derived from the principles of lean manufacturing:

• Levelling schedules between various departments: Working hours betweendepartments are harmonized, at least one person should be present at everydepartment at the start of the working day. A continuous check whether there areparts in need of non-destructive test inspection is part of this category; in the basesimulation, this is only checked twice a day. Reconciling working hours could preventsome specific hold-ups. Theoretically, the higher the incoming flow of new ACMs, thehigher the benefit of levelled schedules should be.

IJLSS7,1

50

• Inventory: The second improvement entails shorter lead times for parts ordering,through negotiation, cooperation and a higher level of supplier involvement, therebyallowing for lower inventory levels. Some other small delays in parts ordering aretaken out as well with this change.

• Introducing additional transactions: The third Level 4 alteration concerns the internalintroduction of a necessary repair; this is, however, only an example of probably manypossible extra internal transactions. Purpose is to show that the number oftransactions does not necessarily need to be minimized for better overall systemperformance or lower total cost. Recall that the total costs are a summation of the totaltransaction costs and the total production costs, and that the total costs need to beminimized.

• Single piece flow on the critical parts level: Single piece flow is not only applied to thecomplete product, but it is projected down to the critical parts level. Mechanics arescheduled per process stage and parts flow through the processes as they becomeavailable. Parts are not batched between process stages, and therefore, it could be thatsome parts already enter a process when other parts of the same unit are still beingdisassembled. This transactional change does not only affect the new ACM, but alsothe other ACMs.

Results of simulation for ATAT and OTPThe simulation showed that the Level 4 changes have a large impact on the measurespresented in the improved simulation results. Figure 7 shows the results on the generalLSS indicators of ATAT and OTP. ATAT is the TAT minus the hold-up time thecustomer is responsible for. Figure 7 also shows the indexed values of average ATATsand OTPs for the following detailed cases:

• the five other ACMs combined;• all new ACMs together;• new ACMs having repair as work scope;• new ACMs having overhaul as work scope; and• new ACMs having test only as work scope.

It can be observed that all indicators have improved for the final simulationcompared with the base simulation. The new ACM process line improved withalmost 19 per cent on average, according to the graph in Figure 7. Level 4improvement of the performance increase must be read relative to 100 per cent of thebaseline simulation.

Scenarios performed with the base and final simulationTo demonstrate how the simulation models behave, five scenarios are performed onboth the base and final simulation. The results are presented:

(1) Scenario change 1: theoretical upper performance boundary of the ACM process:This scenario demonstrates the theoretical upper performance boundary of thenew isolated ACM process line based on the current product budget, the sameinitial inventory levels and also the same amount of approved and authorizedmechanics available for the process.

51

Measuring leanimplementation

(2) Scenario change 2: flooding the ACM process with incoming products: This scenarioentails flooding the simulation with new incoming ACMs to get an indication whenthe system would “choke”, and what the maximum throughput of the assembly linecould be. The numbers of the various available transportation carts are all set to alevel such that they will not be the bottleneck of the system.

(3) Scenario change 3: flooding the new ACM process with a theoretical upper boundary:This scenario is a combination of Scenarios 1 and 2. It entails flooding Scenario 1with incoming new ACMs as specified in Scenario 2. With this scenario, it isdetermined what theoretical maximum throughput, or value flow, the isolated ACMprocess line could handle.

ATAT otherACMs

ATAT newACM alL ATAT repair ATAT

overhaulATAT test

onlyBase simulation 100 100 100 100 100

Improved simulation 65.1 81.3 80.7 79.4 90.8

0

20

40

60

80

100

Inde

x va

lue

of c

alen

dar

days

OTP otherACMs

OTP newACM all OTP repair OTP

overhaulOTP test

onlyBase simulation 100 100 100 100 100

Improved simulation 116.7 109.4 112.3 120.4 103.5

0

20

40

60

80

100

120

Inde

x va

lue

of %

(a)

(b)

Notes: a) ATAT before and after Level 4 improvements; b) OTP before and after Level 4improvements

Figure 7.Indexed ATAT andOTP of thesimulation runsbefore and after theLevel 4improvements

IJLSS7,1

52

(4) Scenario change 4: assigning an additional technician for testing of other ACMs:This scenario shows the impact of extra personnel and/or investing in training forapprovals/authorizations.

(5) Scenario 5: no stock, only housing set: This scenario shows that without inventory, itis for the improved model still possible to achieve relatively high OTPs. This provesthat the JIT practice of the final model compared to the base best practice model ishighly improved.

Results of scenarios with new found indicators T/FA, GM/FA and I/FAThe relational measurement points for each of the changes (1-4) are projected tovisualize their impact within the three graphs in Figure 8. In each graph, alpha line isconnected to the measurement point of the baseline simulation model. The simulationswere modelled according to actual future budgets from the company; as such a restrictedmaximum exists for the total output. As the total output of the new ACM with thebaseline model approaches the total budget already, T/FA remains constant whenimproving the system. The following can be observed from the graphs in Figure 8 for therelations between the lean transaction cost efficiency indicators T/FA, GM/FA and I/FAas a result of the changes:

• T/FA versus GM/FA:The GM is maximized due to additional transactions(Change 4) by WOs up to the boundary capacity of the system. It shows themaximum utilization of the FA for the system. The final performance is locatedon the right upper side of the graph relative to the index 100.

• T/FA versus I/FA: Shorter lead times for parts ordering, through negotiation,cooperation and a higher level of supplier involvement, allow for lower inventorylevels. This causes a better utilization of inventory. The final performance islocated at the left side of the graph relative to the index 100. The value of I/FA hasbeen more than halved due to improvement of inventory turns. Main contributoris Change 2, in which specifically the inventory management is reviewed.

• GM/FA versus I/FA: The shorter lead times of parts as mentioned regardingT/FA-I/FA, the positive effect on GM/FA can be expected. The final performanceis located at the upper left side of the graph relative to the index 100. Relative to thebaseline model, the increase is approximately 50 per cent according to the graph,while the I/FA was halved.

DiscussionAlthough there is a remarkable improvement in performance from the year 2006 to2008, as presented in Figures 3, 4 and 5, the question was raised if this can beaccredited to lean implementation only. In the first place, this measurement needs tobe tested with other cases in the MRO service industry by further research, as thismeasurement concerns one SME; however, the measurement spanned an 11-yearperiod. Second, the moment of lean implementation in 2007 can be identified by thegraphs representing the performance on known indicators ATAT and OTP. Theimprovements measured with these indicators are in line with the improvementmeasured with the new found indicators. Third, exogenous factors such as changingmarket demand may have influenced the results. Although the airline industry hasbeen changed over the past due to liberalization of state monopolies and the coming

53

Measuring leanimplementation

of low-cost carriers, MRO services for high complex products such as ACMs andAPUs are growing due to the production increase of aircraft during the past 15years. Regarding endogenous factors, there were no exceptional additionalinvestments in equipment which can explain for the exceptional performanceimprovement. It is plausible more resources such as personnel have been calledupon to cope with increasing customer demand, which may have influenced theresults. However, this will reflect the gross margin.

Base

mod

el

Chan

ge 1

Chan

ge 2

Chan

ge 3

Chan

ge 4

Fina

l mod

el

y = x

y = 2

.041

1x

020406080100

120

140

020

4060

8010

012

014

0

Turnover per Unit of Fixed Asset [-/-of Fixed Asset]

Inve

ntor

y per

Uni

t of F

ixed

Ass

et

[-/-o

f Fixe

d As

set]

T/FA

vs I/

FA(0

1/04

/200

6 -31

/12/

2011

)

Rela

tion

of T

/FA

vs I/

FAal

pha

line

beta

line

Base

mod

el

Chan

ge 1

Chan

ge 2

Chan

ge 3

Chan

ge 4

Fina

l mod

el

y = x

y = 2

.942

2x

020406080100

120

140

160

180

200

020

4060

8010

012

014

0

Gross Margin per Unit of Fixed Asset [-/-of Fixed Asset]

Inve

ntor

y per

Uni

t of F

ixed

Ass

et

[-/-o

f Fixe

d As

set]

GM/F

A vs

I/FA

(01/

04/2

006 -

31/1

2/20

11)

Rela

tion

of G

M/FA

vs I/

FAal

pha

line

beta

line

Base

mod

elChan

ge 1

Chan

ge 2

Chan

ge 3

Chan

ge 4

Fina

l mod

el

y = x

y = 0

.693

7x

020406080100

120

140

020

4060

8010

012

014

016

018

020

0

t ess A dexi Ff oti nUr ep r evonr uT[-/-of Fixed Asset]

Gros

s Mar

gin p

er U

nit o

f Fix

ed A

sset

[-/

-of F

ixed

Asse

t]

T/FA

vs G

M/FA

(30/

11/2

001 -

31/1

2/20

11)

Rela

tion

of T

/FA

vs G

M/FA

alph

a lin

ebe

ta li

ne

(a)

(c)

(b)

Figure 8.Impact of theindividual Level 4changes andcombined effect inthe final model

IJLSS7,1

54

Furthermore, the indicators are tested by simulation for the continuousimprovement scenarios. The results of the changes to improve on the baselinesimulation are indicated with the alpha line, and the results of the changes ofimprovement on the final simulation are indicated with the beta line in Figure 8,presenting the final Level 4 changes for the final simulation model. As simulationtakes place in a virtual environment, different scenarios– changes can beimplemented and evaluated much easier and cost-effective than it would in real life.After the virtual improvements, new performances result from the simulation. Theoutcome can be projected in the quadrant graphs in Figure 9. Comparing � and �, orthe slopes, gives an indication of whether the relation between the indicators T/FA,GM/FA and I/FA is better than before the change. It depends on the x- and y-axisindicators whether � � � or � � � indicate an improvement. The reasoning hasbeen made on whether the angle improved, got closer to a certain target set or gotbetter than a target that was set. As such, different scenarios can be compared. Thefirst simulation outputs served as baseline measurement to be put in the modelproposed by Figure 9. The x1 and y1 are set by this baseline measurement, whichserves as a local optimum. As such, x1 and y1 axes form the quadrant boundaries.The center line through the point results in the angle �, indicating the presentrelation between the variables on the axes.

Statements can be made on whether the angle improved, got closer to a certain target setor got perhaps better than a target that was set. It is also a possibility to perform multiplesimulation iterations and to project the results in the graph. As such, different scenarios canbe compared. The performance of the (new) process can be evaluated by the simulation.Multiple simulation runs should provide with the same measurements if a process is stable.Quantitatively, the improvement can be assessed to some extent by the statistical outputs ofthe simulation, for example the average, minimum value, maximum value and the half widthvalue. The half width value provides with the range within which 95 per cent of themeasurement averages fall and is therefore an important indicator for stability. An iterativesimulation process evaluated by a general analysis, the lean transaction cost efficiencyindicators and the proposed quadrant model provides with a method to improve andvalidate new processes.

For this research, relations between indicators can be applied as per Figure 8:• T/FA versus GM/FA: The final measurement is located in quadrant 3. With � � �,

this places the measurement in part 3a, which is the most preferred part. For thisrelation, � � � indicates that the improved system performs better on thisrelationship with respect to the base measurement.

• T/FA versus I/FA: The final measurement is located in quadrant 4, which alsoautomatically means that � � �. With � � �, the improved simulated systemperforms better on this relationship with respect to the base measurement, therebyquadrant 4 is preferred the most for the final measurement to be located in.

Figure 9.Measurement and

prediction of a newprocess bysimulation

55

Measuring leanimplementation

• GM/FA versus I/FA: The final measurement is located in quadrant 4, which alsoautomatically means that � � �. With � � �, the improved simulated systemperforms better on this relationship with respect to the base measurement, and similarto the previous relation, quadrant 4 is preferred the most for the final measurement tobe located in.

Managerial implicationsThe simulation perspective opens four interesting implications for managers. Firstly, thesimulation tool offers the management the capability to discuss the effect of making changesto assets and resources before deciding to invest in these. Secondly, managers can determinethe progress in operation performance by monitoring the positions in the graphs through theyears. By monitoring the position in the quadrant according to Figure 8, managers canadjust assets and resources to influence operation performance. The lean implementationstrategy framework (Figure 10) shows for instance that an increase of turnover due togrowth of market share may harm the benefit or GM, which results in a position in quadrant4. Once the operation performance is positioned in quadrant 1, the aim is to move to quadrant3 by an improved future state process. The lean implementation strategy framework(Figure 10) guides the managers for finding the balance between T/FA and GM/FA underinfluence of I/FA. Thirdly, the quadrant perspective can form the basis for developing abenchmark tool for maintenance service companies. In the fourth place, the processsimulation can form the basis for the design of a Greenfield layout of a future state processand factory.

ConclusionsThis research learned from literature that measuring lean implementation through timefrom a financial economic perspective was missing. From literature, it was derived that leanimplementation not always leads to the benefits it claims. However, if there is commitmentby the management with a top-down approach, it could be successful. In this case, the

Figure 10.Lean implementationstrategy framework

IJLSS7,1

56

management took leadership to implement lean in 2007. Support was found for the relationbetween cultural and organizational aspects like lean implementation, operationalperformance and FA. The airline industry showed links between TAT and OTP in relationwith utilization of the FA like Ryanair and EasyJet. From the analysis it can be concluded thatby adding the FA perspective, new indicators could be developed to measure operationalperformance under the influence of lean implementation for a maintenance service company inthe airline industry. Lean implementation is embodied by cycle time (ATAT) and OTP. Byintroducing FA in combination with turnover, inventory and gross margin, operationalperformance can now be measured by the lean transaction cost efficiency indicators.

Contribution to theoryThe MRO service industry can be characterized by regulations and certifications due togovernment laws for safety of the customers making use of airlines. As a result, the productsand services are of a high complex nature in combination with high asset specificity. Theproposed main research question, how to measure operation efficiency from an assetspecificity perspective under the influence of lean implementation for maintenance servicecompanies in the aviation industry, is answered with the indicators T/FA, GM/FA andI/FA. The indicators have an interrelation and form a model. The found indicators areadding asset specificity and complementary to the already known Lean Six Sigmaindicator cycle time or TAT. The indicator OTP can be added to the theory of Lean SixSigma (Cutcher-Gershenfeld, 2004).

The model has been validated on company level by observing significant correlationbetween the outcome of the model and the number of years that lean has beenimplemented at the sample set, and by observing positive trend breaks in theperformance variables. The model has been tested with a discrete event simulation forfour model changes and five experimenting scenarios to identify the upper boundariesof the MRO service provider system. By means of four transactional alterations to asimulated fictive new ACM process, it is shown that both the operation performance onthe general Lean Six Sigma indicators and the lean transaction cost efficiency indicatorsare useful to determine the upper boundaries of the value system. From the analysis andsimulation, the following conclusions can be drawn:

• The found indicators were significant through time over the period 2007-2011. Theindicators are interrelated and form a model.

• Theslopeof thetrendlinewaspositive, indicatingapositivechangeontheperformanceofthe separate indicators. GM/FA increased with 250 per cent, T/FA increased with 433 percent and I/FA increased with 177 per cent over the period 2001-2011.

• Remarkable improvement between the phase before the implementation of Lean SixSigma up to 2007 and the sustainment phase, which emerged from 2008 up to 2011was measured with the new indicators.

• The preliminary model enables the implementation of changes to improveperformance by scenarios and to identify the upper performance boundaries of thevalue system.

Contribution to practiceFrom a practical perspective, the model helps to identify all relevant processes, as well as toquantify these processes and to validate possible future state systems, which is a useful tool

57

Measuring leanimplementation

for Lean Six Sigma experts. With the analysis, it would be possible to further develop thismodel for the managers as well as consultants as a tool for business planning. By testingscenarios, it is possible to design improved processes without disrupting current operations.As such, a new process can be balanced from a lean and transactional perspective before itmay be implemented in reality.

Further researchThe model needs to be validated on more MRO service companies in the airline industry. Itcan also be further developed for other type of service industries like hospitals for patientcare services as well as manufacturing industries.

The costing function underlying the GM used in the indicator GM/FA is limited in itsdefinition. GM is defined as the difference between turnover and the summation of materialcosts and value-added labor cost. Further research could expand the cost function with, forexample, material holding costs, employee idle costs and other indirect costs. By doing so,results would be more accurate.

Asset specificity is the most crucial factor in TCE. To provide for an empiricalmeasurement and indication of transaction costs, the book value of the MRO company’s FAis taken as a proxy for the asset specificity. Intangible FA like company brand name orintrinsic employee knowledge have not been investigated in this research and are notincluded in FA. The key problem of measuring intangible assets is that they influencethe results only in an indirect manner, mostly in the form of multi-level cause-and-effectrelationships. However, if future developments allow for displaying these cause-and-effectrelationships regarding both tangible and intangible value creation potentials, this couldrefine the definition of asset specificity used in this research.

ReferencesAchanga, P., et al. (2006), “Critical success factors for lean implementation within SMEs”, Journal

of Manufacturing Technology Management, Vol. 17 No. 4, pp. 460-471.Agyapong-Kodua, K., Ajaefobi, J.O. and Weston, R.H. (2009), “Modelling dynamic value streams

in support of process design and evaluation”, International Journal of Computer IntegratedManufacturing, Vol. 22 No. 5, pp. 411-427.

Allen, T. (2010), Introduction to Engineering Statistics and Lean Sigma, Springer-Verlag, London.Arnold, U. (2000), “New dimensions of outsourcing: a combination of transaction cost economics

and the core competencies concept”, European Journal of Purchasing & SupplyManagement, Vol. 6 No. 1, pp. 23-29.

Arora, P. (2007), Book Review; The Economic Institutions of Capitalism by Oliver Williamson,Ezine @rticles.

Basem, E. and Raid, A. (2006), Simulation-Based Lean Six-Sigma and Design for Six-Sigma,Wiley-Interscience – John Wiley & Sons, Hoboken, NJ.

Beelaerts van Blokland, W., de Waard, A. and Curran, R. (2008a), “Design of a lean MRO businessprocess for EPCOR – how to decrease process variability and reduce lead time”,AIAA-ATIO 2008 Conference Proceedings, Anchorage.

Beelaerts van Blokland, W.W.A., Huijser, R., Stahls, R. and Santema, S.C. (2008b), Future AirportTurnaround Ground Handling Processes, How to Reduce the Turn-Around Time ofAircraft at the Airport, TRAIL Research School, Delft.

Beelaerts van Blokland, W., Fiksinski, M., Amoa, S., Santema, S., van Silfhout, G.-J. andMaaskant, L. (2012), “Measuring value-leverage in aerospace supply chains”, InternationalJournal of Operations & Production Management, Vol. 32 No. 8, pp. 1-22.

IJLSS7,1

58

Bhasin, S. (2011), “Measuring the leanness of an organisation”, International Journal of Lean SixSigma, Vol. 2 No. 1, pp. 55-74.

Canbäck, S. (1998), “Transaction cost theory and management consulting – why do managementconsultants exist?”, Working Paper, Henley Management College, Oxfordshire.

Canbäck, S., Samouel, P. and Price, D. (2006), “Do diseconomies of scale impact firm size andperformance? A theoretical and empirical overview”, Journal of Managerial Economics,Vol. 4 No. 1, pp. 27-70.

Chün, P. (2009), Sluit Efficiëntie (Lean) een innovatie cultuur uit? De kracht van het EN-paradigma,TSM Business School, Enschede.

Cutcher-Gershenfeld, J. (2004), Six Sigma Systems Principles, MIT Sloan School of Management,Massachusetts.

Doganis, R. (2001), The Airline Business in the Twenty-First Century, Routledge, London.Doganis, R. (2002), “Flying off course”, The Economics of International Airlines, Routledge, London.Dotoli, M., Fanti, M.P., Iacobellis, G. and Rotunno, G. (2012), “A lean manufacturing strategy using

value stream mapping, the unified modeling language, and discrete event simulation”,IEEE International Conference on Automation Science and Engineering (CASE), Seoul.

Douglas, W.A. (2000), “Transactiekosten”, in Boudewijn, B., Gerrit, D.G. (Eds), De Encyclopedievan Law and Economics, Edward Elgar, Chelthenham, Vol. 1, pp. 893-926.

El-Haik, B. and Al-Aomar, R. (2006), Simulation-Based Lean Six-Sigma and Design for Six-Sigma,Wiley-Interscience – John Wiley & Sons, Hoboken, NJ.

European Federation of National Maintenance Societies vzw. (2011), “What does EFNMS standfor?”, available at: www.efnms.org/

George, M.L. (2002), Lean Six Sigma: Combining Six Sigma Quality with Lean Speed, TheMcGraw-Hill Companies, New York, NY.

Hall, G., Hutchinson, P. and Michaelas, N. (2000), “Industry effects on the determinants of unquotedSMEs’ capital structure”, International Journal of the Economics of Business, No. 3, pp. 297-312,available at: www.tandfonline.com/loi/cijb20?open�7/0l vol_7 Volume 7, www.tandfonline.com/toc/cijb20/7/3

Hardt, L. (2009), “The history of transaction cost economics and its recent developments”,Erasmus Journal for Philosophy and Economics, Vol. 2 No. 1, pp. 29-51.

Hines, P., Rich, N., Bicheno, J., Brunt, D., Taylor, D., Butterworth, C. and Sullivan, J. (1998), “Valuestream management”, The International Journal of Logistics Management, Vol. 9 No. 1,pp. 25-42.

Horváth, P. and Möller, K. (2003), “Supply chain performance measurement, ‘a transaction costtheory – and value-based approach’”, Paper Präsentiert auf dem 2nd Workshop onPerformance Measurement and Management, Nizza, Vol. 9 Nos 18/19.

Houshmand, M. and Jamshidnezhad, B. (2006), “An extended model of design process of leanproduction systems by means of process variables”, Robotics and Computer-IntegratedManufacturing, Vol. 22 No. 1, pp. 1-16.

Jacobs, M.A and Swink, M., (2011), “Product portfolio architectural complexity and operationalperformance: incorporating the roles of learning and fixed assets”, Journal of OperationsManagement on ScienceDirect, Journal of Operations Management, Vol. 29 Nos 7/8,pp. 677-691, available at: www.sciencedirect.com/science/article/pii/S0272696311000702,www.sciencedirect.com/science/article/pii/S0272696311000702, www.sciencedirect.com/science/journal/027269630/, www.sciencedirect.com/science/journal/02726963/29/7”, Go toTable of Contents for this Volume/Issue.

Jaggi, B. and Tsui, J. (2001), “Management motivation and market assessment: revaluations offixed assets”, Journal of International Financial Management & Accounting, Vol. 12 No. 2,

59

Measuring leanimplementation

pp. 160-187, available at: http://onlinelibrary.wiley.com/doi/10.1111/jifm.2001.12.issue-2/issuetoc

Jeong, K.-Y. and Phillips, D.T. (2011), “Application of a concept development process to evaluateprocess layout designs using value stream mapping and simulation”, Journal of IndustrialEngineering and Management, Vol. 4 No. 2.

Kelton, W., Sadowski, R. and Swets, N. (2010), Simulation with Arena, 5th ed., McGraw-Hill, NewYork, NY.

Lu, J.C., Yang, T. and Wang, C.Y. (2011), “A lean pull system design analysed by value streammapping and multiple criteria decision-making method under demand uncertainty”,International Journal of Computer Integrated Manufacturing, Vol. 24 No. 3.

Mann, D. (2010), Creating a Lean Culture, Tools to Sustain Lean Conversions, Taylor & FrancisGroup, Productivity Press, New York.

Marvel, J. and Standridge, C. (2009), “A simulation-enhanced lean design process”, Journal ofIndustrial Engineering and Management, Vol. 2 No. 1, pp. 90-113.

Masten, S. (1984), “The organization of production: evidence from the aerospace industry”, Journalof Law and Economics, Vol. 27 No. 2, pp. 403-417.

Monteverde, K. and Teece, D. (1982a), “Supplier switching costs and vertical integration in theautomobile industry”, Bell Journal of Economics, Vol. 13 No. 1, pp. 206-213.

Monteverde, K. and Teece, D. (1982b), “Appropriable rents and quasi-vertical integration”, Journalof Law and Economics, Vol. 25 No. 2, pp. 321-328.

Murman, E., Allen, T., Bozdogan, K., Cutcher-Gershenfeld, J., McManus, H., Nightingale, D. andWidnall, S. (2002), Lean Enterprise Value – Insights from MIT’s Lean Aerospace Initiative,PALGRAVE, Houndmills, Basingstoke, Hampshire, NY.

Rother, M. and Shook, J. (2003), Learning to See: Value Stream Mapping to Create Value andEliminate Muda, The Lean Enterprise Institute, Massachusetts, MA.

Scherrer-Rathje, M., Boyle, T.A. and Deflorin, P. (2009), “Lean, take two! Reflections from thesecond attempt at lean implementation”, Business Horizons, Vol. 52 No. 1, pp. 79-88.

Schmidberger, S., Bals, L., Hartmann, E. and Jahns, C. (2008), “Ground handling services at Europeanhub airports: development of a performance measurement system for benchmarking”,International Journal for Production Economics, Vol. 117 No. 1, pp. 104-116.

Shelanski, H. and Klein, P. (1995), “Empirical research in transaction cost economics: a review andassessment”, Journal of Law, Economics and Organization, Vol. 11 No. 2, pp. 335-361.

Singh, B., Garg, S.K., Sharma, S.K. and Grewal, C. (2010), “Lean implementation and its benefits toproduction industry”, International Journal of Lean Six Sigma, Vol. 1 No. 2, available at:www.emeraldinsight.com/loi/ijlss, www.emeraldinsight.com/toc/ijlss/1/2

Williamson, O.E. (1979), “Transaction-cost economics: the governance of contractual relations”,Journal of Law and Economics, Vol. 22 No. 2, pp. 233-261.

Williamson, O.E. (1981), “The economics of organization, ‘the transaction cost approach’”,American Journal of Sociology, Vol. 87 No. 3, pp. 548-577.

Williamson, O.E. (1985), The Economic Institutions of Capitalism, The Free Press, New York, NY.Williamson, O.E. (1998), “Transaction cost economics: how it works, where it is headed”, The

Economist, Vol. 146 No. 1, pp. 23-58.Womack, J. and Jones, D. (2005), Lean Solutions – How Companies and Customers Can Create

Value and Wealth Together, Simon & Schuster, New York, NY.Womack, J.P. and Jones, D.T. (2003), Lean Thinking: Banish Waste and Create Wealth in Your

Corporation, Simon & Schuster, Inc, New York.Womack, J.P., Jones, D.T. and Roos, D. (1990), The Machine that Changed the World, Simon &

Schuster, Inc, New York.

IJLSS7,1

60

Further readingMasten, S.E. (2000), “Transaction-cost economics and the organization of agricultural transactions”, in

Baye, M.R. (Ed.), Industrial Organization, Emerald Group Publishing Limited, UK, Vol. 9,pp. 173-195.

Appendix

About the authorsStephan J. de Jong is currently a Senior Consultant at EY’s Operational Transaction Services. Hisfocus is on Operational, Commercial and Technology (IT) Due Diligence. Stephan J. de Jong is alsoexperienced in integration support and carve-out readiness assessments in multiple industries. Heobtained MSc in Aerospace Engineering, specializing in Aerospace Management and Operationsin 2012. At the time, Stephan conducted the research at the MRO company where he was as aBusiness Analyst. Afterwards, he worked at TNO (The Netherlands Organization for AppliedScientific Research) and at Deloitte as a consultant in the field of Operations Excellence. The maindisciplines at both companies concerned Process Enhancement, Operations Strategy and DemandFlow and Lean Manufacturing.

Wouter W.A. Beelaerts van Blokland joined the Delft University of Technology, TheNetherlands, in 2004, to become a Lecturer on value chain modeling and lean air transport systemswithin the chair air transport and operations (ATO) under direction of Prof Dr Ir Santema. Hesuccessfully defended his PhD thesis on “Value-leverage by aerospace original equipmentmanufacturers” October 27, 2010 at Delft University of Technology in The Netherlands. He joinedthe faculty of Mechanical Engineering section Transport Engineering and Logistics underdirection of Prof Dr Ir G. Lodewijks in 2013 as an Assistant Professor. He continued research onthe development of methods to measure operations performance from a lean perspective involvingproduction and service companies in general and in aviation. Wouter W.A. Beelaertsvan Blokland is the corresponding author and can be contacted at: w.w.a.beelaertsvanblokland@tudelft.nl

For instructions on how to order reprints of this article, please visit our website:www.emeraldgrouppublishing.com/licensing/reprints.htmOr contact us for further details: permissions@emeraldinsight.com

Table AI.Level 1: Statistical

significance for theindicators over time

Level N Indicators R2-value R-value Significance level Critical value Significant Actual p

1. 11 T/FA 0.9454 0.9723 0.001 0.847 Yes 0.000001. 11 GM/FA 0.7661 0.8753 0.001 0.847 Yes 0.000421. 11 I/FA 0.7686 0.8767 0.001 0.847 Yes 0.00040

Table AII.Level 1: Statistical

significance relationsbetween the

indicators

Level N Indicators R2-value R-value Significance level Critical value Significant Actual p

1. 11 T/FA vs GM/FA 0.8867 0.9416 0.001 0.847 Yes 0.000021. 11 T/FA vs I/FA 0.7703 0.8777 0.001 0.847 Yes 0.000381. 11 GM/FA vs I/FA 0.711 0.8432 0.01 0.735 Yes 0.00111

61

Measuring leanimplementation