quality improvement – the six sigma way
TRANSCRIPT
Quality Improvement – The Six Sigma Way
Mala Murugappan and Gargi KeeniTata Consultancy Services
Abstract
Six Sigma provides an effective mechanism to focuson customer requirements, through improvement ofprocess quality. In the Global Engineering DevelopmentCenter of Tata Consultancy Services (TCS – GEDC) atChennai, India, Six Sigma projects are being carriedout with the objective of improving on time delivery,product quality and in-process quality.
This paper describes the application of the SixSigma methodology comprising the five phases –Define, Measure, Analyze, Improve and Control, bytaking the example of a project, and demonstrates thebenefits attained.
1. Background
The Global Engineering Development Center ofTata Consultancy Services (TCS–GEDC) located atChennai, India, has been executing projects for variousbusinesses of General Electric Company (GE) since1995. The projects span the areas of EngineeringDesign, Computer Aided Design, Finite ElementAnalysis, Software Development for EngineeringAutomation and Implementation of Product DataManagement Solutions. TCS – GEDC has a number ofinitiatives in quality and process management and is onits journey to Level 5 of the Capability Maturity Model(CMM) enunciated by the Software EngineeringInstitute of Carnegie Mellon University [1]. TCS –GEDC's long association with GE has influenced it toadopt the Six Sigma Approach for making disciplinedand rigorous progress in quality improvement. TCS –GEDC learns and leverages from the experience of GEin the implementation of Six Sigma Initiatives.
Six Sigma Quality quantitatively means thatthe average review process generates 3.4 defects permillion units – where a unit can be anything rangingfrom a component to a line of code or an administrativeform. This implies that nearly flawless execution of keyprocesses is critical to achieve customer satisfaction andproductivity growth [2].
The Six Sigma Initiatives in TCS – GEDC startedin 1998, and, since then, 10 Six Sigma projects havebeen completed. These projects include Improvement ofSchedule Compliance, Quality Compliance, InputQuality, Error Reduction, Cycle Time Reduction andDesign Improvement.
At present the Six Sigma Program addressesProductivity Improvement and Defect Prevention in theprojects carried out in the TCS – GEDC. The Six Sigmaprojects have been identified in the areas ofimprovement after analysis of the process and productmetrics against the center level specification limits.Currently, TCS – GEDC has 6 Black Belt and 30Green Belt ongoing projects in different BusinessGroups. The Green Belt Projects are derived from theanalysis phase / improvement phase of the Black Beltprojects.
The approach, methodology and benefits of SixSigma is explained in this paper by taking the case of aSix Sigma project. This project is on the improvementof product quality compliance carried out in one of theBusiness Groups of TCS – GEDC in 1998–1999.
2. The Six Sigma approach
The Six Sigma Approach is customer-driven. For abusiness or a manufacturing process, the SigmaCapability is a metric that indicates how well theprocess is being performed. The higher the SigmaCapability, the better, because it measures the capabilityof the process to achieve defect-free-work (where adefect is anything that results in customerdissatisfaction).
The Six Sigma Approach is also data-driven. Itfocuses on reducing process variation, centering theprocess and on optimizing the process. The emphasis ison the improvement of process capability rather than the
A defect is anything thatresults in customer
dissatisfaction
control of product quality, which includes theimprovement of quality and reduction of cost of quality.
In short, the Six Sigma Approach focuses on: Customer needsData-driven improvementsThe inputs of the processAnd this results in:Reducing or eliminating defectsReducing process variationIncreasing process capabilityIn this context, the customer requirements for our
centre are -On-Time, Accurate and Complete Customer
DeliverablesCustomer Responsiveness Marketplace CompetitivenessTo achieve these goals the Six Sigma approach was
adopted in all the projects to pinpoint sources of errorsand ways of eliminating them.
3. Deployment
All employees are trained on Six Sigma Quality toincrease their awareness, understanding, and the day-to-day use of Six Sigma tools and processes, and theirapplication to projects.
Six Sigma projects (quality projects) are chosen,based on customer feedback and analysis of the processmetrics. Projects that have a significant customer impactand financial savings are given top priority.
To successfully execute these Six Sigma projects,an organization structure as shown in Figure 1-1 hasbeen formulated. This structure consists of the followingroles:Champion: The Center Manager is the Champion, whofacilitates the implementation / deployment of the SixSigma Program. The Champion creates the vision,defines the path to Six Sigma Quality, measures theprogress and sustains improvements.
Master Black Belt: The Master Black Belt is aMentor, who develops a Six Sigma network, provides
training on strategies and tools, gives one-to-onesupport on utilization and dissemination of Six Sigmatools, and supervises the Six Sigma projects. The MasterBlack Belt also facilitates sharing of best practices andactively participates in the change process.Black Belts / Green Belts: Black Belts / Green Belts leadprocess improvement teams, demonstrate credibleapplication of Six Sigma tools, train their team membersand are accountable for Six Sigma project results. WhileBlack Belts work full-time on Six Sigma projects, theGreen Belts work only part-time on the Six Sigmaprojects and devote the rest of their time to otherprojects. The Black Belts report the progress of theirprojects to the Master Black Belt and the Green Beltsreport the progress of their projects to the Black Belts.The Green Belts are expected to complete two SixSigma projects while the Black Belts are to completefive.
The criteria for a successful Six Sigma projectcould be any one of the following:
G reen B e lts
B lac k B e lt(G E -P S g rou p )
G reen B e lts
B lac k B e lt(G E -TS g rou p )
G reen B e lts
B lac k B e lt(G E -N P g rou p )
G reen B e lts
B lack B e lt(A U TO M A TIO N g rou p )
G reen B e lts
B lack B e lt(F E A g rou p )
M as ter B lack B e lt
C H A M P IO N(CENTER MANAG ER)
Figure 1-1 Organization structure for Six Sigma
i. There is an improvement of one sigma in theprocess capability, if the project has started theprocess with less than one Sigma
ii. There is a 50% reduction in defects, if the processhas started with more than three Sigma
iii. There is a Return on Investment of 20%.
PotentialFailure Modes
PotentialCauses
CurrentControls
RiskPriorityNumber
Recommended Action
Dispatchincomplete
WorkPressure
No Control 200 Development of tools for completedelivery
PDM structurenot followed
Lack ofknowledge
No Control 180 PDM file handling procedure to beevolved.
Nonconformance ofinputs withStandards
Informationnot given toall teammembersagainstblindlyfollowingSIMTOs
No Control 160 Maintain a text file in pkg. directory,where a complete track of pkg. progressis recorded to address any deviations /non-conformances up front.
Incompletedispatch
Lack ofknowledge
No Control 140 Process Procedure to include all thedispatch procedures.
Noncompliance toStandards
QA toollimitations
QA Tool 120 Limitations will be fixed by Jan ’99end
Table 5 -1 Failure Modes and Effects Analysis
4. Methodology
A Six Sigma Project consists of the following phases:Define: The product or process to be improved isidentified. Customer needs are identified and translatedinto Critical to Quality Characteristics (CTQs). Theproblem/goal statement, the project scope, team rolesand milestones are developed. A high-level process ismapped for the existing process.Measure: The key internal processes that influence theCTQs are identified and the defects generated relative tothe identified CTQs are measured.Analyze: The objective of this phase is to understandwhy defects are generated. Brainstorming and statisticaltools are used to identify key variables (X’s) that causedefects. The output of this phase is the explanation ofthe variables that are most likely to affect processvariation.Improve: The objective of this phase is to confirm thekey variables and quantify the effect of these variableson the CTQs. It also includes identifying the maximumacceptable ranges of the key variables, validating themeasurement systems and modifying the existingprocess to stay within these ranges.
Control: The objective of this phase is to ensure that themodified process now enables the key variables to staywithin the maximum acceptable ranges, using tools likeStatistical Process Control (SPC) or simple checklists.
5. Six Sigma project on quality compliance
This section discusses a Six Sigma project on theimprovement of product quality compliance, which willgive a better understanding of the approach,methodology and benefits of Six Sigma.
5.1. Define phase
The project teams had the problem of field errorsbeing reported in the deliverables. Based on the metricsfor 1997, the long-term process capability for productquality was at 3.48σ, while the short-term capabilitywas at 4.98σ, and the Defects Per Million Opportunities(DPMO) were at 253.
A Six Sigma approach was initiated to improve thequality of deliverables. The goal of the project was toimprove the long-term process capability to more than4σ and to reduce the DPMO by more than 50%.
The project started the Define Phase with theidentification of Product Quality as the CTQ. Team
members from different levels namely Project Leaders,Module Leaders, Team Members and the Quality Teamwere identified for the Six Sigma project.
A high-level process mapping for the existingprocess was made as shown in Figure 5-2.
The next step was to define the measurement system.Any attribute in the deliverable sent to the customer,which does not meet the customer’s requirements, orwhich is not as per the Customer’s Standards wasdefined as a defect. For the purpose of calculatingDPMO, the opportunity, which is a product or process
characteristic that adds or deducts value from theproduct, was defined, based on the opportunity
definition by the client [3].
5.2. Measure phase
During this phase, the key processes in the projectlifecycle that affect the CTQ (in this case, productquality), were identified to be project study, executionand delivery. Measurements related to the CTQ weremade in these phases. The field errors reported by theclient were classified as defects that have occurred ineach of these processes as shown in Figure5-3.
Measurements in Key Processes
0
2
4
6
8
10
12
Study Execution Delivery
Key Processes
Def
ects
Figure 5-2: Measurement of defects in key processes
ProjectInitiation ONSITE
Project Study
Effort Estimation /Scheduling
CLIENT
QualityAssurance Execution
STUDY
ResourceAllocation
Delivery Feedback
Figure 5-1: High-level process mapping of the existing process
In each of these processes, the input processvariables (controllable or critical-those that show
statistical significance) that affect the CTQ wereidentified as shown in Figure 5-4.
The Input Variable — Quality of Inputs, had variousattributes namely, clarity of scope definitions,
completeness of inputs, and conformance of inputs tostandards. This variable became a critical variable thataffected the product quality, since the inputs for theprojects were obtained from the customer. The projectsthat were affected by poor quality of inputs weremeasured. As depicted in Figure 5-5, a substantialnumber of projects were affected by poor quality ofinputs in 1998.
Therefore, it was decided to analyse the effect ofquality of inputs separately. A Black Belt project named
'Analysis of Input Quality' was executed in 1998. As aresult of this project, process control in the form of aninput checklist was introduced at the project start-upphase, to prevent defects occurring in the projectdeliverables due to wrong inputs.
ProjectInitiation ONSITEProject Study
Effort Estimation /Scheduling
CLIENTQualityAssurance Execution
STUDY
ResourceAllocation
Delivery Feedback
Figure 5-3 Input process variables in the Key
05
101520253035404550
Jan Feb Mar Apr May Jun Jul Aug
Total no.of projects
Projects affected
Figure 5-4: Projects affected by Poor Input quality in 1998
The Input Variable — Quality of Inputs, had variousattributes namely, clarity of scope definitions,completeness of inputs, and conformance of inputs tostandards. This variable became a critical variable thataffected the product quality, since the inputs for theprojects were obtained from the customer. The projectsthat were affected by poor quality of inputs weremeasured. As depicted in Figure 5-5, a substantialnumber of projects were affected by poor quality ofinputs in 1998.
Therefore, it was decided to analyse the effect ofquality of inputs separately. A Black Belt project named'Analysis of Input Quality' was executed in 1998. As aresult of this project, process control in the form of aninput checklist was introduced at the project start-upphase, to prevent defects occurring in the projectdeliverables due to wrong inputs.
5.3. Analyze phase
Process performance was assessed using Cause-and-Effect diagrams, to isolate key problem areas, to studythe causes for the deviation from ideal performance, and
to identify if there is a relationship between thevariables. Extensive brain-storming sessions were heldwith team members to evolve these diagrams. Figure 5-6 shows the Cause-and-Effect diagrams for one of theproject teams.
The probable causes that can lead to quality non-conformance in a project during different phases of aproject life cycle were listed
The Failure Modes and Effects Analysis (FMEA)was subsequently carried out to arrive at a plan forprevention of causes for failure.
FMEA is a tool that helps prevent the occurrence ofproblems by:
Identifying the potential failure modes in which aprocess or product may fail to meet specifications, andrating the severity of the effect on the customer.
Providing an objective evaluation of the occurrenceof causes
Determining the ability of the current system todetect when those causes or failure modes will occur.
Based on the above factors, a Risk Priority Number(RPN) for each failure mode is calculated [3].
Using FMEA, five most severe causes that affectthe CTQs and the recommended actions were
identified. Figure 5-7 shows the results of FMEAconducted in one of the project groups.
5.4 Improve phase.
.
.
C & E D iag ram - P ip in g Q u a lity C o m p lia n c e
L a c k o fk n o w le d g e
P kg . S tu d y n o tfo llo w e d s tric t lyin p k g p la n
T ig h t s c h e du le
In itia l s tu d y o f p k gn o t d o n e
C la r if . n o tra is e d in t im eN o p r io ri ty
L a c k o f k n o w le d g e
S h ift in g p e op lea c ro s s p k g .fre q u e n tly
T ig h t s c h e du le
A llo c a tio n ba s e do n sk ill s e t
Im p rop e r p la n n in gL a c k o f c o m m n .a m o n g te a m m e m b e rs
Im p ro p e r e x e c u tio n
C h k lis ts /IQ An o t fo llo w e d
N e w ava ila b ilityo f re so u rce s
M is in te rp re ta tio no f in s tru c tion s
tig h t sc h e d u leL a c k o f k n o w le d g e
N o IQ A
R e p e tit iv ee rro rs
N o s ha r in g o fle s s o ns le a rn t
N o c a u s al a n a ly s is m e etin g
U n p a ra m e tricfe a tu re s
P o o r q u a lity o fin p u ts /ins tn s .
N e wp e o p le
L a c k o ftra in ing
N o in du c tio ntra in in g
N o p r io ri ty M is in te rp re ta tio no f in s tru c tion s
L a c k o f k n o w le d g e
N o t g iv in g re le v a n td o c . fo r E Q A
tig h t s ch e du le
N o n av a ila b ilityo f re so u rce s
N o E Q A C h an g e in G E S T D s
In fo . no t g iv e n to a ll te a m m e m be rs
N o t u pd a tin gc h e ck lis ts /g u id e lin e s
In it ia to rsp re fe re n ce s
U p d a te o fd a ta file s
Q A T oo l, P ip em a s te r
s /w to o ls n o tru n n in g o n th em a ch in e s
Q u ality -n o n c o m p lian c e
Figure 5-5 Cause & effect diagram for product quality
Based on the recommended actions from FMEA,several process improvements were introduced in thephases - Project Study, Execution and Delivery. Theseimprovements include developing process control anderror proofing tools amongst others. Figure 5-7 showssome of the improvements carried out in differentphases.
In addtion to this, a customer feedback form wasintroduced. The various quality attributes of the
deliverables are rated by the customer on a 1 to 5 scale.Causal analysis is carried out by teams when the ratingis below 4 and preventive actions is initiated.
5.5 Control phase
The process improvements that were introduced resultedin the reduction of field errors. The process capabilityfor quality of deliverables improved from 3.48σ to
Figure 5-6 Process Improvements
Project Study• Project Start-up meetings• Input Checklist• Traceability Matrix
Delivery
• Error Proofing Tools• Guidelines, Procedures
Execution• Checklists (Process / Product)• Guidelines, Procedures• Methodologies• Training• Error Proofing Tools• Standardisation - Templates• Regular Team Meetings for Defect Prevention• Log files for usage of Tools
Trend in Quality Compliance
94%
95%
96%
97%
98%
99%
100%
101%
3rd Qtr '98 4th Qtr '98 1st Qtr '99 2nd Qtr'99 3rd Qtr '99 4th Qtr '98 1st Qtr '00
Period
Qu
alit
y C
om
plia
nce
Figure 5-7: Trends in quality compliance
3.98σ in the long-term and from 4.98σ to 5.48σ in theshort-term by the end of 1998, and the DPMO reducedfrom 253 to 34. This trend continued in 1999 and 2000,resulting in the improved trend in the qualitycompliance as shown in Figure 5-8. The best practicesand lessons learnt in this Six Sigma project forengineering design were applied in other project teamsand other types of projects.
Since the field errors reduced and the processcapability for quality deliverables increased to morethan five sigma, the emphasis shifted to improvement ofin-process quality. This is to be achieved throughreduction of in-process quality cost as explained below.
The project life cycle has a phase for QualityAssurance (QA). As a continuous improvement
initiative, Green Belt projects have been initiated indifferent project teams, to reduce the rework cost afterQA. These Green Belt projects used the results of theCause and Effect, and FMEA of the Six Sigma projecton Quality Compliance.
A number of Defect Prevention practices wereidentified in these GB projects and were built into theprocesses of the project life cycle. To measure thequality cost, a metric called "rework" index was used.This metric is calculated as percentage project effortspent in rework. Control charts (I-MR) were drawn totrack the process level (process characteristic withinprojects) and process variation (process characteristicbetween projects) simultaneously, and also to detect thepresence of special causes.
Figure 5-9shows the Control charts drawn over aperiod of time for rework index for one of the projectteams. The assignable causes for variations from thetarget values were analyzed, and corrective actions wereinitiated to remove the causes and to prevent them fromoccurring again. From the control charts, it is evidentthat the process improvement initiatives have beeneffective in reducing the rework index. This is indicated
by the converging control limits in the individual's chartand the reduction in process variation as seen in themoving range chart.
Using these values, process capability studies arealso undertaken. Figure 5-10 shows the processcapability study for rework index in one of the projectteams. This study can be used to assess whether theprocess is centered on the target and whether it is
5040302010Subgroup 0
8
4
0
Ind
ivid
ual
Val
ue 1
1
Mean=0.9435
UCL=2.735
LCL=-0.8475
1 2 3
876543210
Mov
ing
Ran
ge 1
1
R=0.6734
UCL=2.200
LCL=0
1 2 3
I-MR Chart for Rework Index
April 2000 May 2000 June 2000
Figure 5-8 I-MR chart for rework index
capable of consistently meeting process specifications.For the metric — rework index, the specification limitswere set at 4% to –2% of the project effort as rework
after QA. As seen from the chart, the expected overallperformance has a DPMO of 23480, which is equal to3.47σ.
6. Conclusion
The thrust on Six Sigma Quality has helped increating and sustaining customer focus in the TCS —GEDC, leading to improved customer satisfaction asindicated in the feedback from the customer. At thesame time, active participation of the team membersfrom all levels in the Six Sigma projects has evolved aculture of effective and creative team work.
The goal is to achieve Six Sigma level not only inproduct quality, which is currently at 5.85σ, but also inthe other client specified metrics of on-time deliveryand estimate compliance. To achieve this goal, TCS —GEDC plans to have about 60 Six Sigma projectscompleted by the second quarter of 2001.
7. References
[1] Paulk M, Weber C, Curtis B, Chrissis M.B. (1995)The Capability Maturity Model: Guidelines for
Improving the Software Process. Addison WesleyPublishing Company, Reading, MA.[2] Forrest W. Breyfogle Implementing Six Sigma:Smarter Solutions Using Statistical Methods :[3]. 6σ OJT Six Sigma Application – GEPS Playbook,GE Power Systems
8. Author information
Mala Murugappan ([email protected]) is theSEPG coordinator at Offshore EngineeringDevelopment Center, Chennai, India. A CertifiedQuality Analyst, her interests include process andquality improvement and Six Sigma Quality. She hasundergone Six Sigma Training for Green/Black Beltconducted by GE, India. Mala received her Bachelor ofEngineering in Mechanical Engineering from College ofEngineering, Guindy, Chennai and Master ofTechnology in Machine Design from the Indian Instituteof Technology, Chennai.
Dr. Gargi Keeni ([email protected]) is theCorporate Quality Head at Tata Consultancy Services,
86420-2
USLLSL
Process Capability Study
PPM Total
PPM > USL
PPM < LSL
PPM Total
PPM > USL
PPM < LSL
PPM Total
PPM > USL
PPM < LSL
Ppk
PPL
PPU
Pp
Cpm
Cpk
CPL
CPU
Cp
StDev (Overall)
StDev (Within)
Sample N
Mean
LSL
Target
USL
23480.44
16267.78
7212.66
3063.80
2429.42
634.38
21276.60
21276.60
0.00
0.71
0.82
0.71
0.76
*
0.94
1.07
0.94
1.01
1.30883
0.99351
47
1.20203
-2.00000
*
4.00000
Exp. "Overall" PerformanceExp. "Within" PerformanceObserved PerformanceOverall Capability
Potential (Within) Capability
Process Data
Within
Overall
Figure 5-9 Process capability study
responsible for managing process improvementactivities. A Certified Quality Analyst and CMM LeadAssessor, she has extensive experience in softwareproject management, software tools development,managing training activities, managing human resourceallocation and managing quality processesimplementation. She has conducted several courses andworkshops on Quality Management, Software CMM,
People CMM, ISO 9000 and JRD Quality Value Award(the Malcolm Baldridge National Quality Award criteriaadapted by the Tata Group of companies). A Doctoratein Nuclear Physics from Tohoku University of Japan,she was a Research Fellow at Saha Institute of NuclearPhysics, Calcutta, and a Systems Engineer at Fujitsu,Japan. Dr. Keeni is a member of Computer Society ofIndia and IEEE.