Six Sigma Implementation to reduce rejection rate of Pump Casings at local Manufacturing Company

IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE)
e-ISSN: 2278-1684,p-ISSN: 2320-334X, Volume 7, Issue 4 (Jul. - Aug. 2013), PP 96-112
www.iosrjournals.org
www.iosrjournals.org 96 | Page
Six Sigma Implementation to reduce rejection rate of Pump
Casings at local Manufacturing Company
Faheem Yousaf¹, Dr. Shahid Ikramullah Butt¹, Dr. Riaz Ahmad¹
¹National University of Science and Technology, Islamabad, Pakistan
Abstract: Six Sigma is being Implemented all over the World as a successful Quality Improvement
Methodology. This article provides a description of Six Sigma Project implemented at Local manufacturing
Company. The Company Manufactures Pump Casings where it was receiving high nonconformance rate that
resulted in to Rejection of Product. This study deals with Six Sigma DMAIC methodology implementation and
gives a frame work of how the non-conformance rate was first monitored and then brought in to acceptance
limits. A complete Coverage of the statistical analysis performed during the study is given and results are
shown to describe that how Six Sigma helped the Project members to Improve Quality of Pump casings at
manufacturing facility.
Keywords: Statistical Process Control (SQC), Process Capability Indices, Six Sigma, Variable control chart,
Pareto charts, Standard deviation, Critical to Quality (CTQ), Analysis of Variance (ANOVA), DMAIC, Total
quality Management (TQM)
I. Introduction
In this Era of changing customer needs and demand of highly reliable products have pushed many
manufacturing companies to adopt Total Quality Management (TQM) principles. Globalization and extension
of Product Market has also increased the need of Quality Products at Reasonable cost to Customers. To respond
to these Demands many Companies are implementing different Quality Management Principles at their
manufacturing facilities such as ISO 9000, Just in Time (JIT), Lean Manufacturing, and Kaizan etc. A new and
Improved Quality Improvement Approach called Six Sigma is also becoming Popular in Controlling the Defect
rate and managing the Quality as overall Process Function.
The roots of Six Sigma as a measurement standard can be traced back to Carl Frederick Gauss (1777-
1855) who introduced the concept of the normal curve. Six Sigma as a measurement standard in product
variation can be traced back to the 1920′s when Walter Shewhart showed that three standard deviations from the
process mean is the point where a process requires correction. Many measurement standards (Zero defects etc.)
Later came on the scene but credit for coining the term “Six Sigma” goes to a Motorola engineer named Bill
Smith. (Incidentally, “Six Sigma” is a federally registered trademark of Motorola).
In the early and mid-1980s with Chairman Bob Galvin, Motorola engineers decided that the traditional
quality levels; measuring defects in thousands of opportunities didn’t provide enough granularity. Instead, they
wanted to measure the defects per million opportunities. Motorola developed this new standard and created the
methodology and needed cultural change associated with it. Since then, hundreds of companies around the
world have adopted Six Sigma as a way of doing business.
II. Overview of Statistical Quality Control (SQC)
Quality may be defined as that characteristic which Renders a product or service as having “fitness for
Purpose or use”. There are different reasons why a Product may have poor quality. Statistical Methods play a
central role in Quality improvement Efforts and recognized as an efficient and powerful tool in dealing with the
process control aspects (Montgomery 2003).
Statistical Quality Control (SQC) is the term used to describe the set of statistical tools used by quality
professionals. SQC is used to analyze the quality problems and solve them. Statistical quality control refers to
the use of statistical methods in monitoring and maintaining of the quality of products and services. All the tools
of SQC are helpful in evaluating the quality of services. SQC uses different tools to analyze quality problems.
Descriptive Statistics, Statistical Process Control (SPC) and Acceptance Sampling are the major SQC
Techniques used in Improving Quality of Processes.
2.1. Concept of Variation
Variation is part and parcel of life. The concept of Variation states that no two products will be
perfectly identical even if extreme care is taken to make them Identical in some aspect. The variation in the
quality of Product in any manufacturing process results because of two reasons namely, Chance cause and

Six Sigma Implementation to reduce rejection rate of Pump Casings at local Manufacturing Company
Assignable Cause. A process that is operating with only chance causes of variation is said to be in a state of
control with minor variations in the process. The major objective of SPC is to quickly detect the occurrence of
assignable causes so that investigation of process and corrective action may be taken before many non-
conforming units are manufactured. Finally the eventual goal of SPC is the elimination of variability in the
processes.
2.2. Process Capability
The Process Capability studies are helpful in analyzing the quality and efficiency of the process. The
Process Capability analysis has been widely adopted as the ultimate measure of Performance to evaluate the
ability of a process to satisfy the customers in the form of specifications (English et al 1993).
The Process Capability study helps in designing the product, deciding the acceptance limits of the
process and operators in the operations management (Feigenbaum 1994). The evaluation of process capability is
an important step in process quality improvement (Juran 1991). It is customary to take the Six Sigma Prime
spread in the distribution of the product Quality Characteristic as a measure of process capability. In Process
Capability study of particular process, Six Sigma Prime spread is compared with the difference of Upper
Specification Limit (USL) and the Lower Specification Limit (LSL). The following are the three possible cases.
1. 6s ` > (USL – LSL): In this case, the process spread is greater than the tolerance. So the process is incapable
of meeting the Specification.
2. 6s `= (USL – LSL): In this case, the process spread is exactly equal to the tolerance. So the process is exactly
capable of meeting the specifications.
3. 6s ` < (USL – LSL): In this case, the process spread is less than the tolerance. So the process is capable of
meeting the specifications.
2.3. Process Capability Indices
It is frequently convenient for Engineers to have a simple and quantitative way to express process
capability. One way to do so is through process capability indices. Process capability indices (PCI) are powerful
means of studying the process ability for manufacturing a product that meets specifications (Chen et al 2001).
PCI is defined as the ratio of tolerances to the process spread. Equation (1) describes the way to calculate PCI.
Process Capability Index (PCI) = (USL – LSL)/6s (1)
Where PCI is process Capability indices, USL is upper specification limit, LSL is lower specification
limit and 6s is the six sigma abbreviation.
2.4. Six Sigma
It is the set of practices originally developed by Motorola to systematically improve process by
eliminating defects. Defect is defined as non-conformity of a product or service to its specification. Like its
previous quality improving methodologies six sigma focuses on the following points.
 A continuous effort to reduce variation in process outputs is essential to business success.
 Manufacturing and business processes can be Measured, Analyzed, Improved and Controlled.
 In order to achieve best Quality Improvement results, role of upper management is very critical.
The term Six Sigma refers to a highly capable process that can produce products within specifications.
Process that achieves Six Sigma levels produces only 3.4 defective Products per million opportunities. Main
focus of Six Sigma is to improve all key processes of manufacturing setup and takes quality as a function of
Processes Capability to produce items with in specification.
Six Sigma mainly uses two main methodologies one is called Define, Measure, Analyze, Improve and
Control, usually known as DMAIC and other is Define, Measure, Analyze, Design and Verify, known as
DMADV. Both the methodologies are based on Edwards Deming’s, Plan- Do-Check-Act cycle.
DMADV is used for creating new processes to produce products with minimum defect rate.
Other methodologies that are being used during six sigma implementation are given as.
 DCCDI (Define, Customer, Concept, Design And Implement)
 CDOC (Conceptualize, Design, Optimize and Control)
 DCDOV (Define, Concept, Design, Optimize and Verify)
 DMADOV (Define, Measure, Analyze, Design, Optimize and Verify)
 DMEDI (Define, Measure, Explore, Develop and Implement)
 IDOV (Identify, Design, Optimize and validate)
 IIDOV (Invent, Innovate, Develop, Optimize and validate)

DMAIC methodology has been used as effective methodology in before mentioned organization;
details of this methodology are given as.
2.5. Define, Measure, Analyze, Improve and Control (DMAIC)
Basically this methodology comprises of following five key points.
 Define the process improvement goals that are aligned with the customer demands and company’s strategy.
 Measure the current process and make a strategy for making further improvement.
 Analyze to verify the relationship and causality of factors. Determine what the relationship is and attempts
to ensure that all the factors have been considered.
 Improve and optimize process based on findings of analysis phase using different techniques.
 Control to ensure that any variances are corrected before they result in defects.
2.6. DMAIC Implementation at Pump Casing manufacturing company
DMAIC is chosen as potential methodology for implementation at pump casings manufacturing
company. At this company quality management systems like ISO 9000 are already implemented that’s why
quality improvement focus is to only reduce nonconformities. Figure 1 describes the five phases of DMIAC that
are implemented at manufacturing facility with different statistical techniques.
Define Phase
Problem description, project goal and
charter, communication plan, process
flow chart, SIPOC, Defining customer
and process requirements, Pareto chart,
Voice of customer (VOC), Critical to
quality tree (CTQ)
Measure Phase
X bar and R chart to analyze process
before improvement, calculating process
mean, standard deviation, cumulative
distribution function and sigma
capability, other analysis include Cause
and effect matrix for rotor casing
assembly, Measurement system
Analysis (MSA), Gage R&R study and
its Analysis of Variance
Analyze Phase One Sample Z test for average cylinder
inner diameter, Multi Variant Analysis,
Analysis of Variance (ANOVA)
Improve Phase
Multi-level factorial design, Main effect
and interaction effect analysis of factors,
Regression Analysis, Analysis of effect
Achieved after improvement, One
sample z-test and two sample
T-test Analysis
Control Phase Potential problem Analysis, Overall
Control Plan, Sample process control
plan, Project Conclusion
Figure 1. DMAIC Flow Chart
III. Define Phase
It is the first stage of Six Sigma DMAIC project, the phase determines the objectives and scope of the
project and it defines the problem & communicates the problem to others.
3.1. Problem Description
Company ABC is a manufacturing organization that produces pump casings and different sub-
assemblies related to Pump. For the few weeks company is receiving high rejection rate in pump casings, this
results in high rework rate and due to which higher production time is spent and also credibility of the company
and profits are being affected.
Approximately 25% pump casings produced by manufacturing facility are either required to be
reworked or rejected at final inspection (Rework and rejection are both considered as defects), this was found
when the last two lots were sampled for inspection.

3.2. Project Goal and Charter
Management of the company wanted a detail process study of different sections of manufacturing
Facility. For this purpose a project team was made whose primary responsibility was to identify the key factors
playing role in rejection and rework and then to take certain improvement measures to reduce the rejection rate.
Table 1 clearly defines the Project Team Charter. Clearly the project goal is to reduce Current rejection rate
from 25% to 1% (acceptance limit).
Table 1. Project Team Description
3.3. Communication Plan for DMAIC Project
Table 2 shows the complete communication plan for Project.
Table 2. Communication Plan for Six Sigma Project
3.4. SIPOC for Pump Casings
Supplier, Inputs, Processes, Outputs, Customers i.e. SIPOC is a process improvement and analysis tool
that summarizes the input and output of one or more processes. This tool helps to sum up the inputs and outputs
of a certain process according to the customer point of view that helps to analyze the requirements of each
process in one table format.
For the current facility the SIPOC is shown in Table 3. From the current system study it was revealed
that Pump casings key supplier is material store accompanied with manufacturing shops. Inputs to the process
are manufacturing process parameters and piece of bars in rectangular and cylindrical form. Complete process
details and outputs along with customers are shown in Table 3.
Project Team Charter
Team Members Status Involvement in Project
Sponsor and Champion Define, Project Charter,
Project Closure
Black Belt Project Implementation
Production Department Improve and Control
Quality control Head Data collection and Analysis
Project Title: Implementation of Six Sigma to reduce rejection
rate of Pump Casings.
Subject Matter Experts: Black Belt, Champion Six Sigma.
Stake Holders: Employees of Company, Production
Department, and quality control Department.
Project Milestones
S.NO. PHASE
1 Define
2 Measure
3 Analyze
4 Improve
5 Control
What To
whom
When Who How Where
Project
meetings
Project
team
invitees
Every
Thursday
Black
belt
Via
mail
Conference
room
Minutes of
meetings
Distribu
tion list
Next day of
meeting
Black
belt
Via
mail
Office of black
belt
Team work Project
team
As per
requiremen
t
Black
Belt
Via
mail
Office of black
belt
Status reports Custom
er
Every
Friday
Black
Belt
Via
mail
Office of black
belt

Table 3. SIPOC for Pump Casing
3.5. Process flow chart for Pump casings
Complete flow chart of the manufacturing process at manufacturing facility of pump casings was
developed and shown in Figure 2. This chart provides good understanding of process and material flow.
Figure 2. Pump Casing’s Process Flow Chart
3.6. Defining Process boundaries and customer Rrequirements
Main manufacturing processes are setup to process three items named as cover, cylinder and pads.
Complete flow diagram is illustrated in Figure 3. According to detailed study of the processes it was revealed
that chamber inner diameter, nozzle throat diameter and some other defects were contributing to high
nonconformance rate . Figure 3 shows a Pareto chart which summarizes the key defects in process outputs, this
show that chamber Inner Diameter accounts for 71.4% of total defects and comes to be potential Key Process
Input Variables (KPIV) to be addressed.
Suppliers Inputs Processes Outputs Customers
Material
store
Welding
section,
Milling
section,
Turning
section,
Drilling
section
Piece of round/
rectangular bar,
Tool angle,
Tool,
Feed,
Depth of cut,
Coolant flow
Facing,
Turning,
Taper turning,
Boring,
Threading,
Milling,
Drilling,
Welding,
Assembly.
Rotor cylinder,
Pads with holes,
Padded Rotor
cylinder,
Nozzle end,
Front Stopper,
Chuck
Ring,
Hexagonal
section of nozzle
end,
Assembled rotor
casings
Inspection and
quality
department,
Assembly
section,
Manufacturing
shops,
End users

Figure 3. Pareto chart for defects monitoring
Customer data i.e. Voice of Customer (VOC) revealed that internal customers are being affected by the
defects rate of Chamber inner diameter that accounts for major portion of defects as explained in Figure 3. Thus
Critical To Quality (CTQ) tree can be established based on voice of customer and project objectives. Figure 4
shows the CTQ tree based on the voice of customer
Figure 4. Critical To Quality (CTQ) Tree
IV. Measure Phase
The primary requirement of this phase is the complete process map of manufacturing process that is
given in Figure 2 of define phase. Measure Phase focuses on the following points about Six Sigma Project.
4.1. Determines What to Measure
The Six Sigma project team needs to come out with an explicit list of vital inputs, desired outputs and
process metrics that they are planning to track. This decision has large scale implications on the performance of
the project and is usually taken by the Project Leader on the basis of data provided by the Six Sigma team and
the Process Owner.
4.1.1. Determine How to Measure
There have been many cases of measurement bias in Six Sigma history. The bias may have its roots in
the complexity of the calculation, the wrong method of data collection or the bias of the person performing the
exercise.
4.1.2. Manage the Measurement Process
One of the biggest challenges of this phase is to validate the fact that the measurement system in place
is good for the purpose. This challenge has been discussed in detail in the section titled Measurement Systems
Analysis.

4.1.3. Calculating the Current Sigma Levels
A common practice in the measure phase is to put a figure in front of the losses that the organization is
currently facing due to inefficient processes. This helps the management to evaluate the process and the Six
Sigma team can bargain for more resources to successfully implement the project.
4.2. Current Base line for the Process
Mean of cylinder inner diameter = 50.4512mm
Standard deviation = 0.0708445mm
Upper specification limit = 50.6 mm
lower specification limit = 50.4 mm
4.2.1 Cumulative Ddistribution Function of sample taken
Normal with mean =50.4512mm and standard deviation = 0.0708445mm.
Cumulative Distribution Function
x P(X<=x)
50.4 0.234929
50.6 0.982152
4.2.2. Defective parts per million opportunities (DPPMO)
Equation 2 gives us the formula to calculate defective parts per million opportunities.
DPPMO = (% of parts less than 50.4) + (%of parts greater than 50.6)/1, 000000 (2)
Using Equation 2 the defect percentage is calculated as.
= (0.234929) + (1-0.982152)/1000000
= 252,777
= 25.27%
4.2.3. Inverse Cumulative Distribution function
Normal with mean = 0 and standard deviation = 1
Inverse Cumulative Distribution Function
P(X<=x) x
0.252777 -0.665776
Measure of distance from the mean of sample (Z-score) is calculated to be = 0.665776
Process Capability index (Cpk) is calculated to be
= 0.665776/3
= 0.2219
Sigma capability = Z-score + 1.5 (3)
Using equation 3 sigma capability for current process before improvement is calculated to be = 2.16
4.3. Cause and Effect matrix for Pump casing Assembly
Cause and effect matrix for pump casings was made in detail and relationship between process inputs
and outputs is described. From Table 4 it is clear that depth of cut, feed rate and tool angle are the important
factors affecting the quality of pump casings.

Table 4. Cause and Effect Matrix (C&E)
Where A= cylinder length, B= cylinder inner diameter, C = cylinder wall thickness, D = cylinder front prepped,
E = cylinder rear prepped, F = Nozzle throat Diameter, G = D-Nozzle Outer diameter, H = D-Nozzle angle.
4.4. Measurement System Analysis [MSA]
The next step in measurement phase is measurement system analysis. Basic measurement tool is
Venire caliper. Two measurement analysis techniques named Gage Linearity and bias Study and Gage R and r
study were used and details are given.
4.4.1. Control Charts
In statistical process control XBar and R charts are type of control charts that are used to monitor
samples data when data is collected at different intervals. Usually standard Deviations are monitored by Range
chart commonly known as R chart and sample mean is monitored by X-bar chart.
In Figure 5 the XBar-R chart shows that process is out of control with maximum observations out of
limits. Also a certain pattern is followed by the data points which may be due to machine error or operators shift
change. Reasons of out of Control manufacturing process are given in Analyze Phase of the Project.
Rating of
Iimportance to
Customers
7 8 8 7 7 9 9 9
S/N 1 2 3 4 5 6 7 8
Process output
Process
input
A B C D E F G H Total
Depth of cut 9 9 9 9 9 9 9 9 576
Feed 0 0 3 3 3 3 3 3 147
Tool Angle 1 3 1 3 3 3 1 3 144
Tool 1 3 1 1 1 3 1 1 98
Speed 1 1 1 3 3 1 1 1 92
Coolant Flow 0 1 1 1 1 3 1 1 75
Material Type 1 1 1 1 1 1 1 1 64
Figure 5. Xbar-R chart of cylinder inner Diameter before improvement

Table 5. Gage Linearity Analysis
Predictor Coefficient SE Coefficient P
Constant 0.01434 0.01442 0.324
Slope 0.0000145 0.0001223 0.906
S 0.070711 R-sq 0.0%
Linearity 0.0000051 %Linearity 0.0
Table 6. Gage Biasness Analysis
Reference Bias %Bias P
Average 0.0156667 4.4 0.017
1.5 -0.031666 8.9 0.445
25 0.0675000 19.1 0.004
80 0.0125000 3.5 0.006
150 0.0216667 6.1 0.012
200 0.0083333 2.4 0.001
Results shown in Table 5 and 6 provides enough evidence that bias of the instrument is constant over
its measuring range and the gage measures constantly regardless of the part size.
4.5. Gage Repeatability and Reproduce ability (R&R) Study-ANOVA Method
Another study of the measurement system was made by analysis of variance of gage. Complete data of
the analysis is Table 7 and 8.
Table 7. Two-Way ANOVA Table with Interaction
Source DF SS MS F P
Part ID 9 1.4244 0.158271 2347.9 0.00
Operator 1 0.00003 0.0000 0.4 0.54
Part ID*Operator 9 0.00061 0.000067 2.25 0.03
Repeatability 40 0.0012 0.000030
Total 59 1.42627
Table 8. Gage R&R Analysis
Source VarComp %Contribution (Varcomp)
Total Gage R&R 0.0000425 0.16
Repeatability 0.0000300 0.11
Reproducibility 0.0000125 0.05
Operator 0.000000 0.00
Operator*Part ID 0.0000125 0.05
Part-To-Part 0.0263673 99.84
Total Variation 0.0264098 100.0
Source St Dev Study %study Tolerance
Variation Variation
Total Gage R&R 0.006517 0.033562 4.01 6.71
Repeatability S 0.005477 0.028208 3.37 5.64
Reproducibility 0.003531 0.018186 2.17 3.64
Operator 0.00000 0.00000 0.0 0.00
Operator*Part ID 0.003531 0.018186 2.17 3.64
Part-To-Part 0.16238 0.836257 99.92 167.25
Total Variation 0.162511 0.836931 100.0 167.39
No. of Distinct Categories = 35

Figure 6. Gage R&R (ANOVA) for Measurement
By the complete ANOVA study and Gages run chart and R&R chart analysis shown in Figure 6 it is
clear that percentage contribution of repeatability and reproducibility is very minimum and these are not making
significant impact in variation data. Main cause of variation is found to be part to part variation.
Figure 7. Multi-Vari Chart for Cylinder Inner Diameter by Depth of Cut-Tool Angle

By Table 8 it is clear that Measurement Gage
%contribution is 0.16<1,
%study variation = 4.01 < 10%,
%Tolerance = 6.71 < 10%,
%Distinct categories = 35>10
Hence the measurement system is acceptable and reliable for measurement purpose.
V. Analyze Phase
This phase focuses on analyzing the cause of defects defined in the before mentioned project phases.
The first analysis was made on cylinder inner diameter. Table 9 shows One sample z test performed to check
whether average inner cylinder diameter was 50.4512 or not.
Table 9. One-sample Z-Test of Cylinder inner diameter.
Factor Type Levels Values
Depth random 3 1.0, 1.5, 2.0
Of cut
Feed random 3 8,9,10
Tool angle random 3 25, 30, 35
Source DF SS MS F P
Depth of cut 2 0.015528 0.0077 14.23 0.00
Feed 2 0.005140 0.0002 0.470 0.62
Tool angle 2 0.162343 0.08117 148.75 0.00
Error 74 0.040380 0.00054
Total 80 0.287650
S = 0.0233598 R-sq. = 81.54 % R-sq. (adj.) = 80.05%
As given in Table 9 the P-Value of 1 is greater than alpha = 0.05, the decision is not to reject null
hypothesis. There is strong evidence that the average cylinder inner diameter is 50.4512 mm. Next in the
analyze phase is Multi-vari Analysis that was performed with three variables i.e. depth of cut with three settings
1.0,1.5,2.and tool angle with settings 25,30,35 degrees and feed rate with settings 8,9 and 10mm per minute.
Figure 7 shows the Analysis. The settings of variable that gave best responses are
 Depth of cut = 1 mm
 Feed = 8 mm/min
 Tool angle = 35 degrees
Next in the analyze phase is ANOVA performed on cylinder inner Diameter versus Depth of cut, Feed,
Tool Angle. Exhibit Table 10 shows complete Analysis data.
Table 10. Analysis of Variance of Cylinder Inner Diameter
Factors 2 Replicates: 3
Base runs 9 total runs 27
Base blocks1 Total Blocks 1
Number of levels: 3, 3
Depth of Cut fixed 3 1.0, 1.5, 2.0
Tool angle fixed 3 25, 30, 35
Source DF AdjSS Adj MS F P
Depth of cut 2 0.01111 0.00555 11.2 0.001
Tool angle 2 0.073563 0.036781 74.11 0.626
Tool angle* 4 0.010148 0.000253 5.11 0.006
Depth of cut

Error 18 0.008933 0.000496
Total 26 0.103763
S = 0.0222777 R-sq. = 91.39% R-sq. (adj) = 87.56%
From data given in Table 10 it is clear that depth of cut and tool angle has significant effect over the
response variable whereas feed has no significant effect. Hence these variables were considered in the improve
phase for improving the response variables.
VI. Improve Phase
Key findings from the analyze phase are that depth of cut and tool angle have the significant effect on
the cylinder inner diameter. This was further analyzed by Multilevel factorial design in Table 11 in which
cylinder inner diameter was tested against different levels of tool angle and depth of cut data obtained is given
in Table 11.
Table 11. Analysis of variance of Cylinder Inner
Diameter using adjusted SS for tests
Factors 2 Replicates: 3
Base runs 9 total runs 27
Base blocks1 Total Blocks 1
Number of levels: 3, 3
Depth of Cut fixed 3 1.0, 1.5, 2.0
Tool angle fixed 3 25, 30, 35
Source DF AdjSS Adj MS F P
Depth of cut 2 0.01111 0.00555 11.2 0.001
Tool angle 2 0.073563 0.036781 74.11 0.626
Tool angle* 4 0.010148 0.000253 5.11 0.006
Depth of cut
Error 18 0.008933 0.000496
Total 26 0.103763
S = 0.0222777 R-sq. = 91.39% R-sq. (adj) = 87.56%
Figure 8. Main Effect Plots for Cylinder Inner Diameter

Figure 9. Interaction Plot (Data Means) for Cylinder Inner Diameter
Figure 8 and 9 show the main effect and interaction effect of depth of cut and tool angles. Form plots it
is clear that depth of cut, and the interaction have significant effect on the response variable best results are
obtained for the settings:
 Depth of cut = 1.5 mm
 Tool angle= 30 degrees
Regression analysis was also made to get the correlation between different response variables. The
regression equation is given by equation 4.
Cylinder inner diameter = 50.1 + 0.0058* depth of cut + 0.126 *Tool angle (4)
Table 12. Multiple Regression Analysis
Predictor Coef SE Coef T P
Constant 50.0974 0.042 1192.23 0.00
Depth of cut 0.00583 0.01244 0.47 0.642
Tool angle 0.01258 0.001244 10.12 0.000
S = 0.0304622 R-sq. = 75.7% R-sq. (adj) = 74.2%
Table 13. Analysis of variance (ANOVA)
Source DF SS MS F P
Regression 2 0.095528 0.047764 51.3 0.000
Residual 33 0.030622 0.000957
Error
Total 35 0.125831
Using the multiple regression analysis in Table 12 and 13 it is clear that tool angle has significant
effect on response variable whereas depth of cut has also an effect but not to significant. This is further
confirmed by the best subset regression considering the maximum r-sq and minimum mallows c-p criteria, tool
angle has the most effect and the interaction has the significant effect on the second number. The settings
obtained from the analyze phase were implemented and data was monitored for some time.
6.1. Analysis of the effect achieved after improvement
Mean of cylinder inner diameter = 50.4972mm
Standard deviation of cylinder inner diameter= 0.032479mm

6.1.1. Cumulative distribution function of sample taken
Normal with mean =50.4972mm
Sstandard deviation = 0.0324795mm
Cumulative distribution function
x P(X<=x)
50.4 0.001383
50.6 0.999225
6.1.2. Defective parts per million opportunities (DPPMO)
Using formula given in equation 2, DPPMO is given as
= (0.001383) + (1-0.999225)/1000000
= 2,158
= 2.158%
6.1.3. Inverse Cumulative Distribution function
Normal with mean = 0mm and standard deviation = 1mm
P(X<=x) x
0.002158 -2.85409
Measure of distance from the mean of sample (Z-score) is calculated to be Z-score = 2.85409
Process Capability index (Cpk) is calculated to be
Cpk = 2.85409/3 = 0.951363
Using equation 3, Sigma capability for current process before improvement is calculated to be
Sigma capability = Z-score + 1.5
= 2.85409 +1.5
= 4.354
6.2. Ensuring the sampling errors with test of Hypothesis
To ensure that average cylinder inner diameter is 50.4972 or not a one sample Z test was performed.
Exhibit 23 shows findings as follows:
 Ho: µ=50.4972 (claim) and
 H1: µ≠50.4972
Table 14. One-sample Z: Cylinder Iinner diameter
Test of mu = 50.4972 vs. not = 50.4972
The assumed standard deviation = 0.0324795
Variable N mean stdev SE Mean 95%CI Z P
Cylinder 36 50.49 0.03 0.0054 (50.48, 0.0 0.9
inner D 50.50)
As given in Table 14 the P-value of 0.997 is greater than α = 0.05 the decision here is not to reject the
null hypothesis. there is enough evidence to support the claim that the average cylinder inner diameter is
50.4972.
6.3. Testing difference between two Means (Large independent systems)
 Ho: µ1 = µ2 and
 H1: µ1 ≠ µ2 (claim)
6.3.1. Two sample T-Test and CI: Cylinder inner diameter
Table 15 shows test results.

Table 15. Two sample T-Test
N Mean StDev SEMean
Cylinder Inner D 50 50.4512 0.0708 0.010
Cylinder Inner D 36 50.4972 0.0325 0.0054
Difference =
mu (cylinder inner Dia. Before Imp) – (Cylinder Inner Dia. After Imp)
Estimate of difference : -0.046022
95% CI for difference : (-0.068718, -0.023326)
T-Test of difference = 0 (vs. not =): T-Value = -4.04
P-Value = 0.000, DF=73.
In Table 15 the P value of 0.00 is less than alpha=0.05 so the decision is to reject the null hypothesis.
There is enough evidence to support the claim that the two means are not equal. There is significant difference
in cylinder inner diameter before and after process improvement.
VII. Control Phase
Tables 16 and 17 show the overall control plan for the Project. Overall control plan includes three
phases. First one is solution implementation Phase, in this phase settings of DMAIC Implementation phase were
carried and observed for 2 months and were communicated to Project sponsor, process owner and Quality
control head. Main objective of this phase is to monitor the process and decide whether the process is settled
down and has absorbed the changes by giving the required results or it does not continue to give the improved
results because of some other factors setting’s that might be overlooked in the Improve Phase.
Next is training Phase which includes training of working staff as well as process owners and Quality
Head to make them able to provide improved results if working conditions are slightly changed e.g. induction of
new Machine in Assembly line or workers shift changes etc. Cost and benefits analysis is the last and project
conclusion Phase, that was done after solution was Implemented and observed for 2 months.
Process Control plan is given in Table 17. It includes control of Pump Casings Inner diameter boring Control.
Machine used is turning and 100% inspection was carried on to observe this Process In detail. XBar
and R charts were used to observe processes deviation and Mean shifts.
By observing XBar-R chart in Figure 10 it is clear that the process is well in control i.e. the variation
with in groups is controlled. There are no points in X-bar chart that are more than 3 standard deviation away
from Process Mean, also R chart shows that variation of process is well under control.
Table 16. Overall control plan
S/No. Description Communicated to Approximate duration
1. Solution
implementation
schedule
Sponsor,
process owner,
Qc head
02 months
2. Training Process owner, Qc head 02 weeks
3. Cost and Benefits
Analysis
Sponsor Was carried Out after Implementing the solution and
running the process for 2 months
Table 17. Sample process control plan
Process/
operation
description
Machine
Jigs, tools,
Fixtures for
manufacturing
Methods
Category Number Spec.
limit
Sample
size
Control
method
Inner
Dia.
Boring
Turning Jig T/jig/
002
100%
inspect
ion
5 X-bar,
R-charts

Fig.10. XBar-R chart of cylinder inner Diameter after improvement
VIII. Conclusion
Local pump casing manufacturing facility was receiving high defect rate due to which quality of the
production lots was being effected and customer complaints had to be addressed. This was turning in to high
financial loses to the company. An effective quality improvement technique called Six Sigma was decided to be
implemented in the facility.
At this stage the defect rate was 25.27% with sigma capability of 2.16. Pareto charts defined that
chamber inner diameter defects were more than 71.4% of the total defects and was potential KPIV to be
addressed. Further analyses cleared that depth of cut; feed rate and tool angle were contributing in defect rate.
After carrying out detailed analysis of the process ,tool settings came out as main reason for this defect rate.
feed rate had not significant effect as compared to depth of cut and tool angle. Best results were achieved at:
 Feed rate = 8mm/min
 Depth of cut = 1.5 mm
 Tool angle = 30 degrees
Some jigs and fixture were required to achieve the above settings during production of rotor casings.
These were designed and manufactured and then were Incorporated to reduce setups times and eliminate
adjustments.
After getting the right settings from the analyze phase these settings were implemented and results
achieved were in favor of good quality of products. Further results were monitored and controlled and finally
defect rate observed is 2.158% with sigma capability of 4.354.
References
[1]. Douglas, C. M. (2003). Introduction to Statistical Quality Control. New York, NY: John Wiley Publications.
[2]. Mahesh, B. P., Prabhuswamy, M. S. (2010). Process Variability Reduction through Statistical Process Control for Quality
Improvement. International Journal for Quality research, 4(3), 1621-1635.
[3]. James, M. J., Mary, E. J. (2006). Lean and Six Sigma. LABMEDICINE, 37(3), 407-408.
[4]. John, J. F., Clark, W. A., San, C. A. (2006). Selecting Optimal Specification Limits. Quality Technology & Quantitative
Management, 3(2), 207-216.
[5]. Pleck, A., Vjica, H. (2009). An Application of six sigma in manufacturing company. Advances in production engineering and
Management, 4, 243-254.
[6]. John, J. F. (2009). A unifying process capability metric. JIEM, 2(1), 48-59.
[7]. Mahesh, B. P., Prabhuswamy, M. S. (2010). Improvement of Quality Awareness using Six Sigma methodology for achieving
higher CMMI Level. International journal of advance research in management, 1(1), 20-41.
[8]. Alessandro, L., Jiju, A., Alex, D. (2010). Lean six sigma in a call center: a case study. International Journal of Productivity and
Performance Management, 59(8), 757-768.
[9]. Richard, J. L., Fugee, T., Lau, W. C. (2006). Supplier Selection Based on Process Capability and Price. Quality Engineering, 18,
123–129.
[10]. Charles, R. Theodore, T. A. (2003). An Alternative Desirability Function for Achieving Six Sigma Quality. Quality and Reliability
Engineering, 19, 227–240.
[11]. Chen, K. S., Huang, M. L., Li, R. K. (2001). Process capability analysis for an entire product. International Journal of Production
Research, 39(17), 407-408.
[12]. Van, D. H., Edwin, R., Ion, S., Roxana, A. (2003). Control chart for Markov Dependent Sample Size. Quality Engineering, 12(4),
593-601.
[13]. Jeroen, D. M., Kit, C. B., Ronald, J. M. (2001). The multi vary chart: A systematic approach. Quality Engineering, 13(3), 437-439.

Author’s Biographies
Faheem Yousaf is Masters Student at National University of Science and Technology in the School of
Mechanical and Manufacturing Engineering. He received his Bachelor’s degree in Industrial Engineering from
University of Engineering and Technology Pakistan. His research interests include Quality Management,
Operation Management and Manufacturing Engineering.
Shahid Ikramullah Butt, PhD, is Head of Mechanical Engineering at National University of Science
and Technology in the School of Mechanical and Manufacturing Engineering. He has fifteen years of Industrial
and Teaching Experience with different Organizations. Dr. Shahid received his Masters degree in Industrial
Engineering from Mississippi State University and later on got his PhD. in industrial and manufacturing
engineering from Beijing Institute of technology, China.
Riaz Ahmad, PhD, is acting as director research at National University of Science and Technology.
He has blend of industry and academic experience spanning over 27 years. Besides being at senior management
posts at university, he is supervising research and teaching undergraduate and post graduate courses at
university. He has published number of research papers in international journals and conferences in field of
Mechanical Engineering.
Contact: Faheem Yousaf, School Of Mechanical and Manufacturing Engineering, National University
of Science and Technology, Islamabad, Pakistan, Engr_faheem37@yahoomail.com.

Six Sigma Implementation to reduce rejection rate of Pump Casings at local Manufacturing Company

More Related Content

Viewers also liked (20)

Similar to Six Sigma Implementation to reduce rejection rate of Pump Casings at local Manufacturing Company (20)

More from IOSR Journals (20)

Recently uploaded (20)

Six Sigma Implementation to reduce rejection rate of Pump Casings at local Manufacturing Company