# quality management systems and statistical quality control

of 37/37
Quality Management Systems and Statistical Quality Control 2 Contents 2.1 Introduction to Quality Management Systems . . . 17 2.2 International Standards and Speciﬁcations ...... 19 2.3 ISO Standards for Quality Management and Assessment .............................. 19 2.3.1 Quality Audit, Conformity, and Certiﬁcation 19 2.3.2 Environmental Standards ................. 21 2.4 Introduction to Statistical Methods for Quality Control ........................... 23 2.4.1 The Central Limit Theorem ............... 23 2.4.2 Terms and Deﬁnition in Statistical Quality Control ................................ 24 2.5 Histograms .................................. 25 2.6 Control Charts ............................... 25 2.7 Control Charts for Means ..................... 26 2.7.1 The R-Chart ............................ 26 2.7.2 Numerical Example, R-Chart ............. 29 2.7.3 The N x-Chart ........................... 29 2.7.4 Numerical Example, N x-Chart ............. 30 2.7.5 The s-Chart ............................ 30 2.7.6 Numerical Example, s-Chart and N x-Chart . . . 33 2.8 Control Charts for Attribute Data .............. 33 2.8.1 The p-Chart ............................ 35 2.8.2 Numerical Example, p-Chart .............. 36 2.8.3 The np-Chart ........................... 37 2.8.4 Numerical Example, np-Chart ............ 37 2.8.5 The c-Chart ............................ 37 2.8.6 Numerical Example, c-Chart .............. 39 2.8.7 The u-Chart ............................ 40 2.8.8 Numerical Example, u-Chart .............. 40 2.9 Capability Analysis ........................... 40 2.9.1 Numerical Example, Capability Analysis and Normal Probability .................. 42 2.9.2 Numerical Examples, Capability Analysis and Nonnormal Probability ............... 46 2.10 Six Sigma ................................... 48 2.10.1 Numerical Examples .................... 51 2.10.2 Six Sigma in the Service Sector. Thermal Water Treatments for Health and Fitness .... 51 Organizations depend on their customers and there- fore should understand current and future customer needs, should meet customer requirements and strive to exceed customer expectations... Identifying, un- derstanding and managing interrelated processes as a system contributes to the organization’s effective- ness and efﬁciency in achieving its objectives (EN ISO 9000:2006 Quality management systems – fun- damentals and vocabulary). Nowadays, user and consumer assume their own choices regarding very important competitive factors such as quality of product, production process, and production system. Users and consumers start making their choices when they feel they are able to value and compare ﬁrms with high quality standards by them- selves. This chapter introduces the reader to the main prob- lems concerning management and control of a qual- ity system and also the main supporting decision mea- sures and tools for so-called statistical quality control (SQC) and Six Sigma. 2.1 Introduction to Quality Management Systems The standard EN ISO 8402:1995, replaced by EN ISO 9000:2005, deﬁnes “quality” as “the totality of characteristics of an entity that bear on its ability to satisfy stated and implied needs,” and “product” as R. Manzini, A. Regattieri, H. Pham, E. Ferrari, Maintenance for Industrial Systems 17 © Springer 2010

Post on 14-Feb-2017

222 views

Category:

## Documents

Embed Size (px)

TRANSCRIPT

• Quality Management Systemsand Statistical Quality Control 2

Contents

2.1 Introduction to Quality Management Systems . . . 17

2.2 International Standards and Specifications . . . . . . 19

2.3 ISO Standards for Quality Managementand Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.3.1 Quality Audit, Conformity, and Certification 192.3.2 Environmental Standards . . . . . . . . . . . . . . . . . 21

2.4 Introduction to Statistical Methodsfor Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.4.1 The Central Limit Theorem . . . . . . . . . . . . . . . 232.4.2 Terms and Definition in Statistical Quality

Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.5 Histograms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.6 Control Charts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.7 Control Charts for Means . . . . . . . . . . . . . . . . . . . . . 262.7.1 The R-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.7.2 Numerical Example, R-Chart . . . . . . . . . . . . . 292.7.3 The Nx-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.7.4 Numerical Example, Nx-Chart . . . . . . . . . . . . . 302.7.5 The s-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.7.6 Numerical Example, s-Chart and Nx-Chart . . . 33

2.8 Control Charts for Attribute Data . . . . . . . . . . . . . . 332.8.1 The p-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.8.2 Numerical Example, p-Chart . . . . . . . . . . . . . . 362.8.3 The np-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . 372.8.4 Numerical Example, np-Chart . . . . . . . . . . . . 372.8.5 The c-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372.8.6 Numerical Example, c-Chart . . . . . . . . . . . . . . 392.8.7 The u-Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.8.8 Numerical Example, u-Chart . . . . . . . . . . . . . . 40

2.9 Capability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.9.1 Numerical Example, Capability Analysis

and Normal Probability . . . . . . . . . . . . . . . . . . 422.9.2 Numerical Examples, Capability Analysis

and Nonnormal Probability . . . . . . . . . . . . . . . 46

2.10 Six Sigma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482.10.1 Numerical Examples . . . . . . . . . . . . . . . . . . . . 512.10.2 Six Sigma in the Service Sector. Thermal

Water Treatments for Health and Fitness . . . . 51

Organizations depend on their customers and there-fore should understand current and future customerneeds, should meet customer requirements and striveto exceed customer expectations... Identifying, un-derstanding and managing interrelated processes asa system contributes to the organizations effective-ness and efficiency in achieving its objectives (ENISO 9000:2006 Quality management systems fun-damentals and vocabulary).

Nowadays, user and consumer assume their ownchoices regarding very important competitive factorssuch as quality of product, production process, andproduction system. Users and consumers start makingtheir choices when they feel they are able to value andcompare firms with high quality standards by them-selves.

This chapter introduces the reader to the main prob-lems concerning management and control of a qual-ity system and also the main supporting decision mea-sures and tools for so-called statistical quality control(SQC) and Six Sigma.

2.1 Introduction to QualityManagement Systems

The standard EN ISO 8402:1995, replaced by ENISO 9000:2005, defines quality as the totality ofcharacteristics of an entity that bear on its ability tosatisfy stated and implied needs, and product as

R. Manzini, A. Regattieri, H. Pham, E. Ferrari, Maintenance for Industrial Systems 17 Springer 2010

• 18 2 Quality Management Systems and Statistical Quality Control

the result of activities or processes and can be tan-gible or intangible, or a combination thereof. Conse-quently, these definitions refer to production systemsboth in industrial sectors, such as insurance, banking,and transport, and service sectors, in accordance withthe conceptual framework introduced in Chap. 1. An-other synthetic definition of quality is the degree towhich a set of inherent characteristics fulfills require-ments (ISO 9000:2005).

A requirement is an expectation; it is generally re-lated to the organization, customers, or other inter-ested, or involved, parties. We choose to name allthese entities, i. e., the stakeholders of the organiza-tion, as customers and, consequently, the basic key-word in quality management is customer satisfaction.Another basic term is capability as the ability of theorganization, system, or process to realize a productfulfilling the requirements.

A quality management system is a particular man-agement system driving the organization with regardto quality. In other words, it assists companies and or-ganizations in enhancing customer satisfaction. Thisis the result of products capable of satisfying the ever-changing customer needs and expectations that conse-quently require the continuous improvement of prod-ucts, processes, and production systems.

Quality management is a responsibility at all levelsof management and involves all members of an organi-

Fig. 2.1 Process-based qual-ity management system(ISO 9000:2005)

Continual improvement of the quality management system

Requirements

Interested parties

Interested parties

Managementresponsability

Resourcemanagementrealization

Mesaurementanalysis &

improvement

Product realization

Product

Satisfaction

IN OUT

Information flow

zation. For this reason, in the 1980s total quality man-agement (TQM) as a business management strategyaimed at embedding awareness of quality in all orga-nizational processes found very great success. Accord-ing to the International Organization for Standardiza-tion (ISO) standards (ISO 9000:2006), the basic stepsfor developing and implementing a quality manage-ment system are:

determination of needs and expectations of cus-tomers and other involved parties;

definition of the organizations quality policy andquality objectives;

determination of processes and responsibilities forquality assessment;

identification and choice of production resourcesnecessary to attain the quality objectives;

determination and application of methods to mea-sure the effectiveness and efficiency of each processwithin the production system;

prevention of nonconformities and deletion of therelated causes;

definition and application of a process for contin-uous improvement of the quality management sys-tem.

Figure 2.1 presents the model of a process-based qual-ity management system, as proposed by the ISO stan-dards.

• 2.3 ISO Standards for Quality Management and Assessment 19

2.2 International Standardsand Specifications

According to European Directive 98/34/EC of 22 June1998, a standard is a technical specification for re-peated or continuous application approved, withouta compulsory compliance, by one of the following rec-ognized standardization bodies:

ISO; European standard (EN); national standard (e. g., in Italy UNI).

Standards are therefore documents defining the char-acteristics (dimensional, performance, environmental,safety, organizational, etc.) of a product, process, orservice, in accordance with the state of the art, andthey are the result of input received from thousands ofexperts working in the European Union and elsewherein the world. Standards have the following distinctivecharacteristics:

Consensuality: They must be approved with theconsensus of the participants in the works of prepa-ration and confirmed by the result of a public en-quiry.

Democracy: All the interested economic/social par-ties can participate in the works and, above all,have the opportunity to make observations duringthe procedure prior to final and public approval.

Transparency: UNI specifies the basic milestonesof the approval procedure for a draft standard, plac-ing the draft documents at the disposal of the inter-ested parties for consultation.

Voluntary nature: Standards are a source of refer-ence that the interested parties agree to apply freelyon a noncompulsory basis.

In particular CEN, the European Committee for Stan-dardization founded in 1961 by the national standardsbodies in the European Economic Community andEFTA countries, is contributing to the objectives of theEuropean Union and European Economic Area withvoluntary technical standards promoting free trade,safety of workers and consumers, interoperability ofnetworks, environmental protection, exploitation of re-search and development programs, and public procure-ment.

CEN works closely with the European Commit-tee for Electrotechnical Standardization (CENELEC),the European Telecommunications Standards Institute

(ETSI), and the ISO. CEN is a multisectorial organi-zation serving several sectors in different ways, as il-lustrated in the next sections and chapters dealing withsafety assessment.

2.3 ISO Standards for QualityManagement and Assessment

The main issues developed by the technical committeefor the area of quality are:

1. CEN/CLC/TC 1 criteria for conformity assess-ment bodies;

2. CEN/SS F20 quality assurance.

Table 2.1 reports the list of standards belonging to thefirst technical committee since 2008.

Similarly, Table 2.2 reports the list of standards be-longing to the technical committee CEN/SS F20 since2008, while Table 2.3 shows the list of standards cur-rently under development.

Quality issues are discussed in several standardsthat belong to other technical groups. For example,there is a list of standards of the aerospace series deal-ing with quality, as reported in Table 2.4. Table 2.5presents a list of standards for quality managementsystems in health care services. Similarly, there areother sets of standards for specific sectors, businesses,or products.

2.3.1 Quality Audit, Conformity,and Certification

Quality audit is the systematic examination of a qual-ity system carried out by an internal or external qual-ity auditor, or an audit team. It is an independent anddocumented process to obtain audit evidence and to al-low its objective evaluation, in order to verify the ex-tent of the fulfillment of the audit criteria. In particular,third-party audits are conducted by external organiza-tions providing certification/registration of conformityto a standard or a set of standards, e. g., ISO 9001 orISO 14001. The audit process is the basis for the dec-laration of conformity.

The audit process is conducted by an auditor, or anaudit team, i. e., a person or a team, with competence

• 20 2 Quality Management Systems and Statistical Quality Control

Table 2.1 CEN/CLC/TC 1 criteria for conformity assessment bodies, standards published since 2008

Standard Title

EN 45011:1998 General requirements for bodies operating product certification systems (ISO/IEC Guide65:1996)

EN 45503:1996 Attestation Standard for the assessment of contract award procedures of entitiesoperating in the water, energy, transport and telecommunications sectors

EN ISO/IEC 17000:2004 Conformity assessment Vocabulary and general principles (ISO/IEC 17000:2004)EN ISO/IEC 17011:2004 Conformity assessment General requirements for accreditation bodies accrediting

conformity assessment bodies (ISO/IEC 17011:2004)EN ISO/IEC 17020:2004 General criteria for the operation of various types of bodies performing inspection

(ISO/IEC 17020:1998)EN ISO/IEC 17021:2006 Conformity assessment Requirements for bodies providing audit and certification of

management systems (ISO/IEC 17021:2006)EN ISO/IEC 17024:2003 Conformity assessment General requirements for bodies operating certification of

persons (ISO/IEC 17024:2003)EN ISO/IEC 17025:2005 General requirements for the competence of testing and calibration laboratories

(ISO/IEC 17025:2005)EN ISO/IEC 17025:2005/AC:2006 General requirements for the competence of testing and calibration laboratories

(ISO/IEC 17025:2005/Cor.1:2006)EN ISO/IEC 17040:2005 Conformity assessment General requirements for peer assessment of conformity

assessment bodies and accreditation bodies (ISO/IEC 17040:2005)EN ISO/IEC 17050-1:2004 Conformity assessment Suppliers declaration of conformity Part 1: General

requirements (ISO/IEC 17050-1:2004)EN ISO/IEC 17050-2:2004 Conformity assessment Suppliers declaration of conformity Part 2: Supporting

documentation (ISO/IEC 17050-2:2004)

Table 2.2 CEN/SS F20 quality assurance, standards published since 2008

Standard Title

EN 45020:2006 Standardization and related activities General vocabulary (ISO/IEC Guide 2:2004)EN ISO 10012:2003 Measurement management systems Requirements for measurement processes and

measuring equipment (ISO 10012:2003)EN ISO 15378:2007 Primary packaging materials for medicinal products Particular requirements for the

application of ISO 9001:2000, with reference to good manufacturing practice (GMP)(ISO 15378:2006)

EN ISO 19011:2002 Guidelines for quality and/or environmental management systems auditing(ISO 19011:2002)

EN ISO 9000:2005 Quality management systems Fundamentals and vocabulary (ISO 9000:2005)EN ISO 9001:2000 Quality management systems Requirements (ISO 9001:2000)EN ISO 9004:2000 Quality management systems Guidelines for performance improvements

(ISO 9004:2000)

Table 2.3 CEN/SS F20 quality assurance, standards under development as of October 2008

Standard Title

ISO 15161:2001 Guidelines on the application of ISO 9001:2000 for the food and drink industry(ISO 15161:2001)

prEN ISO 9001 Quality management systems Requirements (ISO/FDIS 9001:2008)prEN ISO 19011 rev Guidelines for auditing management systemsprEN ISO 9004 Managing for the sustained success of an organization A quality management

approach (ISO/DIS 9004:2008)

• 2.3 ISO Standards for Quality Management and Assessment 21

Table 2.4 Aerospace series, quality standards

Standard Title

EN 9102:2006 Aerospace series Quality systems First article inspectionEN 9103:2005 Aerospace series Quality management systems Variation management of key

characteristicsEN 9110:2005 Aerospace series Quality systems Model for quality assurance applicable to

maintenance organizationsEN 9120:2005 Aerospace series Quality management systems Requirements for stockist distributors

(based on ISO 9001:2000)EN 9104:2006 Aerospace series Quality management systems Requirements for Aerospace Quality

Management System Certification/Registrations ProgramsEN 9111:2005 Aerospace series Quality management systems Assessment applicable to

maintenance organizations (based on ISO 9001:2000)EN 9121:2005 Aerospace series Quality management systems Assessment applicable to stockist

distributors (based on ISO 9001:2000)EN 9132:2006 Aerospace series Quality management systems Data Matrix Quality Requirements

for Parts MarkingEN 4179:2005 Aerospace series Qualification and approval of personnel for nondestructive testingEN 4617:2006 Aerospace series Recommended practices for standardizing company standardsEN 9101:2008 Aerospace series Quality management systems Assessment (based on

ISO 9001:2000)EN 9104-002:2008 Aerospace series Quality management systems Part 002: Requirements for Oversight

of Aerospace Quality Management System Certification/Registrations Programs

Table 2.5 CEN/TC 362, health care services, quality management systems

Standard Title

CEN/TR 15592:200 Health services Quality management systems Guide for the use ofEN ISO 9004:2000 in health services for performance improvement

CEN/TS 15224:2005 Health services Quality management systems Guide for the use ofEN ISO 9001:2000

to conduct an audit, in accordance with an audit pro-gram consisting of a set of one or more audits plannedfor a specific time frame. Audit findings are used to as-sess the effectiveness of the quality management sys-tem and to identify opportunities for improvement.Guidance on auditing is provided by ISO 19011:2002(Guidelines for quality and/or environmental manage-ment systems auditing).

The main advantages arising from certification are:

improvement of the company image; increase of productivity and company profit; rise of contractual power; quality guarantee of the product for the client.

In the process of auditing and certification, the docu-mentation plays a very important role, enabling com-munication of intent and consistency of action. Severaltypes of documents are generated in quality manage-ment systems.

2.3.2 Environmental Standards

Every standard, even if related to product, service,or process, has an environmental impact. For a prod-uct this can vary according to the different stages ofthe product life cycle, such as production, distribu-tion, use, and end-of-life. To this purpose, CEN hasrecently been playing a major role in reducing envi-ronmental impacts by influencing the choices that aremade in connection with the design of products andprocesses. CEN has in place an organizational struc-ture to respond to the challenges posed by the devel-opments within the various sectors, as well as by theevolution of the legislation within the European Com-munity. The main bodies within CEN are:

1. The Strategic Advisory Body on the Environment(SABE) an advisory body for the CEN TechnicalBoard on issues related to environment. Stakehold-ers identify environmental issues of importance

• 22 2 Quality Management Systems and Statistical Quality Control

to the standardization system and suggest corre-sponding solutions.

2. The CEN Environmental Helpdesk provides sup-port and services to CEN Technical Bodies on howto address environmental aspects in standards.

3. Sectors some sectors established a dedicatedbody to address environmental matters associatedwith their specific needs, such as the Construc-tion Sector Network Project for the Environment(CSNPE).

4. Associates two CEN associate members providea particular focus on the environment within stan-dardization:

European Environmental Citizens Organizationfor Standardization (ECOS);

European Association for the Coordination ofConsumer Representation in Standardization(ANEC).

Table 2.6 Technical committees on the environment

Technical commitee Title

CEN/TC 223 Soil improvers and growing mediaCEN/TC 230 Water analysisCEN/TC 264 Air qualityCEN/TC 292 Characterization of wasteCEN/TC 308 Characterization of sludgesCEN/TC 345 Characterization of soilsCEN/TC 351 Construction Products Assessment of release of dangerous substances

Table 2.7 Committee CEN/SS S26 environmental management

Standard Title

EN ISO 14031:1999 Environmental management Environmental performance evaluation Guidelines(ISO 14031:1999)

EN ISO 14001:2004 Environmental management systems Requirements with guidance for use(ISO 14001:2004)

EN ISO 14024:2000 Environmental labels and declarations Type I environmental labeling Principles andprocedures (ISO 14024:1999)

EN ISO 14021:2001 Environmental labels and declarations Self-declared environmental claims (Type IIenvironmental labelling) (ISO 14021:1999)

EN ISO 14020:2001 Environmental labels and declarations General principles (ISO 14020:2000)EN ISO 14040:2006 Environmental management Life cycle assessment Principles and framework

(ISO 14040:2006)EN ISO 14044:2006 Environmental management Life cycle assessment Requirements and guidelines

(ISO 14044:2006)prEN ISO 14005 Environmental management systems Guidelines for a staged implementation of an

environmental management system, including the use of environmental performanceevaluation

Table 2.6 reports the list of technical committees onthe environment.

There are several standards on environmental man-agement. To exemplify this, Table 2.7 reports the listof standards grouped in accordance with the commit-tee CEN/SS S26 environmental management.

ISO 14000 is a family of standards supporting theorganizations on the containment of the polluting ef-fects on air, water, or land derived by their operations,in compliance with applicable laws and regulations. Inparticular, ISO 14001 is the international specificationfor an environmental management system (EMS). Itspecifies requirements for establishing an environmen-tal policy, determining environmental aspects and im-pacts of products/activities/services, planning environ-mental objectives and measurable targets, implemen-tation and operation of programs to meet objectivesand targets, checking and corrective action, and man-agement review.

• 2.4 Introduction to Statistical Methods for Quality Control 23

x

x

uniform n = 2 n = 10

n = 10

exponential

n = 2

x x

n = 10n = 2

x x

Gauss

x

Fig. 2.2 Central limit theorem, examples

2.4 Introduction to Statistical Methodsfor Quality Control

The aim of the remainder of this chapter is the intro-duction and exemplification of effective models andmethods for statistical quality control. These tools arevery diffuse and can be used to guarantee also thereliability,1 productivity and safety of a generic pro-duction system in accordance with the purpose of thisbook, as illustrated in Chap. 1.

2.4.1 The Central Limit Theorem

This section briefly summarizes the basic result ob-tained by this famous theorem. Given a population orprocess, a random variable x, with mean and stan-dard deviation , let Nx be the mean of a random sam-ple made of n elements x1; x2; : : : ; xn extracted fromthis population: when the sample size n is sufficientlylarge, the sampling distribution of the random vari-

1 Reliability, properly defined in Chap. 5, can be also defined asquality in use.

able Nx can be approximated by a normal distribution.The larger the value of n, the better the approximation.

This theorem holds irrespective of the shape of thepopulation, i. e., of the density function of the vari-able x.

The analytic translation of the theorem is given bythe following equations:

M. Nx/ D NNx D O; (2.1). Nx/ D Op

n; (2.2)

where O is the estimation of and O is the estimationof .

Figure 2.2 graphically and qualitatively demon-strates these results representing the basis for the de-velopment and discussion of the methods illustratedand applied below. In the presence of a normal distri-bution of population, the variable Nx is normal too foreach value of size n.

Figure 2.3 quantitatively demonstrates the centrallimit theorem starting from a set of random valuesdistributed in accordance with a uniform distribution0; 10: the variable Nx is a normally distributed vari-able when the number of items used for the calculusof mean Nxi is sufficiently large. In detail, in Fig. 2.3the size n is assumed be 2, 5, and 20.

• 24 2 Quality Management Systems and Statistical Quality Control

9.07.56.04.53.01.50.0

0.16

0.12

0.08

0.04

0.009.07.56.04.53.01.5

0.20

0.15

0.10

0.05

0.00

87654321

0.3

0.2

0.1

0.06.66.05.44.84.23.6

0.8

0.6

0.4

0.2

0.0

DATA (n=1)

Den

sity

Mean (n=2)

Mean (n=5) Mean (n=20)

Histogram of DATA (n=1) and means (n>1)

Fig. 2.3 Central limit theorem, histogram of Nx for n D f1; 2; 5; 20g. Uniform distribution of variable x

2.4.2 Terms andDefinition in StatisticalQuality Control

Quality control is a part of quality management(ISO 9000:2005) focused on the fulfillment of qualityrequirements. It is a systematic process to monitor andimprove the quality of a product, e. g., a manufacturedarticle, or service by achieving the quality of theproduction process and the production plant. A list ofbasic terms and definitions in accordance with the ISOstandards follows:

Process, set of interrelated activities turning inputinto output. It is a sequence of steps that results inan outcome.

Product, result of a process. Defect, not fulfillment of a requirement related to

an intended or specified use. Measurement process, set of operations to deter-

mine the value of a quantity. Key characteristic, a quality characteristic the prod-

uct or service should have to fulfill customer re-quirements and expectations.

Value of a key characteristic. For several productsa single value is the desired quality level for a char-acteristic.

Nominal or target value. It is the expected valuefor the key characteristic. It is almost impossible tomake each unit of product or service identical to the

next; consequently it is nonsense to ask for identi-cal items having a key characteristic value exactlyequal to the target value. This need for flexibilityis supported by the introduction of limits and toler-ances.

Specification limit, or tolerance, conformanceboundary, range, specified for a characteristic.The lower specification limit (LSL) is the lowerconformance boundary, the upper specificationlimit (USL) is the upper conformance bound-ary.The following equation summarizes the relation-ship among these terms:

Specification limits D (nominal value) tolerance:(2.3)

One-sided tolerance. It relates to characteristicswith only one specification limit.

Two-sided tolerance. It refers to characteristics withboth USLs and LSLs.

Nonconformity. It is a nonfulfillment of a require-ment. It is generally associated with a unit: a non-conformity unit, i. e., a unit that does not meet thespecifications.

Nonconforming product or service. A product orservice with one or more nonconformities. A non-conforming product is not necessary defective, i. e.,no longer fit for use.

• 2.6 Control Charts 25

2.5 Histograms

Histograms are effective and simple graphic tools forthe comprehension and analysis of a process behav-ior with regards to the target value and the specifica-tion limits. The histograms illustrate the frequency dis-tribution of variable data: the values assumed by thevariable are reported on the abscissa, while the verti-cal axis reports the absolute or relative frequency val-ues. The specification limits are generally included inthe graph and give warnings of possible process prob-lems. Figure 2.4 exemplifies a few histogram shapes.The control charts illustrated in the next section repre-sent a more effective tool for the analysis of a produc-tion process.

2.6 Control Charts

Control charts, introduced by W.A. Shewhart in 1924,are effective tools for the analysis of the variation ofrepetitive processes. They are able to identify possi-ble sources of process variation in order to control andeventually eliminate them. In a generic process, twodifferent kinds of variations can be distinguished:

reasonsconfigurations

process shifted to the "right" ormeasurements are out of calibration

LSL USL

mix of two different processes, e.g. datafrom two operators, two machines, orcollected at different points in time

LSL USL

LSL USL

"Special (assignable) causes" ofvariation, i.e. errors of measurement or

in the activity of data collection

process shifted to the "left" ormeasurements are out of calibration

LSL USL

granularity, i.e. "granular process"

LSL USL

stable process within specifications

LSL USL

configurations reasons

process variation too large for thespecification limits

LSL USL

"truncated" data

LSL USL

Fig. 2.4 Exemplifying histograms shapes. LSL lower specification limit, USL upper specification limit

1. Common causes variations. They are the noise ofa production system and are uncontrollable varia-tions.

2. Assignable (or special) causes variations. Theycan be properly identified and controlled. Someexamples are turnover in workman load, break-downs, machine or tool wear out, and tool change.

Control charts are a family of tools for detecting theexistence of special causes variations in order to avoidthem, i. e., eliminate all anomalous controllable pat-terns, and bring the process into a state called of sta-tistical control, or simply in control, whose ran-dom behavior is justified by the existence of commoncauses variations. The in control state is necessary toobtain conforming products, as properly discussed inthe following sections on capability analysis and SixSigma.

Control charts can be constructed by extracting suc-cessive samples from the variable output of the pro-cess. These samples, also called subgroups, all havesize n and have to be taken at regular intervals of time.For each group a summary statistic is calculated andplotted as illustrated in Fig. 2.5.

Typical statistical measures calculated for each sub-group are reported in Table 2.8, where the related sta-tistical distribution is cited together with the values of

• 26 2 Quality Management Systems and Statistical Quality Control

Subgroup number

Subg

roup

stat

istic

(e.g

. R, p

, u)

1 2 3 4 5

UCL

LCL

centerline

Fig. 2.5 Control chart

the centerline and control limits, as properly defined inthe next subsections.

A control chart is made of three basic lines as illus-trated in Fig. 2.5:

1. Centerline. It is the mean of the statistic quantifiedfor each subgroup, the so-called subgroup statistic(e. g., mean, range, standard deviation).

2. Control limits. These limits on a control chart de-limit that region where a data point falls outside,thus alerting one to special causes of variation.This region is normally extended three standarddeviations on either side of the mean. The controllimits are:

upper control limit (UCL), above the mean; lower control limit (LCL), below the mean.

The generic point of the chart in Fig. 2.5 may representa subgroup, a sample, or a statistic. k different sam-ples are associated with k different points whose tem-poral sequence is reported on the chart. Control limitsare conventionally set at a distance of three standardserrors, i. e., three deviations of the subgroup statis-tic, from the centerline, because the distribution ofsamples closely approximates a normal statistical dis-tribution by the central limit theorem. Consequently,the analyst expects that about 99.73% of samples liewithin three standard deviations of the mean. This cor-responds to a probability of 0.27% that a control chartpoint falls outside one of the previously defined con-trol limits when no assignable causes are present.

In some countries, such as in the UK, the adoptedconvention of three standard deviations is different.

Figure 2.6 presents eight different anomalous pat-terns of statistic subgroups tested by Minitab Statis-

tical Software to find reliable conditions for the in, orout, control state of the process.

A process is said to be in control when all sub-groups on a control chart lie within the control lim-its and no anomalous patterns are in the sequence ofpoints representing the subgroups. Otherwise, the pro-cess is said to be out of control, i. e., it is not ran-dom because there are special causes variations affect-ing the output obtained.

What happens in the presence of special causes?It is necessary to identify and eliminate them. Conse-quently, if a chart shows the possible existence of spe-cial causes by one of the anomalous behaviors illus-trated in Fig. 2.6, the analyst and the person responsi-ble for the process have to repeat the analysis by elim-inating the anomalous subgroups. Now, if all the testsare not verified, the process has been conducted to thestate of statistical control.

2.7 Control Charts for Means

These charts refer to continuous measurement data,also called variable data (see Table 2.8), becausethere are an infinite number of data between twogeneric ones.

2.7.1 The R-Chart

This is a chart for subgroup ranges. The range is thedifference between the maximum and the minimumvalues within a sample of size n:

Ri D maxj D1;:::;nfxij g minj D1;:::;nfxij g; (2.4)

where i is a generic sample and xij is the j th value inthe sample i .

Consequently, the centerline is

R D NR D 1k

kX

iD1Ri : (2.5)

This value is a good estimation of the mean value ofvariable Ri , called R . We also define the statisticmeasure of variability of the variable Ri , the standarddeviation R. By the central limit theorem, the distri-bution of values Ri is normal. As a consequence, the

• 2.7 Control Charts for Means 27

Table2.8

Stat

isti

calm

easu

res

and

cont

rolc

hart

clas

sific

atio

n

Nam

eTy

peSt

atis

tica

lSt

atis

tic

Cen

terl

ine

Con

trol

lim

its

Stan

dard

ofda

stri

buti

onm

easu

rede

viat

ion

R-c

hart

Var

iabl

eC

onti

nuou

s

norm

aldi

stri

buti

onR

ange

RN R

D1 k

k X iD1

Ri

UC

LD

R

C3

R

D4

N RL

CL

D

R

3R

D

3N R

s-c

hart

Var

iabl

eC

onti

nuou

s

norm

aldi

stri

buti

onSt

anda

rdde

viat

ion

sO S

DNsD

1 k

k X iD1

s iU

CL

DB

4Ns

LC

LD

B3Ns

Nx-ch

art

Var

iabl

eC

onti

nuou

s

norm

aldi

stri

buti

onM

ean

NxO

DN N X

D1 k

k X iD1

N X iU

CL

DOC

3O p n

N N X

C3

N R=d

2p n

DN N X

CA

2N R

LC

LD

O

3O p n

N N X

3

N R=d

2p n

DN N X

A

2N R

ODN R d 2

UC

LD

OC3

O p n

N N XC

3Ns=

c4

p nD

N N XC

A3Ns

LC

LD

O

3O p n

N N X

3

Ns=c

4p n

DN N X

A

3Ns

ODNs c 4

p-c

hart

Att

ribu

teD

iscr

ete

bi

nom

ial

dist

ribu

tion

Non

conf

orm

ing

prop

orti

onp

NpD

1 k

k X iD1

pi

UC

LD

NpC3s

Np.1

Np/

n

LC

LD

Np3s

Np.1

Np/

n

NpD

x1

Cx

2C

C

xk

1

Cx

k

n1

Cn

2C

C

nk

1

Cn

k

UC

Li

DNpC

3sNp.

1

Np/n

i

LC

Li

DNp

3sNp.

1

Np/n

i

np

-cha

rtA

ttri

bute

Dis

cret

e

bino

mia

ldi

stri

buti

onN

umbe

rof

nonc

onfo

rmit

ies

np

nNp

UC

LD

nNpC

3pn

Np.1

Np/

LC

LD

nNp

3pn

Np.1

Np/

c-c

hart

Att

ribu

teD

iscr

ete

Po

isso

ndi

stri

buti

onN

umbe

rof

nonc

onfo

rmit

ies

NcD1 k

k X iD1

ci

UC

LD

NcC3p

NcL

CL

DNc

3pNc

u-c

hart

Att

ribu

teD

iscr

ete

Po

isso

ndi

stri

buti

onN

onco

nfor

mit

ies

per

unit

uNuD

1 k

k X iD1

ui

UC

Li

DNuC

3sNu n i

LC

Li

DNu

3sNu n i

UC

Lup

per

cont

roll

imit

,LC

Llo

wer

cont

roll

imit

,UC

Li

uppe

rco

ntro

llim

itfo

rsa

mpl

ei,

LC

Li

low

erco

ntro

llim

itfo

rsa

mpl

ei

• 28 2 Quality Management Systems and Statistical Quality Control

Test 1 1 point beyond 3 std.dev. (zone A)

Test 2 9 points in a row on same sideof the center line

Test 3 6 points in a row all increasing(or decreasing)

Test 4 14 points in a row alternangup and down

Test 6 4 out of 5 points more than1 std.dev.

Test 5 2/3 points in a row more than2 std.dev.

Test 7 15 points in a row within1 std.dev. (either side)

Test 8 8 points in a row more than1 std.dev. (either side)

Fig. 2.6 Eight tests for special causes investigation, Minitab Statistical Software

control limits are defined as

UCLR D R C 3R D4 NR;LCLR D R 3R D3 NR;

(2.6)

where R is the standard deviation of the variable Rand D4 is a constant value depending on the size of

the generic subgroup. The values are reported in Ap-pendix A.2.

The following equation represents an estimation ofthe standard deviation of the variable and continuousdata xij :

O DNRd2

(2.7)

• 2.7 Control Charts for Means 29

2.7.2 Numerical Example, R-Chart

This application refers to the assembly process ofan automotive engine. The process variable is a dis-tance, D, measured in tenths of millimeters, betweentwo characteristic axes in the drive shafts and heads.Table 2.9 presents the data collected over 25 days ofobservation and grouped in samples of size n D 5.

By the application of Eqs. 2.5 and 2.6, we have

NR D 1k

kX

iD1Ri D 1

25.R1 C CR25/ 6:50;

UCL D R C 3R D4 NR DD4.nD5/D2:114

13:74;

LCL D R 3R D3 NR DD4.nD5/D0

0:

The R-chart obtained is reported in Fig. 2.7. Previ-ously introduced tests for anomalous behaviors are notverified. As a consequence, the process seems to berandom and coherent with itself and its characteris-tic noise and variance. There are no special causes ofvariation.

Table 2.9 Data 25 subgroups, numerical example

Sample Month Day D (mm=10)

1 7 25 0:387 5.192 1.839 0.088 1.7742 7 26 4.251 3.333 4.398 6.082 1.7063 7 27 2:727 2:806 4.655 0.494 2:8074 7 28 6.980 3.280 3.372 1:914 2.4785 7 29 3.978 3.479 7.034 4.388 1:7906 7 30 3.424 1.758 0.009 0:216 1.8327 7 31 4:285 2:369 2:666 2.639 3.0818 8 1 1:756 1:434 1.887 1:678 7.0609 8 2 4.184 1.005 0.825 6:427 4:598

10 8 3 3:577 1:684 1.800 4.339 0.02711 8 4 2:467 2:752 4:029 2:798 2:15212 8 5 1.199 0.817 0:213 0:737 1:75713 8 6 4.312 1.127 2.534 1.618 0:66514 8 7 3.282 3.319 3:564 3.430 1.55615 8 8 2.000 3:364 1:996 1:830 0.01516 8 9 3.268 1.519 2.704 0.138 0:05017 8 10 3.356 3:335 3:358 4:302 2:85618 8 11 0:240 3:811 1:615 3:510 4:37719 8 12 4:524 0:091 1.945 4.515 1:66720 8 13 0.837 4:536 4.249 0.114 0:08721 8 14 1:016 2.023 4.539 0.075 2:72422 8 15 4.547 0.262 4:108 1:881 0:00423 8 16 0.159 3.786 1:951 6.315 5.16124 8 17 0.842 3:550 1:805 2:731 1:61025 8 18 4.435 1.730 0:185 0.242 4:689

2.7.3 The Nx-Chart

This is a chart for subgroup means. In the Nx-chart,also called x-chart, the problem is the estimationof the standard deviation of the population of val-ues. In Sect. 2.7.1, Eq. 2.7 is an effective estimation.Consequently, this chart is generally constructed afterthe creation of the R-chart and reveals the process tobe in the state of statistical control. The centerline ofthe statistic variable Nx is the average of the subgroupmeans:

O D O. Nx/ D NNx D 1k

kX

iD1Nxi D

kX

iD1

nX

j D1xij : (2.8)

The control limits are

UCL Nx D OC 3 Opn

NNx C 3NR=d2pn

D NNx C A2 NR;

LCL Nx D O 3 Opn

NNx 3NR=d2pn

D NNx A2 NR;(2.9)

where d2 andA2 are constant values as reported in Ap-pendix A.2.

• 30 2 Quality Management Systems and Statistical Quality Control

DayMonth

Sample

18161412108642312927258888888887777252321191715131197531

14

12

10

8

6

4

2

0

Sam

ple

Ran

ge

_R=6.50

UCL=13.74

LCL=0

R Chart of D

Fig. 2.7 R-chart, numerical example. Minitab Statistical Software. UCL upper control limit, LCL lower control limit

2.7.4 Numerical Example, Nx-Chart

Consider the application introduced in Sect. 2.7.2. ByEqs. 2.8 and 2.9,

O D NNx D 1k

kX

iD1Nxi D 1

25. Nx1 C C Nx25/ D 0:389;

UCL D OC 3 Opn

NNx C 3NR=d2pn

D NNx C A2 NRD

A2D0:5770:389 C 0:577 6:5 4:139;

LCL D O 3 Opn

NNx 3NR=d2pn

D NNx A2 NRD

A2D0:5770:389 0:577 6:5 3:361:

The chart obtained is reported in Fig. 2.8. Test 6 foranomalous behaviors is verified in sample 5, month 7,and day 29, i. e., there are four of five points in zone Bor beyond. As a consequence, the process seems to beout of control. There is in fact a very scarce proba-bility of having a sample in those points when the pro-cess is in control. We assume we are able to properlyidentify this special cause of variation and to elimi-nate it. Figure 2.9 presents the charts obtained from thepool of samples without the anomalous subgroup 5.The chart shows another potential anomalous behav-

ior regarding subgroup 4. In this way, assuming weidentify and eliminate new special causes, we obtainFigs. 2.10 and 2.11. In particular, Fig. 2.11 presentsa process in the state of statistical control: subgroups 2,4, and 5 have been eliminated.

2.7.5 The s-Chart

This chart for subgroup standard deviation can be usedto support the construction of the x-chart by the es-timation of the standard deviation of the continuousvariable xij . In particular, the control limits of the x-chart use the centerline of the s-chart.

The average of standard deviation of subgroups, Os,is the centerline of the s-chart:

OS D O.si / D Ns D 1k

kX

iD1si ; (2.10)

where O.si / is the estimation of the mean of the vari-able si , the standard deviation of a subgroup.

The control limits are

UCLs D O.si /C 3 O.si /pn

D B4 Ns;

LCLs D O.si / 3 O.si /pn

D B3 Ns;(2.11)

• 2.7 Control Charts for Means 31

DayMonth

Sample

18161412108642312927258888888887777252321191715131197531

4

2

0

-2

-4

Sam

ple

Mea

n

__X=0.389

UCL=4.136

LCL=-3.359

DayMonth

Sample

18161412108642312927258888888887777252321191715131197531

15

10

5

0

Sam

ple

Ran

ge

_R=6.50

UCL=13.74

LCL=0

6

Xbar-R Chart of D

Fig. 2.8 x-chart from R. Numerical example (25 samples). Minitab Statistical Software

Day [24]Month [24]

Sample [24]

1715131197531302725888888888777

24222018161412108631

4

2

0

-2

-4

Sam

ple

Mea

n

__X=0.262

UCL=3.984

LCL=-3.459

Day [24]Month [24]

Sample [24]

1715131197531302725888888888777

24222018161412108631

15

10

5

0

Sam

ple

Ran

ge

_R=6.45

UCL=13.64

LCL=0

5

Xbar-R Chart [rif.no sub.5]

Fig. 2.9 x-chart from R. Numerical example (24 samples). Minitab Statistical Software

• 32 2 Quality Management Systems and Statistical Quality Control

Day [23]Month [23]

Sample [23]

18161412108642312725888888888777

25232119171513119731

4

2

0

-2

-4

Sam

ple

Mea

n

__X=0.150

UCL=3.847

LCL=-3.546

Day [23]Month [23]

Sample [23]

18161412108642312725888888888777

25232119171513119731

15

10

5

0

Sam

ple

Ran

ge

_R=6.41

UCL=13.55

LCL=0

1Xbar-R Chart [no sub 4 and 5]

Fig. 2.10 x-chart from R. Numerical example (23 samples). Minitab Statistical Software

Day [22]Month [22]

Sample [22]

1715131197531302588888888877

2422201816141210861

4

2

0

-2

-4

Sam

ple

Mea

n

__X=-0.022

UCL=3.729

LCL=-3.774

Day [22]Month [22]

Sample [22]

1715131197531302588888888877

2422201816141210861

15

10

5

0

Sam

ple

Ran

ge

_R=6.50

UCL=13.75

LCL=0

Xbar-R Chart [no sub 2, 4 and 5]

Fig. 2.11 x-chart from R. Numerical example (22 samples). Minitab Statistical Software

• 2.8 Control Charts for Attribute Data 33

where O.si / is the estimation of the standard devia-tion of the variable si , the standard deviation of a sub-group, and B3 and B4 are constant values reported inAppendix A.2.

The standard deviation of the process measurementis

O D O. Nxi / D Nsc4: (2.12)

As a consequence, the control limits of the x-chart are

UCL Nx D OC 3 Opn

NNx C 3 Ns=c4pn

D NNx C A3 Ns;

LCL Nx D O 3 Opn

NNx 3 Ns=c4pn

D NNx A3 Ns;(2.13)

where A3 is a constant value reported in Ap-pendix A.2.

2.7.6 Numerical Example, s-Chartand Nx-Chart

Table 2.10 reports a set of measurement data madefor 20 samples of size n D 5. They are the outputof a manufacturing process in the automotive industry.The last three columns report some statistics useful forthe construction of the control charts and for verifica-tion of the status of the control of the process.

With use of the values of the constant parameters inAppendix A.2, the following control limits and center-lines have been obtained.

Firstly, we propose the results related to the R-chart. By Eq. 2.5 the centerline is

R D NR D 1k

kX

iD1Ri 0:004155:

By Eq. 2.6

UCLR D4 NR D 2:114 0:004155 0:008784:LCLR D3 NR D 0 0:004155 D 0:

These results are very close to those proposed by theR-chart, as constructed by the tool Minitab Statisti-cal Software (Fig. 2.12). From the R-chart the processseems to be in the state of statistical control.

The x-chart is now created assuming the centerlineof theR-chart and in accordance with Eqs. 2.8 and 2.9:

UCLx from R NNx CA2 NRD 0:009237 C 0:577 0:004155D 0:0116;

LCLx from R NNx A2 NRD 0:009237 0:577 0:004155D 0:0068:

The upper section of Fig. 2.12 presents the x-chartwhere some subgroups verify a few tests, as illustratedalso in Fig. 2.13. Consequently, the process is not ina state of control.

Similarly, by the application of Eqs. 2.10, 2.11,and 2.13,

OS D Ns D 1k

kX

iD1si 0:00170:

UCLs B4 Ns D 2:089 0:00170 D 0:00355;LCLs B3 Ns D 0 0:00170 D 0;

UCLx from s NNx C A3 NsD 0:009237 C 1:427 0:00170D 0:01166;

LCLx from s NNx A3 NsD 0:009237 1:427 0:00170D 0:00681:

All these values are also reported in Fig. 2.14, show-ing that the process is not in the state of statisticalcontrol. Consequently, a survey for the identificationand deletion of special causes of variations, and thesubsequent repetition of the control analysis, is re-quired.

2.8 Control Charts for Attribute Data

These charts refer to counted data, also called at-tribute data. They support the activities of monitoringand analysis of production processes whose productspossess, or do not possess, a specified characteristicor attribute. Attributes measurement is frequently ob-tained as the result of human judgements.

• 34 2 Quality Management Systems and Statistical Quality Control

Table 2.10 Measurement data and subgroup statistics. Numerical example

ID sample i Measure X M.xi / Ri si

1 0.0073 0.0101 0.0091 0.0091 0.0053 0.0082 0.0048 0.00192 0.0106 0.0083 0.0076 0.0074 0.0059 0.0080 0.0047 0.00173 0.0096 0.0080 0.0132 0.0105 0.0098 0.0102 0.0052 0.00194 0.0080 0.0076 0.0090 0.0099 0.0123 0.0094 0.0047 0.00195 0.0104 0.0084 0.0123 0.0132 0.0120 0.0113 0.0048 0.00196 0.0071 0.0052 0.0101 0.0123 0.0073 0.0084 0.0071 0.00287 0.0078 0.0089 0.0122 0.0091 0.0095 0.0095 0.0044 0.00168 0.0087 0.0094 0.0120 0.0102 0.0099 0.0101 0.0033 0.00129 0.0074 0.0081 0.0120 0.0116 0.0122 0.0103 0.0048 0.0023

10 0.0081 0.0065 0.0105 0.0125 0.0136 0.0102 0.0071 0.002911 0.0078 0.0098 0.0113 0.0087 0.0118 0.0099 0.0040 0.001712 0.0089 0.0090 0.0111 0.0122 0.0126 0.0107 0.0037 0.001713 0.0087 0.0075 0.0125 0.0106 0.0113 0.0101 0.0050 0.002014 0.0084 0.0083 0.0101 0.0140 0.0097 0.0101 0.0057 0.002315 0.0074 0.0091 0.0116 0.0109 0.0108 0.0100 0.0042 0.001716 0.0069 0.0093 0.0090 0.0084 0.0090 0.0085 0.0024 0.001017 0.0077 0.0089 0.0091 0.0068 0.0094 0.0084 0.0026 0.001118 0.0076 0.0069 0.0062 0.0077 0.0067 0.0070 0.0015 0.000619 0.0069 0.0077 0.0073 0.0074 0.0074 0.0073 0.0008 0.000320 0.0063 0.0071 0.0078 0.0063 0.0088 0.0073 0.0025 0.0011

Mean 0.009237 0.004155 0.0016832

191715131197531

0.011

0.010

0.009

0.008

0.007

Sample

Sam

ple

Mea

n

__X=0.009237

UCL=0.011666

LCL=0.006808

191715131197531

0.008

0.006

0.004

0.002

0.000

Sample

Sam

ple

Ran

ge

_R=0.004211

UCL=0.008905

LCL=0

553

266

6

Xbar-R Chart

Fig. 2.12 R-chart and x-chart from R. Numerical example. Minitab Statistical Software

• 2.8 Control Charts for Attribute Data 35

Test Results for Xbar Chart

TEST 2. 9 points in a row on same side of center line.Test Failed at points: 15

TEST 3. 6 points in a row all increasing or all decreasing.Test Failed at points: 18

TEST 5. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Test Failed at points: 19; 20

TEST 6. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). Test Failed at points: 12; 13; 14; 20

Fig. 2.13 x-chart from R, test results. Numerical example. Minitab Statistical Software

191715131197531

0.011

0.010

0.009

0.008

0.007

Sample

Sam

ple

Mea

n

__X=0.009237

UCL=0.011666

LCL=0.006808

191715131197531

0.004

0.003

0.002

0.001

0.000

Sample

Sam

ple

StD

ev

_S=0.001702

UCL=0.003555

LCL=0

553

266

6

Xbar-S Chart

Fig. 2.14 s-chart and x-chart from s. Numerical example. Minitab Statistical Software

2.8.1 The p-Chart

The p-chart is a control chart for monitoring the pro-portion of nonconforming items in successive sub-groups of size n. An item of a generic subgroup is saidto be nonconforming if it possesses a specified charac-teristic. Given p1; p2; : : : ; pk , the subgroups propor-tions of nonconforming items, the sampling random

variable pi for the generic sample i has a mean anda standard deviation:

p D ;

p Dr.1 /

n;

(2.14)

where is the true proportion of nonconforming itemsof the process, i. e., the population of items.

• 36 2 Quality Management Systems and Statistical Quality Control

The equations in Eq. 2.14 result from the binomialdiscrete distribution of the variable number of noncon-formities x. This distribution function is defined as

p.x/ D n

x

!x.1 /nx ; (2.15)

where x is the number of nonconformities and is theprobability the generic item has the attribute.

The mean value of the standard deviation of thisdiscrete random variable is

DX

x

xp.x/ D n;

DsX

x

.x /2p.x/ D n.1 /:(2.16)

By the central limit theorem, the centerline, as the esti-mated value of , and the control limits of the p-chartare

Op D O.pi / D Np D 1k

kX

iD1pi ; (2.17)

UCLp D Np C 3r Np.1 Np/

n;

LCLp D Np 3r Np.1 Np/

n:

(2.18)

If the number of items for a subgroup is not constant,the centerline and the control limits are quantified bythe following equations:

Np D x1 C x2 C C xk1 C xkn1 C n2 C C nk1 C nk ; (2.19)

where xi is the number of nonconforming items insample i and ni is the number of items within the sub-group i , and

UCLp;i D Np C 3s

Np.1 Np/ni

;

LCLp;i D Np 3s

Np.1 Np/ni

;

(2.20)

where UCLi is the UCL for sample i and LCLi is theLCL for sample i .

Table 2.11 Rejects versus tested items. Numerical example

Day Rejects Tested Day Rejects Tested

21=10 32 286 5=11 21 28122=10 25 304 6=11 14 31023=10 21 304 7=11 13 31324=10 23 324 8=11 21 29325=10 13 289 9=11 23 30526=10 14 299 10=11 13 31727=10 15 322 11=11 23 32328=10 17 316 12=11 15 30429=10 19 293 13=11 14 30430=10 21 287 14=11 15 32431=10 15 307 15=11 19 2891=11 16 328 16=11 22 2992=11 21 304 17=11 23 3183=11 9 296 18=11 24 3134=11 25 317 19=11 27 302

2.8.2 Numerical Example, p-Chart

Table 2.11 reports the data related to the number ofelectric parts rejected by a control process considering30 samples of different size.

By the application of Eqs. 2.19 and 2.20,

Np D x1 C x2 C C xk1 C xkn1 C n2 C C nk1 C nk D

573

9171

0:0625;

UCLp;i D Np C 3s

Np.1 Np/ni

0:0625 C 3s

0:0625.1 0:0625/ni

;

LCLp;i D Np 3s

Np.1 Np/ni

0:0625 3s

0:0625.1 0:0625/ni

:

Figure 2.15 presents the p-chart generated byMinitab Statistical Software and shows that test 1(one point beyond three standard deviations) occursfor the first sample. This chart also presents the non-continuous trend of the control limits in accordancewith the equations in Eq. 2.20.

• 2.8 Control Charts for Attribute Data 37

17/1114/1111/118/115/112/1130/1027/1024/1021/10

0.11

0.10

0.09

0.08

0.07

0.06

0.05

0.04

0.03

0.02

Day

Pro

port

ion

_P=0.0625

UCL=0.1043

LCL=0.0207

1

P Chart of Rejects

Tests performed with unequal sample sizes

Fig. 2.15 p-chart with unequal sample sizes. Numerical example. Minitab Statistical Software

2.8.3 The np-Chart

This is a control chart for monitoring the number ofnonconforming items in subgroups having the samesize. The centerline and control limits are

Onp D n Np; (2.21)UCLnp D n Np C 3

pn Np.1 Np/;

LCLnp D n Np 3pn Np.1 Np/:

(2.22)

2.8.4 Numerical Example, np-Chart

The data reported in Table 2.12 relate to a productionprocess similar to that illustrated in a previous applica-tion, see Sect. 2.8.2. The size of the subgroups is nowconstant and equal to 280 items. Figure 2.16 presentsthe np-chart generated by Minitab Statistical Soft-ware: test 1 is verified by two consecutive samples(collected on 12 and 13 November). The analyst hasto find the special causes, then he/she must eliminatethem and regenerate the chart, as in Fig. 2.17. Thissecond chart presents another anomalous subgroup:11=11. Similarly, it is necessary to eliminate this sam-ple and regenerate the chart.

Table 2.12 Rejected items. Numerical example

Day Rejects Day Rejects

21=10 19 5=11 2122=10 24 6=11 1423=10 21 7=11 1324=10 23 8=11 2125=10 13 9=11 2326=10 32 10=11 1327=10 15 11=11 3428=10 17 12=11 3529=10 19 13=11 3630=10 21 14=11 1531=10 15 15=11 191=11 16 16=11 222=11 21 17=11 233=11 12 18=11 244=11 25 19=11 27

2.8.5 The c-Chart

The c-chart is a control chart used to track the numberof nonconformities in special subgroups, called in-spection units. In general, an item can have any num-ber of nonconformities. This is an inspection unit, asa unit of output sampled and monitored for determina-tion of nonconformities. The classic example is a sin-gle printed circuit board. An inspection unit can bea batch, a collection, of items. The monitoring activ-ity of the inspection unit is useful in a continuous pro-

• 38 2 Quality Management Systems and Statistical Quality Control

17/1114/1111/118/115/112/1130/1027/1024/1021/10

35

30

25

20

15

10

Day

Sam

ple

Cou

nt

__NP=21.1

UCL=34.35

LCL=7.85

11

NP Chart of Rejects

Fig. 2.16 np-chart, equal sample sizes. Numerical example. Minitab Statistical Software

19/1116/1111/118/115/112/1130/1027/1024/1021/10

35

30

25

20

15

10

Day (no 12 & 13 /11)

Sam

ple

Cou

nt

__NP=20.07

UCL=33.02

LCL=7.12

1

NP Chart of Rejects (no 12 & 13 /11)

Fig. 2.17 np-chart, equal sample sizes. Numerical example. Minitab Statistical Software

duction process. The number of nonconformities perinspection unit is called c.

The centerline of the c-chart has the following av-erage value:

Oc D O.ci / D Nc D 1k

kX

iD1ci : (2.23)

The control limits are

UCLc D Nc C 3pNc;

LCLc D Nc 3pNc:

(2.24)

The mean and the variance of the Poisson distribution,defined for the random variable number of nonconfor-mities units counted in an inspection unit, are

.ci / D .ci / D Nc: (2.25)The density function of this very important discreteprobability distribution is

f .x/ D ex

x; (2.26)

where x is the random variable.

• 2.8 Control Charts for Attribute Data 39

2.8.6 Numerical Example, c-Chart

Table 2.13 reports the number of coding errors madeby a typist in a page of 6,000 digits. Figure 2.18 showsthe c-chart obtained by the sequence of subgroups andthe following reference measures:

Nc D 1k

kX

iD1ci D 6:8;

UCLc D Nc C 3pNc D 6:8 C 3p6:8 14:62;

LCLc D Nc 3pNc D maxf6:8 3p6:8; 0g 0;

where ci is the number of nonconformities in an in-spection unit.

From Fig. 2.18 there are no anomalous behaviorssuggesting the existence of special causes of variationsin the process, thus resulting in a state of statisticalcontrol.

A significant remark can be made: why does thisnumerical example adopt the c-chart and not the p-chart? If a generic digit can be, or cannot be, an objectof an error, it is in fact possible to consider a binomialprocess where the probability of finding a digit with an

28252219161310741

16

14

12

10

8

6

4

2

0

Day

Sam

ple

Cou

nt

_C=6.8

UCL=14.62

LCL=0

C Chart of errors

Fig. 2.18 c-chart. Inspection unit equal to 6,000 digits. Numerical example. Minitab Statistical Software

Table2.13 Errors in inspection unit of 6,000 digits. Numericalexample

Day Errors Day Errors

1 10 16 82 11 17 73 6 18 14 9 19 25 12 20 36 12 21 57 14 22 18 9 23 119 5 24 910 0 25 1411 1 26 112 2 27 913 1 28 114 11 29 815 9 30 12

error is

pi D cin

D ci6;000

;

where n is the number of digits identifying the inspec-tion unit.

The corresponding p-chart, generated by Minitab

Statistical Software and shown in Fig. 2.19, is verysimilar to the c-chart in Fig. 2.18.

• 40 2 Quality Management Systems and Statistical Quality Control

28252219161310741

0.0025

0.0020

0.0015

0.0010

0.0005

0.0000

Day

Pro

port

ion

_P=0.001133

UCL=0.002436

LCL=0

P Chart of errors

Fig. 2.19 p-chart. Inspection unit equal to 6,000 digits. Numerical example. Minitab Statistical Software

2.8.7 The u-Chart

If the subgroup does not coincide with the inspectionunit and subgroups are made of different numbers ofinspection units, the number of nonconformities perunit, ui , is

ui D cin: (2.27)

The centerline and the control limits of the so-calledu-chart are

Ou D O.ui / D Nu D 1k

kX

iD1ui ;

UCLu;i D NuC 3s

Nuni;

LCLu;i D Nu 3s

Nuni:

(2.28)

2.8.8 Numerical Example, u-Chart

Table 2.14 reports the number of nonconformities asdefects on ceramic tiles of different sizes, expressed infeet squared.

Figure 2.20 presents the u-chart obtained; five dif-ferent subgroups reveal themselves as anomalous. Fig-

ure 2.21 shows the chart obtained by the elimination ofthose samples. A new sample, i D 30, is irregular.

2.9 Capability Analysis

A production process is said to be capable when it is instate of statistical control and products meet the spec-ification limits, i. e., the customers requirements. Inother words, the process is capable when it producesgood products. This is the first time the lower andupper specifications are explicitly considered in theanalysis of the process variations.

Nonconformity rates are the proportions of pro-cess measurements above, or below, the USL, or LSL.This proportion can be quantified in parts per million(PPM), as

PPM > USL D P.x > USL/ Pz >

USL OO

;

(2.29)

PPM < LSL D P.x < LSL/ Pz USL 40000.00PPM Total 80000.00

Observed PerformancePPM < LSL 53034.23PPM > USL 43586.49PPM Total 96620.72

Exp. Within PerformancePPM < LSL 53675.43PPM > USL 44166.87PPM Total 97842.30

Exp. Overall Performance

WithinOverall

Fig. 2.23 Capability analysis process 1, numerical example. Minitab Statistical Software

Table 2.16 reports the process data as a result of theprocess improvement made for a new set of k D 20samples with n D 5 measurements each. Figure 2.25presents the report generated by the six-pack analysis.

It demonstrates that the process is still in statisticalcontrol, centered on the target value, 600 mm, and witha Cpk value equal to 3.31. Consequently, the negligi-ble expected number of PPM outside the specification

• 2.9 Capability Analysis 45

191715131197531

601

600

599

Sam

ple

Mea

n

__X=599.972

UCL=600.778

LCL=599.165

191715131197531

3.0

1.5

0.0

Sam

ple

Ran

ge

_R=1.398

UCL=2.957

LCL=0

2015105

601

600

599

Sample

Val

ues

601.5601.0600.5600.0599.5599.0598.5

LSL Target USL

LSL 599Target 600USL 601

Specifications

602600598

Within

Overall

Specs

StDev 0.60121Cp 0.55Cpk 0.54

WithinStDev 0.603415Pp 0.55Ppk 0.54Cpm 0.55

Overall

Process Capability Sixpack of DATAXbar Chart

R Chart

Last 20 Subgroups

Capability Histogram

Normal Prob PlotAD: 0.481, P: 0.228

Capability Plot

Fig. 2.24 Six-pack analysis process 1, numerical example. Minitab Statistical Software

Table 2.16 Measurement data process 2, numerical example

Sample Data process 2 Mean value Range

2.1 600.041 600.0938 600.1039 600.0911 600.1096 600.08788 0.06862.2 599.8219 599.9173 600.0308 600.07 600.0732 599.98264 0.25132.3 600.0089 600.075 600.0148 599.9714 600.0271 600.01944 0.10362.4 600.1896 600.1723 599.8368 600.0947 599.9781 600.0543 0.35282.5 600.1819 600.0538 599.9957 600.0995 599.9639 600.05896 0.2182.6 599.675 599.9778 599.9633 599.9895 599.8853 599.89818 0.31452.7 600.0521 600.1707 599.9446 599.8487 600.012 600.00562 0.3222.8 600.0002 600.0831 599.9298 599.9329 599.9142 599.97204 0.16892.9 600.02 599.9963 599.9278 599.9793 600.0456 599.9938 0.11782.10 600.1571 600.0212 599.9061 599.9786 600.0626 600.02512 0.2512.11 600.0934 599.9554 599.7975 600.0221 599.8821 599.9501 0.29592.12 599.8668 599.8757 600.0414 599.7939 600.1153 599.93862 0.32142.13 599.9859 599.9269 599.8124 600.0288 600.0261 599.95602 0.21642.14 599.9456 600.0405 600.0576 599.7819 600.0603 599.97718 0.27842.15 600.0487 600.0569 599.9321 599.9164 599.9984 599.9905 0.14052.16 599.8959 599.979 600.1418 600.1157 599.9525 600.01698 0.24592.17 600.1891 600.1168 600.1106 599.9148 600.0013 600.06652 0.27432.18 600.0002 600.1121 599.93 599.9924 600.0458 600.0161 0.18212.19 599.9228 600.092 599.9225 600.1062 600.1794 600.04458 0.25692.20 599.7843 599.9597 600.011 600.0409 600.0436 599.9679 0.2593

Average 600.001124 0.23198

• 46 2 Quality Management Systems and Statistical Quality Control

191715131197531

600.1

600.0

599.9

Sam

ple

Mea

n

__X=600.0011

UCL=600.1362

LCL=599.8660

191715131197531

0.4

0.2

0.0

Sam

ple

Ran

ge

_R=0.2342

UCL=0.4952

LCL=0

2015105

600.2

600.0

599.8

Sample

Val

ues

600.9600.6600.3600.0599.7599.4599.1

LSL Target USL

LSL 599Target 600USL 601

Specifications

600.4600.0599.6

Within

Overall

Specs

StDev 0.100682Cp 3.31Cpk 3.31

WithinStDev 0.101659Pp 3.28Ppk 3.28Cpm 3.28

Overall

Process Capability Sixpack of DATA2Xbar Chart

R Chart

Last 20 Subgroups

Capability Histogram

Normal Prob PlotAD: 0.408, P: 0.340

Capability Plot

Fig. 2.25 Six-pack analysis process 2, numerical example. Minitab Statistical Software

limits is quantified as

Total PPM DPz >

USL OO

C Pz USL 6666.67PPM Total 6666.67

Observed Performance

PPM < LSL *PPM > USL 2344.33PPM Total 2344.33

Exp. Overall Performance

Process Capability of dataCalculations Based on Weibull Distribution Model

Fig. 2.26 Capability analysis Weibull distribution, numerical example. Minitab Statistical Software

151413121110987654321

5

3

1

Sam

ple

Mea

n

__X=3.098

UCL=4.926

LCL=1.270

151413121110987654321

3

2

1Sam

ple

StD

ev

_S=1.874

UCL=3.217

LCL=0.532

15105

10

5

0

Sample

Val

ues

9.07.56.04.53.01.50.0

USL

USL 10Specifications

10.001.000.100.01

Overall

Specs

Shape 1.70519Scale 3.47777Pp *Ppk 0.93

Overall

Process Capability Sixpack - Weibull distributionXbar Chart

S Chart

Last 15 Subgroups

Capability Histogram

Weibull Prob PlotAD: 0.248, P: > 0.250

Capability Plot

Fig. 2.27 Six-pack analysis Weibull distribution, numerical example. Minitab Statistical Software

2.9.2.1 Weibull Distribution

Table 2.17 reports data regarding the output of manu-facturing process of tile production in the ceramics in-dustry. This measurement refers to the planarity of thetile surface as the maximum vertical distance of cou-

ples of two generic points on the surface, assuming asthe USL a maximum admissible value of 1 mm.

Figures 2.26 and 2.27 present the report gener-ated by Minitab Statistical Software for the capabilityanalysis. The production process generates products,i. e., output, that are well fitted by a Weibull statisti-

• 48 2 Quality Management Systems and Statistical Quality Control

cal distribution, shape parameter D 1:71 and scaleparameter D 3:48. The process is therefore in sta-tistical control but it does not meet customer require-ments in terms of an admissible USL. In other words,the process is predictable but not capable. In par-ticular, the number of expected items over the USL isabout 6,667 PPM.

2.9.2.2 Binomial Distribution

This application deals with a call center. Table 2.18 re-ports the number of calls received in 1 h, between 3and 4 p.m., and the number of calls that were not an-swered by the operators. The measurement data can bemodeled by assuming a binomial distribution of val-ues. Figure 2.28 presents the results of the capabilityanalysis conducted on this set of values, called dataset 1. The process is not in statistical control becausesample 15 is over the UCL. As a consequence, it is notcorrect to quantify the production process capability.This figure nevertheless shows that the process is dif-ficultly capable, also in the absence of sample 15. Inorder to meet the demand of customers properly it isuseful to increase the number of operators in the callcenter.

191715131197531

0.26

0.24

0.22

0.20

Sample

Pro

port

ion

_P=0.22769

UCL=0.25684

LCL=0.19853

2015105

23.0

22.5

22.0

21.5

21.0

Sample

% D

efec

tive

Upper CI: 0.7605

%Defective: 22.77Lower CI: 22.35Upper CI: 23.19Target: 0.20PPM Def: 227686Lower CI: 223487

Upper CI: 231926Process Z: 0.7465Lower CI: 0.7325

(95,0% confidence)

Summary Stats

200019201840

26

24

22

20

Sample Size

% D

efec

tive

24201612840

8

6

4

2

0

% Defective

Freq

uen

cy

Tar

1

Binomial Process Capability Analysis of no answerP Chart

Tests performed with unequal sample sizes

Cumulative % Defective

Rate of Defectives

Histogram

Fig. 2.28 Binomial process capability, numerical example. Minitab Statistical Software

Table 2.18 Number of calls and no answer, numerical ex-ample

day 1 421 1935 day 11 410 1937day 2 392 1945 day 12 386 1838day 3 456 1934 day 13 436 2025day 4 436 1888 day 14 424 1888day 5 446 1894 day 15 497 1894day 6 429 1941 day 16 459 1941day 7 470 1868 day 17 433 1868day 8 455 1894 day 18 424 1894day 9 427 1938 day 19 425 1933day 10 424 1854 day 20 441 1862

2.10 Six Sigma

Six Sigma stands for six standard deviations andcan be defined as a business management strat-egy, originally developed by Motorola, that enjoyswidespread application in many sectors of industryand services. Six Sigma was originally developed asa set of practices designed to improve manufacturingprocesses and eliminate defects. This chapter presentsa synthetic recall of the basic purpose of Six Sigma,assuming that a large number of the models and meth-ods illustrated here and in the following can properly

• 2.10 Six Sigma 49

support it. Nevertheless, there are a lot of ad hoctools and models specifically designed by the theoristsand practitioners of this decisional and systematic ap-proach, as properly illustrated in the survey by Blackand Hunter (2003).

Six Sigma is a standard and represents a measureof variability and repeatability in a production process.In particular, the 6 specifications, also known as SixSigma capabilities, ask a process variability to be ca-pable of producing a very high proportion of output

Lower specification limit

Upper specification limit

capability 6

x

x

x

x

capability 6

a

b

c

d

Outside lower specification

Out of specification

Out of specification

The process is very capable with 6

• 50 2 Quality Management Systems and Statistical Quality Control

i. e., 0.002 PPM:

1 C6Z

6f .x/dx D 21 .z D 6/

0:00000000198024 2 109; (2.35)where is the standard deviation of the process, f .x/is the density function of the variable x, a measure

Fig. 2.30 Four sigma (Cp DCpk D 1:33) versus SixSigma (Cp D Cpk D 2)

5 10 155 10 155 10 15

0 1 2 3 4 5 6-6 -5 -4 -3 -1-2

six sigma design specification width

mean

LSULSL

Cp=2 Cpk=1.5

Cp= Cpk= 2

Cp=2 Cpk=1.5 virtually zero defects

5 10 150 1 2 3 4 5 6-6 -5 -4 -3 -1-2

5 10 155 10 15

LSU LSL

four sigma design specification width

defectdefect

Cp=1.33 Cp=1.33

1.5-1.5

of the output of the process (process characteristic) x is assumed to be normally distributed and isa cumulative function of the standard normal distribu-tion.

Figure 2.30, proposed by Black and Hunter (2003),compares the performance of a capable process withCp D Cpk D 1:33, known also as four sigma ca-pability, and a process with Cp D Cpk D 2, whichguarantees Six Sigma capability.

• 2.10 Six Sigma 51

2.10.1 Numerical Examples

Among the previously illustrated numerical exam-ples only the one discussed in Sect. 2.9.1 (process 2)verifies the Six Sigma hypotheses, because Cp DCpk D 3:31.

2.10.2 Six Sigma in the Service Sector.ThermalWater Treatmentsfor Health and Fitness

In this subsection we present the results obtained bythe application of the Six Sigma philosophy to thehealth service sector of thermal water treatments. Thisinstance demonstrates how this methodological ap-proach is effective also for the optimization of ser-vice processes. In particular, in this case study severalhealth and fitness treatments are offered and they are

Customer contact

Boo

king

Boo

king

/Rec

eptio

nC

usto

mer E-mail

deliveredRemote request

Fax delivered

Local request

Call Missed request

Recognize customer request

Phonerequest?

Unoccupiedcollegue?

Forward to a collegue

Luckycustomer?

Call customer back

Apply phone number for recall

Acquaintcustomer

yes

no

Consistentrequest?

no

yes

no

no

no yes

yes

Being out tobook?

Missed booking

RetrieveCustomer

Master Data

no

yes

Customerin Master

Data?

Input Customer Data

no

noSingle customer?

Request for wellness?

Requestfor thermal treat-

ment?

Single hotel reservation

Single thermal reservation

Single wellness reservation

yes yes

noyes

yes

no no Is it a group?Request forwellness?

no

Group thermal reservation

Group wellness reservation

Group hotel reservation

yes yes

yes

Tour operator reservation

no

Legend: Event Activity Note Decision Process Subprocess

Take customer request

Requestaccepted?

yes

Requestfor thermal treat-

ment?

Fig. 2.31 Booking procedure

grouped in three divisions, each with a proper bookingoffice and dedicated employees: hotel, wellness, andthermal services.

Employees are nominated to have contact with thecostumers, to identify their requirements, to accept therequests, and to finalize the booking process. Cus-tomers can have contact via telephone, e-mail, Website, or by presenting themselves at the reception. Ev-ery kind of service has its own booking procedure, de-pending on the customer request. Before the applica-tion of Six Sigma methodologies the process was di-vided into the following five subroutines, dependingon the service:

single thermal booking; group thermal booking; single hotel booking; tour operator hotel booking; wellness booking.

• 52 2 Quality Management Systems and Statistical Quality Control

Once the booking procedure has been completed thestaff will wait for the customer. On his/her arrival, therelated booking data are recalled from the system andthe customer is sent to the so-called welcome pro-cess, which is common to hotel, wellness, and ther-mal services. By the next check-in stage the customeris accepted and can access the required service. Thereis a specific check-in stage with its own dedicated rulesand procedures for every kind of service. Once the cus-tomer has enjoyed the service, he/she will leave thesystem and go to the checkout stage, with its own pro-cedures too.

The whole process, from the admittance to theexit of the customer, can be displayed as a flowchart;Fig. 2.31 exemplifies the detail of the booking proce-dure.

The analysis of the whole process has emphasizedthe existence of significant improvement margins, re-lated to costs and time. For example, a particular ser-vice, e. g., thermal mud, may need a medical visit be-fore the customer is allowed to access the treatment.By the Six Sigma analysis it was possible to reduce thelead time of the customer during the visit, through theoptimization of the work tasks and processes. Some-times this can be performed by very simple tricks andexpedients.

For example, the aural test can be invalidated bythe presence of a plug of ear wax in a patient. Teach-ing the technician how to recognize and remove thisobstruction reduces the probability of null tests, andconsequently there is a reduction of costs and leadtimes.

• http://www.springer.com/978-1-84882-574-1