200 000 users 8 000 corporate clients 55 countries 21 languages

200 000 Users8 000 Corporate Clients

55 Countries21 Languages1 Goal: Satisfied Customers

A Publication by Q-DAS® GmbH & Co. KGand TEQ Training & Consulting GmbH

ISSN 0949-958X

Issue 2013/2014

25 Years

Information for Innovative Quality Management International Issue

PARTNERINFOQUALITY

b PIQ International

Contents

Categories

1 Editorial/Advertisers1 Publication Details78 Q-DAS® Subsidiaries

and Distributors world-wide

Q-DAS® Products

Q-DAS® News

25 Q-DAS® Software alsoavailable in Thai

27 Global Q-DAS®

Network 28 Tom Stewart Receives

Ford Global PTM Eng.TMM Excellence Award

29 Q-DAS® SoftwareComplies with FDA’s21CFR Part 11 Standard

20 Q-DAS CAMERA®

Concept30 Specification Limits and

Costs31 Limits in Measurement

Process CapabilityAnalyses

34 Reducing the Uncer tain -ty through suitableMeasurement Processes

35 A Simple Introductionto the World of Q-DAS®

and the CAMERAConcept

42 Software Validation

48 Interaction of Q-DAS®

Software Products andMES Solutions

50 Version 1154 Quality in Manu -

facturing, Two indus-tries - Same problem

56 Evaluation of Indivi -duals Based on theExtended ToleranceClassification by UsingQ-DAS® Products

58 AQDEF® – AdvancedQuality Data ExchangeFormat

TEQ® Training & Consulting

64 Risks of an IntuitiveDefinition of ControlLimits for a ShewhartControl Chart

67 Which Information doCapability IndicesProvide?

71 An Approach toProcess Improvement -Multiple LinearRegression

75 Roles andResponsibilities in SixSigma Projects

76 The Crux of the ndc

Cover Story

2 25 Years Q-DAS - anInterview with EdgarDietrich

4 25 Years Q-DASGermany

14 Statistical Procedures inthe Course of Time

44 BMW MINI Project inEngland

46 OCB: Have a DifferentLook at Your MeasuredData

Theory and Practice

60 Interaction of iqs and Q-DAS® Software Products

62 Coordinate Metrology,Statistic and QualityManagement throughthe ages - QUINDOS®

congratulates Q-DAS®

Specialist Articles

200 000 Users8 000 Corporate Clients

55 Countries21 Languages1 Goal: Satisfied Customers

A Publication by Q-DAS® GmbH & Co. KGand TEQ Training & Consulting GmbH

ISSN 0949-958X

Issue 2013/2014

25 Years

Information for Innovative Quality Management International Issue

PARTNERINFOQUALITY

1PIQ International

Editorial

Dear Readers,

"Statistics Drives Success"

I cannot think of a better motto toCelebrate the 25th year anniver-sary of Q-DAS® GmbH & Co. KG.For 25 years now, we have beenserving the global manufacturingindustry with integrity, listening tothe voice of the customer and pro-viding creative solutions that areutilized now in 55 Countries and21 National languages.

As I reflect now on the many chal-lenges faced and solved, the result-ing relationships formed and thegrowth of the Q-DAS® Family ofemployees, I am amazed at the lev-el of synchronization between allthese components. I wonder abouthow the systematic topic of

Statistics has fostered such valuedHuman relationships across theglobe. We place a high value oneach of these relationships and rec-ognize the great responsibility thatwe have to deliver the very bestsolution for each application whilemaintaining the core values of aWorld Class Organization.

This year, with the release ofVersion 11 we enter into a new eraof diverse technology. The chal-lenges to come are not yet known,but the process of solving them isknown to us and is part of ournature. Keeping an open and cre-ative mind, maintaining the idealsof the many International Norms

and Company Guidelines andalways, always building the rela-tionship among our Customers andPartners. This is our nature.

So as we Celebrate our 25th year,please join with us and know thatwe recognize the vital role ourCustomers and Partners have inour success. We are committed todemonstrating every day, howStatistics can Drive the Success ofyour organization and look forwardto the challenges and solutions thatwill give us the opportunity to havea meaningful role in the success ofyour organization.

Advertisers

26,69 ScanALLY63 Hexagon Metrology PTS61 iqs Software GmbH

29 STEINWALD datentechnikGmbH

On behalf of the Q-DAS® und TEQ® Team

Publisher:Dr.-Ing. Edgar DietrichQ-DAS® GmbH & Co. KGEisleber Str. 2 69469 Weinheim

Editors:Stephan Sprink, Thomas StewartQ-DAS® GmbH & Co. KG

Layout:Heide MesadQ-DAS® GmbH & Co. KG

Printing Press:DruckhausDiesbach GmbHWeinheimCirculation: 1,500ISSN 0949-958X

Publication Details

2 PIQ International

Cover Story

PIQ: It is the 25th anniversary of Q-DAS® so theremust be plenty to tell!ED: Yes, indeed, there certainly is. Pages 4 to 13 providea well-arranged summary of the major milestones in the25 years of our company history. You can see the devel-opment of Q-DAS® perfectly.

PIQ: But there are certainly some particularities youwould like to point out.

ED: There are three aspects that have helped Q-DAS®

become what it is today. First of all, it is our range ofproducts, i.e. the software and the corresponding ser-vice. Secondly, it is the internationalization and last butnot least it is the corporate culture of Q-DAS®, i.e. theway we deal with one another and how we support ourcustomers every day.

PIQ: What is special about Q-DAS® products?ED: At first, our cofounder Alfred Schulze, who wasresponsible for software development during the first 20years, perfectly knew how to apply statistical proce-dures in a way that users are able to fulfill their tasks inpractice. The automatic evaluation and the user-specificgraphical visualization of results provide the basis ofthese functionalities in our software. Our functionalitieshelp users to describe, assess and evaluate theirprocesses and issues comprehensibly by means of sta-tistical procedures. There is not any other software onthe market doing it better than our software. On, thisbasis, we developed all the products offered by Q-DAS®

today. All our software products access the availablenumeric and graphics library we evaluate and extendcontinuously.

PIQ: When did the internationalization of Q-DAS®

start?ED: When we founded the company in 1988, we did noteven anticipate that we might ever need the software inany other language than German. This is the reason whythe source code contained our texts as was the casewith almost every other software back then. However,something happened in 1990. The INA company inHerzogen aurach had already purchased several qs-STAT® licenses and applied them successfully. So whenI visited INA in 1990, the former quality manager toldme that the company wanted to use our software world-wide in a total of 20 plants. “But we need the softwarein several languages”, he told me and listed 10 differentlanguages. At this time, it became clear to me that Q-DAS® will only stay competitive if we are able to pro-vide the software in several languages and offer the

required support in the respective countries, i.e. train-ing, consulting and, of course, distribution of our soft-ware. Luckily, Alfred Schulze had the same opinionabout these requirements as I did. He redesigned thesoftware in a way that it supported various languages in1992. What he had done proved to be a masterstrokesince, still today, the Q-DAS® software can switch thelanguage of the graphical user interface, masks andprint-outs in real time. As an example, you can make anevaluation in Chinese but show or print the results in adifferent language. This is a huge advantage for a globalcompany and particularly for suppliers. The switch oflanguage provided the basis for our internationalizationand we first started to create a partner network outsideof German-speaking countries. Later, we founded ourown subsidiaries in strategically important markets. Dueto our efforts, we are able to support software installa-tions in 55 countries today.

PIQ: You talked about the corporate culture of Q-DAS®. What do you mean by this?ED: In addition to lean processes, it is important to rec-oncile employee satisfaction, product or service qualityand customer satisfaction. These three aspects arelinked closely to one another and, in my opinion, allhave the same significance. You can never observe one

Dr.-Ing. Edgar Dietrich, Q-DAS® GmbH & Co. KG

25 Years Q-DAS - an Interview with Edgar Dietrich (ED)

3PIQ International

Cover Story

of these without considering the other two aspects sinceevery single aspect affects the other two. We do notconsider these three aspects to be a triangle relationshipbut a control cycle working properly for years. We offerour products and services to the customers. Theybecome acquainted with both, product and service, andappreciate our expertise and the cooperative collabora-tion. Customers do not feel left alone and notice that theQ-DAS® software is highly accepted by users and theirown customers. My personal slogan is: “Q-DAS® cus-tomers trust us because of the reliability they experi-enced!” And the positive relation between customersand suppliers also benefits employee satisfaction, ofcourse, and frequently leads to recommendations whichare the best kind of advertising.

PIQ: What are you particularly proud of?ED: Humbly, I do not want to use the word ”proud“.However, there are many things I am pleased with. Theemployees of the entire Q-DAS® Group rank firstbecause they care for our customers around the worldday and night, every day. I am also glad that our firstcustomer – Autoliv – still applies Q-DAS® productstoday. Since we have acquired more than 300 new cus-tomers per year for years, I am quite confident of ourfuture.

PIQ: When talking about the future, what will Q-DAS® look like in 25 years?ED: Fortunately, nobody can gaze into the crystal balland see the future, least of all the future in 25 years’time. Otherwise, life will be boring. However, what wecan do very well is defining a strategic alignment for thecompany based on past experiences and current marketinvestigations. This is what we did for Q-DAS® until2017. We have to check the chosen strategy continu-ously, of course, to find out whether it still makes senseand sometimes we have to adapt it. However, if we stillsucceed in maintaining the corporate culture of Q-DAS®

I mentioned before I am not worried about Q-DAS®.

PIQ: Coming back to the subject of product, after allthere have been many changes in the past 25 years?ED: Yes, there are also three individual aspects. First ofall, the IT environment has changed fundamentally fromDOS to Windows. This new environment gave us theopportunity to realize a comprehensive system includ-ing data collection, data management, data evaluationand the distribution of information to respective work-stations. And, of course, the application of statisticalprocedures has changed, too.

PIQ: How significant are the influences of the IT envi-ronment?ED: The IT environment issues a new challenge to usevery day. Technically speaking, it was all very easy 25years ago. Back then, there were the DOS, OS/2 andUNIX operating systems but we knew that most of ourcustomers only applied DOS. Nowadays, we have tosupport all the different releases of the Windows oper-ating system as well as the 32-bit and 64-bit versions. Inaddition, tablets and smartphone including an ownoperating system become increasingly popular and peo-ple use web applications. However, this variety and per-manent changes are not only a challenge but our soft-ware house benefits from this development. Our cus-tomers also go with these trends and, thus, they ordernew versions of our software or conclude maintenancecontracts in order to be always up-to-date.

PIQ: How did statistical procedures change?ED: The basic statistical procedures were not reinventedover the last 25 years, of course. However, the way peo-ple apply them changed. I already mentioned the auto-matic evaluation. We have adapted our software contin-uously to various customer requirements and today, weare able to better cater to specific demands in order tosatisfy the respective customer wishes. Data manage-ment plays an important role. All required informationmust be available to analyze data from different per-spectives. This is the only way to provide answersregarding processes and different issues. We succeededbrilliantly in doing so in recent years.

PIQ: Q-DAS® often participates in creating guidelinesand standards, too?ED: That is right. These activities for DIN, ISO, VDA,VDI or any conglomerates helped us a lot in becomingwhat we are today. Since we are in contact with decisionmakers, we are among the first to learn about new stan-dards and know very well how to implement the con-tents of these documents in our software. However, weparticularly want to intensify our efforts to participate inthe creation of international ISO standards. As an exam-ple, ANSI is likely to transfer the ISO/TC 69 SC 4 secre-tariat (applied statistics) to DIN in the coming monthsand I will take the chair

PIQ: Thank you for the interview and all the best forthe next 25 years!

25th Anniversary of Q-DAS® GermanyHow it all began... – the first years

1988: Motivated by the feedback ofparticipants of the first computer-assisted statistics seminar offeredby the DGQ in February 1988, uni-versity lecturers Edgar Dietrich andAlfred Schulze founded the “GDSGesellschaft für Datenverarbeitungund Systemtechnik mbh” in Birkenauon November 25. Christine Dietrichwas appointed manager of the com-pany.

1993: Edgar Dietrich and AlfredSchulze took over the managementin January. Since then ChristineDietrich has been responsible foraccounting and human resources.After another change of name hadbeen required, henceforth the com-pany operated under the familiarname ”Q-DAS® GmbH“.

1995: The transfer of knowledgehas always played an essentialrole for Q-DAS®. The Carl HanserVerlag published the first editionof the book “StatisticalProcedures for Machine andProcess Qualification“ by theauthors Dietrich and Schulze in1995. Today, it is regarded as astandard reference and its 6th edi-tion has already been published.

1991: Both founders of the compa-ny Edgar Dietrich and AlfredSchulze finally quit teaching at theBA Mannheim and BFW Heidelbergand started working full-time forthe company renamed as “GeDaSGesellschaft für Datenverarbeitungund Systemtechnik mbH” in themeantime.

1996: Q-DAS® and several othercompanies developed the Q-DASASCII transfer format together.This data format provided thebasis for a cross-system exchangeof quality information and soonestablished itself as an industrystandard. Since 2006, Q-DAS® andan industrial work group have con-tinued to develop the format. It isknown as AQDEF (AdvancedQuality Data Exchange Format)today.

1994: Being a quality-orientedcompany, Q-DAS® implemented aQM system based on EN ISO 9001early. DQS audited the systemsuccessfully. So Q-DAS® wasamongst the first German soft-ware houses possessing a certifiedQM system.

2004: Q-DAS® convinced the juryand earned the “TOP 100 –Excellent Innovators in the GermanMiddle Class“ award for their innov-ative achievements.

2007: Edgar Dietrich and AlfredSchulze made it to the finals ofthe “Entrepreneur of the Year2007” inter-company competition– an appreciation of entrepreneur-ial performance.

2000: Due to a lack of space inthe former family-owned hotel inBirkenau/Nieder-Liebersbach, thecompany moved into the newoffice building in Weinheim provid-ing plenty of space for trainingparticipants and employees.

2007: Alfred Schulze retired fromactive business and left the compa-ny at the end of the year. EdgarDietrich took over the companyshares and became the sole owner.

2006: Since Q-DAS® has continuedto expand and to go global, therestructuring of the Q-DAS® Groupwas required: foundation of Q-DAS®GmbH & Co. KG and Q-DAS® AssetGmbH as part of Q-DAS®Verwaltungs GmbH; foreign Q-DAS®subsidiaries now belonged to Q-DAS® Holding GmbH.

25th Anniversary of Q-DAS® GermanyFrom Local to Global Player – the Q-DAS Group

In order to satisfy demands of the interna-tional market, Q-DAS® cooperated withforeign partners and founded subsidiaries inthe Czech Republic (1996), USA (1998),Italy (2002) and France (2005). Branchesin China (2006), India (2013) and Korea(2013) are in charge of the growth marketsin the Asia-Pacific region.

Q-DAS Incorporated, USA

Q-DAS s.r.l., Italy

Q-DAS SARL, FranceQ-DAS Software Technology, China

Q-DAS spol s.r.o., Czech Republic

Since 2008, Q-DAS Academy GmbH hasbeen responsible for trainings and semi-nars. The TEQ® Technologietransfer &Qualitätssicherung GmbH, Chemnitz, alsojoined the Q-DAS® Group in the sameyear. In 2010, both companies, Q-DASAcademy GmbH and TEQ®, merged intoTEQ® Training & Consulting GmbH head-quartered in Weinheim with offices inChemnitz and Berlin.

2008: Management of Q-DAS® and TEQ®

TEQ® headquarter in Weinheim

TEQ® office in Chemnitz

TEQ® office in Berlin

2013: TEQ® strategy workshop

25th Anniversary of Q-DAS® GermanyQ-DAS QM Tools – statistics for knowledge acquisition

1989: The first complete qs-STAT®version including the Sample Analysisand Process Capability Analysis mod-ules was compiled with Turbo PascalCompiler 3.0 and saved to a 5 ¼“floppy disk. First customer was theAutoliv company in Elmshorn.

1999: procella monitoring addeda new tool to the product line.This tool offered real-time visual-ization of product and processdata in the Q-DAS® ASCII trans-fer format.

1992: The first multilingual ver-sion of the software opened thedoors for Q-DAS® to enter theinternational market in February1992. At first, the software wasonly available in German andEnglish; however, today it sup-ports more than 20 languages.

2001: The 32bit version qs-STAT®millennium and procella® millenniumrepresented the shift from a purestatistics tool to a comprehensivesystem for the recording, visualiza-tion, automatic evaluation andreporting of product-related qualityinformation. Data management in acentral database provided a new,comfortable way of data processing.

1996: qs-STAT® 3.0 pavedthe way for the change fromDOS to Windows.

2002: The procella –My.SPC campaign estab-lished the software as aleading SPC tool in industrialproduction. A variety ofinterfaces, user-friendlyvisualization and task-orient-ed measurement processcontrol were the character-istic features of procella®.

2003: MCA/CMM Reportingcomplemented the tools fora production-related onlinevisualization and evaluation.

2006: The Q-DAS CAMERA® Conceptwas designed as a set of tools consistingof software and services for the creationof a performance measurement system. Q-DAS first presented it to the profession-al audience at the Control internationaltrade fair.

2008: The new product structure of version ME 8 adapted tothe tasks of the users. There were software packages avail-able for any important application. They provided practice-oriented solutions without any complicated additionaloptions. In addition to the standard tools qs-STAT®, procel-la®, solara and destra®, the CAMERA tools O-QIS, M-QIS andQ-DBM supplemented the software package.

2004: Since version ME 4, the newmodules for data compression andlong-term analysis are added to thesoftware solution. This leads to a con-sistent information and reporting sys-tem of quality statistics in industrialproduction.

2007: Version ME 7 offereddestra as a new statistical pack-age including modules like DOEand Analysis of Variance. Thistool for experts was mainlydesigned for users in the SixSigma environment.

2013: M-QIS Dashboard enhances theoptions to visualize statistics clearly inthe form of an interactive web appli-cation meeting the demands of therespective target group. Version 11provides a modern and user-friendliergraphical user interface. It is availableas a 32-bit or 64-bit version.

25th Anniversary of Q-DAS® GermanyA strong Team

Q-DAS® becomes celebratory: staff on the occasion of the 20th anniversary of Q-DAS® (2008)

Q-DAS® Startup (1988-92)

The first manager and the only employ-ee during the first years was ChristineDietrich. In 1990, the companyemployed Gerlinde Kinscherf as anoffice worker. After another year, LeilaDietz and Robert Disser reinforced theteam in the fields of software deliveryand support/training. Heide Mesad com-plemented the “original team” in 1992.

at Q-DAS® since …Edgar Dietrich 1988Christine Dietrich 1988Gerlinde Kinscherf 1990Heide Mesad 1992Michael Wagner 1994Martin Oswald 1997Thomas Götz 1998Stefan Weber 1998Ulrike Wehebrink 1998Stephan Sprink 1999Oliver Kleefoot 1999Michael Sommer 2000Thomas Clemens 2000Marco Franz 2001Michael Roth 2002Michael Radeck 2002Frank Paetzold 2004Boris Schiller 2004Klaus Tasch 2005Martin Werner 2005Thomas Gastgeb 2005Markus Pfirsching 2006Dirk Dingler 2007Marco Hardardt 2007Inge Welker 2007Mike Pfeiffer 2008Nadja Dvoretski 2010Malte Steinkopf 2010Matthias Gerstenkorn 2011Thomas Schäfer 2011Thomas Schäfer 2011Stephanie Dietrich 2011Uwe Brang 2011Alexander Rudolf 2011Mubarak Bai 2012Ming Wu 2012Johannes Pudleiner 2012Melanie Feuerstein 2012Daniel Hamer 2013Andreas Diefenbach 2013

Q-DAS® gets active: geocaching (2013)

Q-DAS® is present: trade fair team at Control 2002

Q-DAS® goes sporty: BASF company cup (2013)

Q-DAS® turns ladylike: Control 2008

Due to the continuous growth, Q-DAS®employed further staff working in differentfields in the following years. Today, a totalof 39 employees work for Q-DAS® GmbH& Co. KG. Team leaders are responsiblefor the single departments and support themanagement.

25th Anniversary of Q-DAS® GermanyFit for the future – The experts in statistics establish standards. Worldwide.

For 25 years now we have listened closely to our cus-tomers’ suggestions. We understand their requirements andconvert them into market-leading solutions. Thus our enthu-siastic customers position our company in the market.

Q-DAS® channels data streams in industrial production inorder to acquire knowledge. The instruments we developare able to convert characteristic values and processparameters reliably into statistics and to convey them.

We pave the way for a structured, customized evaluationand control of industrial processes. Our software productsand know-how for the creation of performance measure-ment systems ensure that the entire potential of statisti-cal evaluations may contribute to the increase in efficien-cy at any time – a valuable acquisition of knowledge lead-ing to the success of our customers.

Our values provide strength. The aim of our company iscontinuous personal growth and steady economic develop-ment. Our success shall be sustainable and create satisfac-tion. For this reason, clear values form the basis of ouractions.

In our vision, no industrial product all over the world is pro-duced without encountering Q-DAS® products directly orindirectly. This applies to the manufacturers of end prod-ucts and their suppliers but also to contractors of machinesand production facilities or measuring instruments.

Unique worldwide. About 200 000 users trust in the Q-DAS® software and apply our tools for enhancing prod-uct and process quality successfully.

More than 8 000 business customers in 55 countries havealready decided to apply Q-DAS® software products andtap the potential of correct and reliable statistical evalua-tions profitably.

21 supported languages overcome language barriers. Thesame applies to our network of 28 subsidiaries and part-ners ensuring professional advice, installation, support andtraining all over the world.

Before 1980Statistical procedures had already been applied before Istarted to encounter them. Here I would like to refer tothe lecture of professor Masing that he gave on theoccasion of the Q-DAS® user meeting on November 26,2003. Unfortunately, this was his last public lecturebefore his death. The subject of his last public presenta-tion was “Statistics as a Fundament of a Quality MethodBased Way of Thinking and Acting“. I would like to rec-ommend this lecture to anyone interested in the historyof statistical procedures. Find his lecture in the 5th edi-tion of his “Handbuch Qualitätsmanagement“ (qualitymanagement). It is also available as a PDF file or videoon the Q-DAS® website under http://www.q-das.de/de/kompetenzcenter/masing-video/.

Principles of Statistical ProceduresIn the middle of the 1990s the following books coveredthis topic in Germany: ● “Formeln und Tabellen der angewandten mathema-

tischen Statistik“ (formulas and tables for appliedmathematical statistics) by Graf, Henning, Stange,Wilrich, Springer Verlag

● “Statistische Verfahren für Technische Messreihen“(statistical procedures for technical measurementseries) by Dr. John, Carl Hanser Verlag.

● “Statistical Methods for Quality Assurance“ byRinne, Mittag, Carl Hanser Verlag.

These books were quite theoretical and initially hard tounderstand for readers from practice.

When dealing with statistical procedures in industrialproduction, you have to mention the American physicistWalter Shewhart. While working at Bell TelephoneCompany he developed statistical procedures for theevaluation of quality-related processes and applied themin practice. Thus he is considered to be the “father ofSPC” (Statistical Process Control).

Walter Shewhart, born in New Canton, Illinois, on March 18,1891; † in Troy Hills, New Jersey, on March 11, 1967

Still today both his books ● Economic Control of Quality of Manufactured

Product (1931) ISBN 0-87389-076-0 or ISBN 978-0-87389-076-2

● Statistical Method from the Viewpoint of QualityControl (1939) ISBN 0-486-65232-7

provide the basis for the currently applied procedures.

While working in the ISO/TC 69 technical committee, Ihad the opportunity to meet Jack Kayser, the formerchairman of SC 4 who had worked together with WalterShewhart before he died.

Ford was the company to issue the Q101 quality systemstandard at the beginning of the 1990s. It was used forin-house productions and it applied to purchase partsuppliers satisfying the demand for production parts andspare parts. Q101 included the SPC topic (StatisticalProcess Control) based on the theory of WalterShewhart. An own instruction guide explained this topic


Statistical Procedures in the Course of Time

I would like to give an overview of changes in the application of statistical procedures in industrial productionover the last 30 years

Professor Dr. Walter Masing

14 PIQ International

Theory and Practice

15PIQ International

Theory and Practice

in detail. Supplementary guidelines such as “StatisticalProcess Control for Dimensionless Materials“ or “BasicStatistics Using SAS“ were accompanying documentswith which the suppliers of the Ford company had todeal. Audits were used to monitor the extent to whichthe described procedures were implemented andapplied. The application of SPC was a substantial part inthe evaluation of suppliers. Ford Europe created an ownposition in England for the compilation and the continu-ous development of this standard within their company.In the 1990s, Manfred Martelock held this position.

In Germany German statisticians initially challenged SPC and its cor-responding statistical procedures more strongly than inthe USA. They mainly stuck to the following philosophy:“You have to adjust processes in a way that they are nor-mally distributed and that they indicate their stability inthe Shewhart quality control chart.“ Deviations wereimpermissible or had to be an absolute exception.However, this is often unrealistic or not feasible at all forvarious reasons. If you implement the specified proce-dures exactly, record the values of the characteristics bymeans of a measurement system, calculate thedescribed statistics and compare them to the specifiedlimits without questioning whether your results agreewith reality, these results might lead to serious misinter-pretations.

Many German companies compared their results gainedaccording to the philosophy mentioned above with real-ity and had to find out that there was no agreement. Asan example, they performed a 100% inspection and cal-culated a capability index of 1,39, but in reality they hada proportion of rejects that amounted to 10% of theirproduction. Vice versa, they actually had a proportion ofrejects that amounted to 0% of the production in a 100%inspection, but they calculated a capability index of 0,9.The discrepancy between calculated results and realitywas obvious and unacceptable, so this problem causedmany debates on e.g.:

● Which probability distribution does apply?● How do we handle one-sided characteristics?● Do the stability criteria of the Shewhart control

chart also apply to not normally distributed mea-sured values?

● Which formula is applicable in order to calculatethe capability indices?

● Which limits are suitable for which characteristics?● etc.

Since there was no answer to these questions but thesuppliers were still obliged to implement SPC, theyoften just selected characteristics that could be shownto the customer without feeling bad about it. For thisreason, the term “SPC” unfortunately became known asa ”Show Program for Customers“.

DGQ Sets the Course for Statistics inGermanyThe German Society for Quality (DGQ e.V.) in Frankfurtstrongly influenced the topic of statistics in Germany. Inthe 1990s, the DGQ offered the QII training, a four-week training about various statistical subjects (statisti-cal basics, test procedures, sample systems, quality con-trol charts and probability distribution). RainerFranzkowski, who unfortunately died too young, wasresponsible for the contents of this training.

In terms of statistics, his approach was the ultimatesolution in Germany back then. He also worked a lot inthe field of national and international standardizationand he participated in creating several statistical stan-dards. At the end of his QII training, there was an examand participants got the QII certificate. Anyone whowanted to gain further qualifications attended a one-week training in order to become an instructor. Theadvantage of this qualification was a broad knowledge in

Manfred Martelock, formerly Ford Europe

Edgar Dietrich, Alfred Schulze, Rainer Franzkowski


Theory and Practice

statistical procedures; considerably broader than thespecifications given in the SPC Manual issued by theFord company. The correct application of statisticalmethods provided the answer to the discrepancybetween the calculated capability indices and reality asmentioned before. However, a software based on theseconclusions has not been available yet.

Even before the foundation of the Q-DAS® company,Alfred Schulze and I created software componentsbased on general statistical basics due to our knowledgefrom this statistical environment. When our qs-STAT®

product entered the market in 1989, there were alreadymore than 100 SPC software programs available. Theseprograms were offered by the manufacturers of SPC andCAQ systems and by manufacturers of measuringinstruments. This variety was available for reasons ofdata processing. At that time the DOS operating systemwas the state of the art. The communication betweentwo different programs as we know it from the Windowsenvironment today was unusual and technologicallyhard to realize back then. This was the reason why therespective providers were virtually forced to have theirown SPC software. The main function of this kind ofsoftware was the calculation of capability indices basedon a normal distribution and the display of values in aquality control chart according to Shewhart’s theory. Allof them complied with the guidelines of the Ford com-pany, however, none of them provided any answer orsolution to the problems described above.

Q-DAS® Enters the MarketWe offered a wide selection of statistical procedures inqs-STAT®. Consequently, be it machine acceptance orlong-term studies of processes, we were able to providethe statistical procedures for the real situation leading totraceable and correct results in practice. Here are twotypical examples:● One-sided characteristics tend to be not normally

distributed due to their physical properties (typicalexamples are form and location parameters andcoat thicknesses). Thus you cannot use a normaldistribution in order to determine the capabilityindices. Only unimodal skewed distributions pro-vide realistic capability indices, such as the logarith-mic normal distribution, Pearson or the Johnsontransformation.

● In case of tools including several cavities, the varia-tion of the parts within a cavity is normally relative-ly small; however, the location of the respectivecavities strongly differs due to manufacturing toler-ances. Thus it is not possible to use any Shewhartchart in the classical sense. The same applies to thecalculation of capability indices. In order to be ableto calculate realistic capability indices in these situa-tions and in order to monitor processes by meansof reasonable control limits, Q-DAS® refined the

mixed distribution and the Shewhart charts bymeans of extended limits.

When we presented our software in seminars and at theplants of our major customers, the participants quicklyrealized that this tool provided the correct answers tothe questions that have not been solved yet. This factraised the acceptance of the SPC application. During thefollowing years, our conglomerates created more andmore guidelines hence including the functionalitiesoffered by Q-DAS.

Ford Develops Software Test ExamplesEven Ford quickly realized that the people’s believe inthe correctness of everything a computer does - whichwas quite common these days - was inappropriate forthe application of SPC. The SPC systems of variousproviders available on the market back then could notdescribe the actual state of a process run correctly andwithout restrictions. However, the release of the supple-mentary Ford guideline “Process Capability Studies“provided the basis for calculating significant and correctcapability indices even in case of not normally distrib-uted process results. Together with the Ford test exam-ples, which had been developed by the Q-DAS® compa-ny and Mr. Martelock in his position for Ford Europequality methods, it became easy to evaluate whetherSPC systems were qualified for the intended purpose.

The test examples are datasets describing differentprocess types or process situations. By loading the pub-lished test data into the evaluation software you want totest, you just have to compare the numerical and graph-ical results in order to verify the qualification of the eval-uation software. Since Q-DAS® helped to develop thesetest examples, the use of these examples increased theQ-DAS® brand awareness in Europe by leaps andbounds. All producers of SPC systems now had to pro-vide evidence that they implemented the test examplescorrectly. For the calculation of capability indices in caseof Ford purchased parts, neither Ford nor their suppliersaccepted software producers who were not able toprove that they passed the test. Correct calculationresults, significant graphical displays and, of course, auser-friendly handling laid the foundation to extend thenewly attained qs-STAT brand awareness even furtherthrough excellent performance.

Hartmut M.W. Nowack Provides theEvidenceSince 1998, Hartmut M.W. Nowack of the Mercedes CarGroup has been in charge of a representative studyincluding more than 1000 different process situations.He obtained the respective files from different compa-nies and from different applications. The detailed analy-sis and evaluation of the files showed what our gut feel-ing already indicated: Only 2% of the processes are nor-

17PIQ International

Theory and Practice

mally distributed and stable according to Shewhart qual-ity control charts.Due to the tools andstatistical proceduresof Q-DAS® imple-mented in qs-STAT®

we were able todescribe and evaluatethe major part of allprocesses correctly.The results we gainedwere the quality capa-bility indices Cp andCpk as well as Pp andPpk that were close toreality. The relationbetween theory andpractice became com-prehensible.

Uniform Data Format Leads to the Efficiencyof SPCAt the beginning of the 1990s, computer systemschanged from DOS to Windows. This was a big chal-lenge for many software houses since the Windowsoperating system demanded different developmentsfrom software engineers than DOS had done. In addition, Windows provided the option to start and runseveral programs at the same time. Since many produc-ers of SPC systems did not want to make the effort toport their software from DOS to Windows and since the qs-STAT® brand awareness continued to rise, many pro-ducers decided to cooperate with Q-DAS®. Instead ofdeveloping their own software, they recommended qs-STAT® as a software package for statistical evaluationsto their customers or even integrated qs-STAT® into theirown software.

The communication between the third-party system andqs-STAT is based on the Q-DAS® ASCII transfer formatthat was initially developed by Q-DAS® together withconglomerates such as the Mercedes Car Group orFord. Today this data format is an international standardfor the exchange of quality information. Several compa-nies have made this format their standard and still main-tain this standard. Today this data format is referred toas AQDEF (Advanced Quality Data Exchange Format).Please find more information about this format onwww.q-das.de in the ”Service“ section under “Data for-mat”.

Q101 Becomes QS-9000In 1994, the so-called “Big Three” (Chrysler, Ford andGeneral Motors) published the Quality SystemRequirements QS-9000 standard in the USA. Theserequirements only included the application of SPC andMSA since the implementation of these procedures wasalready described in the reference manuals

”Fundamental Statistical Process Control“ and”Measurement System Analysis“. Particularly the SPCManual was widely based on the Ford SPC standard. TheMSA Reference Manual was a newly created standardwhich was considerably more comprehensive than theMSA guidelines of the Ford Company. After the releaseof QS-9000 at the end of the last century, suddenly allsuppliers of Chrysler, Ford and General Motors wereforced to arrange their quality management systemaccordingly and to meet the demands specified in theSPC and MSA manuals.

Q-DAS® Automates Evaluations

Q-DAS® met the described requirements in qs-STAT®

(for SPC) and in solara (for MSA); moreover, we provid-ed further helpful statistical functionalities. A distinctivecharacteristic was the automated evaluation. Q-DAS®

combined and linked different statistical procedures in away that made it possible to automatically determinethe suitable distribution model and the appropriatequality control chart including the reasonable formulasfor the calculation of quality capability indices only byanalyzing the dataset. This raised the acceptance amongmany users to apply statistical procedures. From nowon, you only had to record the data, transfer them to qs-STAT®, evaluate them automatically and eventuallydisplay the results. As a result, the only task left to theuser is to interpret the results without knowing in detailthe statistical procedures that are hidden behind it.Since Q-DAS® already provided their software in morethan 15 languages at the end of the 20th century, wealready set the stage for the international distribution ofour software.

ISO/TS 16949 Develops

At that time, the suppliers of the automotive industry didnot only have to deal with QS-9000 but also had to sat-isfy the requirements of national associations, such asVDA 6 in Germany, the CNOMO standard in France orthe specifications of the ANFIA in Italy. Suppliers had toaccept the challenge to meet the quality requirements ofthe respective customer. For this reason, a Task Force ofthe automotive industry developed ISO/TS 16949 whichwas referred to as ”Quality Management Systems –Particular Requirements for the Application of ISO 9001for Automotive Products and Relevant Service PartOrganizations”.

Nowadays, this technical specification is obligatory forany supplier in the automotive industry and must beimplemented in each company. ISO/TS 16949 combinesdifferent existing quality management system require-ments (mainly of the North American and Europeanautomotive industry). The SPC and MSA requirementscontained in ISO/TS 16949 are broadly defined. For thispurpose, the VDA 6 catalog was created for Germanautomakers. Possible statistical procedures are assem-

Hartmut M.W. Nowack

bled in ISO/TS 10017. However, it is required that theseinstruments are used. You will have to provide evidencethereof in a system audit. ISO/TS 16949 leaves the ref-erence manuals used in each individual case up to therespective applicant. You might either user the companyguidelines of conglomerates or the SPC Manual of theAIAG. In case of the German automotive manufacturers,the VDA Volume 4 and the VDA Volume 5 or the 4th edi-tion of the MSA are used for capability analyses of mea-surement processes.

The Ups and Downs of Statistical Procedures

Statistical procedures certainly had their high point dur-ing the introduction of SPC at the beginning of the1990s. The number of participants who attended QIItraining (see annual report of the DGQ) is a good exam-ple to prove this fact. However, the more companiesdealt with the implementation of their quality manage-ment system based on ISO 9001 or later ISO/TS 16949or QS-9000, VDA 6 etc. during the time in between, themore the meaning of statistical procedures faded intothe background. People thought that sufficient preven-tive action is enough to control processes in a way thatall quality requirements are met. However, this provedto be a mistake.

Systematic improvement of products and processescame to the fore more and more.

For this reason and particularly because of the introduc-tion of Six Sigma at the end of the 20th century, statisti-cal procedures gained in importance again, i.e. peoplestarted to focus on the real process again. By realizingprojects based on the DMAIC 5-step approach (define /measure / analyze / improve / control), the classicalmethods like SPC and MSA were applied more fre-quently again. The improve step additionally containedthe design of experiments subject including test plan-ning, regression analyses and analyses of variance. Inaddition to the DMAIC approach in order to improve

existing processes by means of projects, the DFSS topic(Design for Six Sigma) for the development of new prod-ucts and processes has been put forward over time.Even this methodology required statistical procedures

National/International Standardization ofStatistical Procedures

At ISO, the Task Committee TC 69 is responsible forstandards on statistical topics. The corresponding mirrorcommittee at DIN (German Institute forStandardization) is the NA 147-00-02 AA committee(known as NQSZ 2). The ISO/TR 13425 technical reportdescribes all statistical procedures with which TC 69deals. Find an overview of those standards on www.q-das.de under “Competence Center Statistics“ – “Normsand Guidelines”. I would also like to mention the ISO/TR13007 technical report. It explains a total of 12 typicalstatistical procedures in short. Moreover, it illustratestypical applications and points out the limits of theseprocedures. In terms of SPC and MSA, standards wereonly created during the last 10 years. With the exceptionof two standards about quality control charts (Shewhartcontrol charts and acceptance control charts), therewere hardly any activities referring to this applicationarea in the field of international standardization.Furthermore, it takes a long time to create an interna-tional standard. A standard that takes only two to threeyears to be released are considered to be a very quickdevelopment.

In the meantime, several series of standards have beendeveloped covering topics such as QM systems, SPC,MSA and Six Sigma:● ISO 7880 et sq. about quality control charts● ISO 14253 et sq. about machine and process capa-

bility ● Part 7 about measurement process capability analy-

ses● ISO 13053 et sq. about Six Sigma.

Statistical Procedures in the FutureDue to the software packages available today, usershave to expect a black box in terms of statistical proce-dures. Users are responsible for recording and exchang-ing data by means of this black box. The black box canbe configured in a way that it adapts to the respectivecorporate requirements taken from guidelines and that itmakes evaluations based on these specifications. Themain benefit for users is that they are able to create thedesired statistics, to interpret the statistics, in particular,and to take corrective action, if required. This applies tothe SPC subject area (primarily quality control chartsand capability indices) and to the MSA MeasurementSystem Analysis subject area when purchasing newmeasuring instruments and in continuous inspectionswhile using the measurement systems. The applicationof a validated software is indispensable since the valida-


Theory and Practice

19PIQ International

Theory and Practice

tion is required in order to accept statistical evaluationsas trustful. Especially globally acting conglomerates willthen be able to use this software worldwide in therespective plants and feel certain that the calculationsare based on the same principles. Particularly by stan-dardizing reports and output masks the communicationbetween departments, divisions and plants will becomeeasier. Notably benchmarks will be simple to create.Everything that is valid and binding within a companyand its plants can also be transferred to suppliers. It is ofutmost importance to set a high value on traceability. Incase of any problems with products or componentsoccurring today, the responsible supplier must be ableto trace back where the problem is caused in his supplychain. The problem will be detected quickly and trust-fully if supplier and manufacturer use the same calcula-tion procedures. The results become transparent andtraceable for all parties involved. Standards considerablysupport you in doing so.

Especially in medical technologies, the requirementsbased on the quality system of the FDA (Food and DrugAdministration) are very high, particularly in terms ofthe convertibility of data and evaluations. We have toexpect that one day these requirements will also applyto the environment of the automotive industry.

Due to statistical procedures, the tasks of the users arerestricted to the following:● They know the procedures, their fields of applica-

tion and their meaning.● Then they are able to interpret and evaluate the

results and they are able to make the required deci-sions.

● The information system prepares the results accord-ing to the respective task and in a user-friendly way.

● The information is communicated in a dedicatedmanner in order that, on the one hand, the respec-tive user of the results cannot put too many infor-mation into the results and, on the other hand, theuser does not miss any information.

I will be pleased about any recommendations, sugges-tions and questions about this article. Please write an e-mail to [email protected].

Literature[1.] A.I.A.G. – Chrysler Corp., Ford Motor Co., General

Motors Corp.Measurement Systems Analysis, ReferenceManual, 4. Auflage. Michigan, USA, 2010.

[2.] A.I.A.G. – Chrysler Corp., Ford Motor Co., GeneralMotors Corp.Quality System Requirements, QS-9000, 3.Auflage. Michigan, USA, 1998.

[3.] A.I.A.G. – Chrysler Corp., Ford Motor Co., General

Motors Corp.Fundamental Statistical Process Control, ReferenceManual, 3. Auflage. Michigan, USA, 2005.

[4.] DGQ - Deutsche Gesellschaft für QualitätLehrgangsunterlagen: QII. Statistische Methodendes Qualitätsmanagements. Frankfurt, 2005.

[5.] DIN - Deutsches Institut für NormungDIN EN ISO 9000:2005:Qualitätsmanagementsysteme - Grundlagen undBegriffe. Beuth Verlag, Berlin, 2005.

[6.] DIN - Deutsches Institut für NormungDIN EN ISO 9001:2008:Qualitätsmanagementsysteme - Anforderungen.Beuth Verlag, Berlin, 2008.

[7.] DIN - Deutsches Institut für NormungISO/TS 16949:2002:Qualitätsmanagementsysteme - BesondereAnforderungen bei Anwendungen von ISO9001:2000 für die Serien-und Ersatzteil-Produktionin der Automobilindustrie.Beuth Verlag, Berlin, 2002.

[8.] Graf, Ulrich / Henning, Hans-Joachim / Stange,Kurt / Wilrich, Peter-Theodor. Formeln und Tabellen der angewandten mathema-tischen Statistik. Springer Verlag Berlin, 1987.

[9.] John, B.Statistische Verfahren für Technische Messreihen.Carl Hanser Verlag München, 1979.

[10.] Masing, W. / Pfeifer, T. / Schmitt, R.Handbuch Qualitätsmanagement, 5. Vollst. Neubearb. Auflage. Carl Hanser Verlag München 2007.

[11.] Rinne, H./Mittag, H.-J.Statistische Methoden der Qualitätssicherung.3. Überarbeitete Auflage.Carl Hanser Verlag München, 1995.

[12.] Shewhart, Walter.Economic Control of Quality of ManufacturedProduct.ASQC/Quality Press; WiederveröffentlichungDezember 1980.

[13.] Shewhart, Walter.Statistical Method from the Viewpoint of QualityControl. Dover Publications Inc., 1986.

[14.] VDA - Verband der AutomobilindustrieVDA Band 5: PrüfprozesseignungVDA, Frankfurt, 2011.


Q-DAS® Products

Stephan Sprink, Q-DAS® GmbH & Co. KG

Q-DAS CAMERA® Concept:Performance Measurement System for Knowledge Acquisition

The software tools of the Q-DAS CAMERA® Conceptharness data streams in order to gain knowledge. Theyprocess characteristics values and process parametersobtained in industrial production based on current stan-dards and guidelines, convert them into reliable statis-tics and convey them. However, the Q-DAS CAMERA®

Concept (see Figure 3) does not only consist of softwaretools but also includes different services supportingcompanies in implementing, realizing and maintainingthe system. Q-DAS® also advises companies on any sta-tistical questions starting with the definition and calcu-lation of statistics, the joint development of an evalua-tion configuration and process optimization. All of thesetasks are part of the introduction of a performance mea-surement system ensuring that the entire potential ofstatistical evaluations can contribute to an increase inefficiency at any time – a valuable acquisition of knowl-edge leading to our customers’ success.

What Are the Requirements of Our Customers?Since the demands on the products and the corre-sponding production rise continuously, the processeshave to become more and more transparent in order tomonitor them in a timely manner and to take respectivecorrective action. The closer you approach the techno-logical limits of the manufacturing process, the moresignificant becomes the precise knowledge of the cur-rent status of the production. The Q-DAS CAMERA®

Concept provides this knowledge in the form of reliablestatistics in order to avoid consequential costs causedby rejects, rework, customer complaints or the like. Theprocessed statistics must be displayed user-friendly andclearly (see Figure 2) in order that the respective

(process) owner receives a clear image of the relevantprocesses.

Integration of Information SourcesIn order to gain a comprehensive overview of theprocesses, the integration of the different informationsources is required (see Figure 3). It is of particularimportance to connect measuring and test instrumentsfrom manufacturing, production and the measurementlaboratory to the system in order to make quality-relat-ed statements. Descriptive parameters of the manufac-turing process from controls (PLCs) provide furtherimportant information needed for the assessment / eval-uation of the processes. They help to determine depen-dencies between product and process characteristics.

Many measuring and test instruments and SPC systemssupport the Q-DAS® ASCII transfer format which is

widely used in the market. The AQDEF(advanced quality data exchange format)data format is based on this transfer for-mat and defines an industry standard ofhow to transfer data to the Q-DAS® system (see Figure 4). Severalconglomerates and suppliers even took astep further by regulating that not onlythe measurement results have to betransferred in the AQDEF format buteven the minimum amount of descriptivedata. A certification offered by Q-DAS®

provides the safety that the data trans-mission works properly. Further informa-tion about AQDEF is available on ourhomepage (http://www.q-das.de/ ;Service – Data format). Figure 2: Customer requirements

Figure 1: Statistics for knowledge acquisition

21PIQ International

Q-DAS® Products

Standard measuring equipment, such as calipers, heightmeasuring instruments or external micrometers, whichare only able to pass the measured values, are directlyconnected to Q-DAS® products via serial interface(RS232 or USB) or multiplexer boxes. The respective Q-DAS® test plan selected before starting the measure-ment contains the required header data (work pieceinformation, characteristics description, tolerance limits,…). The characteristics are allocated to the respectivemeasuring equipment in the test plan. Overall, the Q-DAS® software supports more than 150 measuringinstruments or boxes for an easy connection of currentmeasuring equipment. Instruments that have not beenconnected yet are integrated into the standard range ofsupported Q-DAS® interfaces after examination.However, this integration might be subject to a charge.

In addition to these sources of information, the systemcan integrate data from other systems, such as ERP, MESand CAQ, or files (Excel lists, text files, …), dependingon the respective requirements and demands. The stan-dardized SAP interfaces QM-STI and QM-IDI help toexchange measurement and test data between the SAPQM module and the Q-DAS® software products. In thiscase, SAP is the leading system and, depending on therespective application, the Q-DAS® software productstake over the specifications from SAP QM (QM-IDIinterface) for the recording of test data or the Q-DAS®

products evaluate the measured values available in SAPQM and transfer the evaluation results back to SAP QM(QM-STI interface).

The communication between MES and CAQ is eitherbased on the Q-DAS® ASCII transfer format describedbefore or established via call parameters or servicesallowing for a seamless integration. A converter sup-ports customers in quickly integrating files that are notavailable in the Q-DAS® format yet. Even handwritten

control charts or any other manual recordings can betaken over into the Q-DAS® software products, e.g. byusing scanning software for sheets of paper.

In addition to the recorded measured values, you maysave descriptive additional information for each dataset.Examples are batch information, serial number of theindividually produced and / or tested part, machine andspindle information, etc. When analyzing the data lateron, a specific selection of data will be feasible and thetraceability will be ensured.

Real-time MonitoringAfter determining the format the data of different infor-mation sources are available in or the interface transfer-ring information to the Q-DAS® system, the software isable to visualize and evaluate the process data immedi-ately. We distinguish between two control loops in orderto control processes (see Figure 5). The software imme-diately visualizes and monitors data in the “small” con-trol loop while they are transferred to the Q-DAS® sys-tem. A typical application is the display of results for aquick reaction at the operator level. Signals in case ofprocess interventions or alarm violations accompany thedisplay of results.

Figure 3: Q-DAS® CAMERA ConceptFigure 4: Advanced quality dataexchange format

Figure 5: Control Loop


Q-DAS® Products

Classical applications are CMMs‘ real-time displays ofmeasurement results, the permanent visualization ofprocess parameter values and a local or central displayof test data recorded manually or via serial interface. Incase defined alarm criteria are violated while recordingdata, users have to add a comment (event, measure,cause) in order to allocate or identify the process inter-vention later. For the monitoring of processes based onstandardized criteria in real-time monitoring, companiesshould define corporate specifications (calculation ofcontrol limits, stability criteria, tolerance analyses, …).Only these specifications help to guarantee repro-ducibility. The processes are then monitored based onthe same conditions company-wide.

The Q-DAS® software products procella® and O-QIS ful-fill these very tasks of data recording, real-time visual-ization and alarm monitoring.

Evaluation and Reporting SystemThe option to evaluate and assess data automatically inaccordance with certain specifications (standards, cor-porate and association guidelines) provides the basis forthe application of a performance measurement system,the reproducibility of results and for the supply of infor-mation required at the planning and management levels(“big“ control loop). Despite a huge amount of data,users are able to keep track and clearly detect significantdeviations from process specifications. The most impor-tant feature for a validated evaluation in the Q-DASCAMERA® Concept is the evaluation method (see Figure6). Users may either define statistical calculations andspecifications specific to the respective customer oraccess already integrated standards and corporateguidelines that are mainly applied in the automotiveindustry. By applying the Q-DAS CAMERA® Conceptcompany-wide / across locations with a uniform evalua-tion method included, the results become comparable

and reproducible. It is easy to validate the system at thecustomer since Q-DAS® is responsible for the verifica-tion of numerics.

Besides the calculation of statistics, the display of resultsin graphics also plays a major role. Mere columns of fig-ures, even if they are calculated correctly, incur the riskof overlooking important information about criticalprocesses. By contrast, graphics make it easier andquicker to detect changes over time, especially in casethe evaluation results are displayed in task-relatedgraphics. That is the only way to identify the real situa-tion quickly and safely and to take corrective action orevaluate taken measures.

Standardized and clearly structured layouts for reportsincluding evaluation results ensure that users are able tofind the information about desired statistics (see Figure7) quickly. Depending on the recipient of the report ofevaluation results, it is important to compress the datato a certain degree in order not to get lost in the detailsand to obtain the required measures. Respective data-base queries, displays of evaluation results and optionsto sort statistics guarantee a target-oriented processingof the desired process information including the respec-tive level of detail.

The Q-DAS CAMERA® Concept is particularly comfort-able in case statistics are not calculated by being trig-gered manually but the system performs the calculationautomatically in the background. Optionally, it reportsall selected processes or only those processes not meet-ing the requirement adjusted in the current evaluationmethod. For example, users evaluate the data per day,week or shift and the system automatically provides theresults (report, result file, e-mail) to the responsibleoperator. This procedure saves time and raises the

Figure 6: Evaluation Method

Figure 7: Display of Evaluation Results

23PIQ International

Q-DAS® Products

added value of a successfully implemented performancemeasurement system.

Moreover, there is the option to access statistics over awebsite that can be designed individually. This allowsfor a location-independent observation and evaluationof the company’s processes. Predefined dashboards orwebsites adapted to the individual customer require-ments help to display statistics, graphics and reports inthe web browser. Thus, users do not have to install Q-DAS® software products on local workstations butrequire nothing but the authorization to access definedwebsites of the company.

Data Storage / Data BackupOne of the main tasks of the Q-DAS CAMERA®

Concept’s data flow is to merge locally recorded infor-mation on a central server or to make centrally definedspecifications available on local workstations.Measurement and test data as well as values of processparameters are saved to a central server database or, ifrequired and depending on the organizational structure,to several databases. Since databases are part of thesecurity and maintenance concept of the companies‘ ITdepartments, data security can be established withoutany additional effort. In terms of the obligation to keeprecords for critical characteristics / data inaccordance with customer agreements orlegal regulations, data security is a neces-sity in companies, anyway.

In the context of the Q-DAS CAMERA®

Concept’s data flow, the Q-DAS® soft-ware products write / read directly in /from the database or automated servicesfor transferring locally generated datasave them to the structured central data-base. This fact ensures that no data willbe lost in case client PCs breakdown orare exchanged. Q-DAS® database toolsfor outsourcing historical data or formaintaining the database help users towork with a high-performance databasebut having all information still available.Mechanisms of this type have becomeindispensable over time since there aresometimes more than 1 million datasetssaved to the database per day. All thesedata require a well-structured data man-agement.

Types of InstallationThe most common type of installation forQ-DAS® software products is a serverclient installation. In this case, the centralserver contains all program files, configu-rations and system settings. The local

computer accesses all required data at the start of theprogram, e.g. the license, or the program checkswhether there are still licenses available in case of a con-current licensing. Thus it is even feasible to implementcorporate solutions that are quite frequently based onterminal server installations.

Standardization as a Key to SuccessWhy reinvent the wheel when you can access existingand established standards?

The Q-DAS® software products are standardized prod-ucts that have continuously grown in the course of timeby adapting them to customer requirements. Today,they cover a wide range of applications and still contin-ue to grow. Due to the created flexibility and availableconfiguration options, the software is directly geared tothe needs of customers. Even upcoming updates andupgrades of the respective version are easy to imple-ment.

The standardization involves different functionalities ofthe software. When defining an evaluation method – thebasis for statistical calculations of statistics – the cus-tomer selects one of the different integrated evaluationstandards or creates an individual evaluation method.

Figure 8: VDA 5, MSA 4th Edition and GUM as well as “Bosch Heft 10” and“GMPT Measurement System Specification“ as examples of company guidelines“


Q-DAS® Products

Depending on the desired evaluation module / program,standards and guidelines provide the necessary supportto eventually obtain the suitable evaluation method.

The evaluation of machines, production facilities and therunning production process is based on the statisticalevaluation of characteristics values. Measurementprocesses helping to measure predefined characteristicsprovide these characteristics values. In order to avoidmisinterpretations, the recorded measured values haveto reflect the real situation with an adequate accuracy. Inother words, the measured values have to be suitablefor the respective application. Customers may performthese capability analyses in solara.MP.

In solara.MP customers conduct capability analysesaccording to MSA, measurement process capabilityanalyses in accordance with VDA 5 and measurementuncertainty analyses as per GUM. For concretizationand the practical application of these analyses, the auto-motive industry created guidelines specific to therespective company. These guidelines are also availablein solara.MP (see Figure 8).

qs-STAT® is a software package for comprehensive sta-tistical evaluations of quality information relevant in themanufacturing process. Its main purpose is to assessprocesses and systems based on the evaluation results.In case of machine and process qualification, the evalu-ations are based on integrated standards and companyguidelines (e.g. BMW, GMPT, Robert Bosch,Volkswagen, etc.). There are numerous statistical proce-dures available in order to find the best suitable distrib-ution time model and to allocate process modelsaccording to DIN ISO 21747 (now ISO 22514-2) auto-matically. Using a standardized evaluation provides safe-ty and guarantees the reproducibility of results. In addi-tion to the integrated evaluation methods (see Figure 9),customers may even define an evaluation method fortheir own companies based on already available evalua-tion methods or create methods that are completelynew. We will be pleased to support our customers andadvise them on finding a suitable evaluation method. Ifdesired, we include the new definition in our defaultversion in order to provide it to suppliers / customers,too. Thus you all “talk the same language” when evalu-ating and assessing processes which makes the com-munication between customers and suppliers easier

In order to implement the Q-DAS CAMERA® Conceptsuccessfully and quickly, Q-DAS® created a guidelineand checklists offering support in planning and realizingit. This guideline is also available on our website underhttp://www.q-das.de/en/service/project-guideline/.

This guideline describes typically required specificationsand procedures in order to implement the Q-DAS CAM-ERA® Concept forming the basis for the creation, opera-tion and maintenance of a performance measurementsystem. The guideline ensures that all requirements andgeneral conditions for a successful project are fulfilled.

It is not possible to introduce a performance measure-ment system quickly and efficiently with-out a specification that is as accurate aspossible. Moreover, the specification pro-vides a binding basis for customers andsuppliers contributing to the achievementof objectives. Now adays, the servicesrequired for the system integrationalready play a major role compared to thesoftware the customer needs. In order toreduce the effort to a minimum and tooptimize costs, customers require dis-tinctive and clearly defined targets andprocedures. Our experienced consultantshelp to organize an efficient preparationphase and ensure that customers reachtheir targets. The guideline does not havethe purpose to illustrate any possible pro-ject constellations but wants to providesupport and suggestions for the individ-ual handling of the project. Depending on

Figure 9: Statistical Process Control (SPC) and VDA 4 as well as “Bosch Heft 9”,GM Powertrain „Machine and Assembly Runoff Specification“ and Volkswagen AG“Prozessfähigkeitsuntersuch für messbare Merkmale“ as examples of companyguidelines

25PIQ International

Q-DAS® Products

the project, only single aspects of the guideline might berelevant. Many companies applying Q-DAS® productsglobally defined and documented their corporate speci-fications based on this guideline. Due to this standard-ization, customers are able to realize the introduction ofthe Q-DAS CAMERA® Concept in a short time.

A major part of the specification is the definition of thedata format. It describes the interface between the infor-mation sources to be connected and the applied prod-ucts in the Q-DAS CAMERA® Concept. In addition to thestructure of the format, customers also define the con-tents to be transferred. The contents help to select spe-cific data and to give an individual overview of theprocess later. It is advisable to consider the AQDEF for-mat (see Figure 4) for standardization purposes. Evenmany manufacturers of measuring instruments knowthis format by now which allows for a quick and safeconnection.

An Overview of the Q-DAS CAMERA® Concept’sBenefits

After describing the functions of the Q-DAS CAMERA®

Concept and the procedure for implementing the sys-tem, the question that is still open is the question of thebenefits the customers have when introducing such anew system. Of course, cost savings come first in orderto have or gain a permanent competitive edge.

Advantages of the Q-DAS CAMERA® Concept:

• Standardized interfaces to measuring instruments,controls or other data sources provide safety andenable a quick connection.

• The integration of quality and process data gives acomprehensive overview of the current situation.

• A validated statistical evaluation of the data suppliedby any integrated information source is based onstandards and guidelines and thus ensures repro-ducibility and provides a firm basis of decision-mak-ing.

• Automated evaluation systems offer user-friendlydisplays of results. These displays save time andguarantee transparent processes.

• A Q-DAS CAMERA® Concept consisting of singlemodules saves money, true to the motto: Everybodygets what he needs!

• Standardized implementation approaches reduce theintroduction effort.

• Continuous development of standard software prod-ucts protects your investment.

• An international network of subsidiaries and partnersensures local support.

Figure 10: Q-DAS® publication “Kennzahlensystem“ andProject Guideline (available only in German)

Due to the integration of Thai as another language, the Q-DAS®

software products are now applicable in 21 languages. Since ver-sion 10 (110914) Thai has been available as an additional lan-guage. The program texts of all Q-DAS® software products aretranslated in order to ensure an overall consistency. A Thai trans-lation agency translated the software texts. The agency checkedand released the translated texts in collaboration with an interna-tional Q-DAS® customer located in Thailand. This collaborationguarantees the correct translation of the program texts.

Find all available languages at http://www.q-das.de/de/anwendungen/sprachen/

Please inform us about any further language you require!

Q-DAS® Software Also Available in Thai

27PIQ International

Q-DAS® News

Stephanie Dietrich, Stephan Sprink, Q-DAS® GmbH & Co. KG

Global Q-DAS® Network

More than 8000 companies of various sizes worldwidetrust in the Q-DAS® software. These companies are rep-resented by about 200000 users applying our softwareevery day in order to evaluate product and process qual-ity. Different industries use the software; however, theautomotive industry plays a major role. According to theranking of the TOP 100 automotive suppliers (specialedition of the AUTOMOBIL PRODUKTION magazinepublished in July 2013), 92% of the biggest Europeanautomotive suppliers and 2/3 of the TOP 100 automo-tive suppliers worldwide are Q-DAS® customers.The software is available in 21 languages - a fact thathelps to overcome language barriers. In addition, globalplayers (corporate groups and companies) are able tocompare their results worldwide since they use uniformcalculation methods based on international standards,standards of technical associations and corporate stan-dards included in the software. Meanwhile, theseaspects have led us to a total of 55 countries in whichthe Q-DAS® software is currently installed. However,there is another indispensable factor. It is the interna-tional network of subsidiaries and distribution partnersensuring professional consultation, installation, supportand training in the use of the software on site. Globalsupport is essential to most of our customers.

Network Creation and DevelopmentThe Q-DAS® philosophy has always focused on the rel-evance of the network, i.e. maintaining and expandingit. After Q-DAS® had had to establish itself on theGerman market during the first years, in 1995 Q-DAS®

started to develop the first business relations with dis-tribution partners (independent companies abroad dis-tributing our software on site) in Italy, Brazil, Denmarkand France. And in 1997, Q-DAS® finally founded thefirst subsidiary (Czech Republic). Henceforth, the inter-nationalization has continued. Today, the internationalnetwork including seven subsidiaries represents Q-DAS® in 27 countries.

Major Changes during the Past 12 MonthsDuring the last 12 months, we started to operate in fur-ther countries important to us. A new distribution part-ner has represented Q-DAS® in Russia since November2012. Technopolice based in Moscow distributes Q-DAS® software on the Russian market. The companyprovides customer support regarding software imple-mentation and user training.In the second quarter, Q-DAS® founded a subsidiary inIndia since the Indian growth market required the pres-

ence of Q-DAS® through comprehensive customer sup-port on site. Q-DAS Software Pvt. Ltd. is headquarteredin Pune and is also located in Chennai.Only few months later, we were pleased to welcomeanother new member of the Q-DAS® network. Oberon3D Metrologia has been a Polish distribution partnersince August 2013.And last but not least, there is a brand new subsidiary inSouth Korea. South Korean Q-DAS Ltd. has been part ofthe Q-DAS® Group since September 2013 when Q-DAS® established its 7th subsidiary. The company isbased near the capital Seoul. This business locationrecently also required our permanent presence becauseQ-DAS® customers such as Daewoo, Hyundai, Kia,Doosan and Edwards are located in South Korea.

NetworkingThe members of the network meet in Weinheim andobtain further qualifications at annual partner meetingsand in seminars. Regular meetings of subsidiaries anddistribution partners guarantee that the entire networkalways receives the latest information from the head-quarter and foster the communication between oneanother. As in any other area of life, personal communi-cation is crucial and often leads e.g. to projects our part-ners also adapt efficiently to other countries instead ofdefining an entirely new approach. Even long-timemembers of the worldwide network benefit from anynew suggestions based on established solutions andproject reports.

Meeting of Q-DAS partners, 2013

Please find the contact details of our subsidiaries anddistribution partners on the back of this magazine.


Q-DAS® News

Tom Stewart Receives Ford Global Powertrain Manufacturing Engineering TMM Excellence Award

Tom Stewart, president of our NorthAmerican Q-DAS® subsidiary, Q-DAS®

Incorporated, has received Ford MotorCompany’s Global Powertrain Manu -facturing Engineering TMM (TechnicalMaturity Model) Excellence Award for anOperator Control Boards (OCB) QualityFeedback system installed at Ford’s engineplant in Cologne, Germany. Tom was partof a team that developed the solution usingQ-DBM and M-QIS to consolidate and pre-sent critical quality information to machineoperators in the plant’s cylinder head pro-duction area.

The award is one of only 46 conferredworldwide in 2012. It is intended to “rec-ognize individuals or teams who havemade a significant contribution to engi-neering or organizational excellence withinGlobal Powertrain Manufacturing”. In thenominating documents, Ford’s Cologneplant management said that the Q-DAS®

solution has improved quality perfor-mance, reduced scrap, and increased operator morale atthe plant. These aspects are Ford’s major benefits of theQ-DAS® solution.

“The system rationalizes and expedites the presentationof complex CMM data to the production machine oper-ator,” explained Q-DAS® Incorporated president, TomStewart. “In simple terms, the Q-DAS solution consoli-dates data from eight different 20-page CMM reportsand presents it to the operator on a single, prioritizedOCB screen.

Prior to installation of the Q-DAS® solution, the systemincluded a significant time lag between the machiningoperation and the CMM lab tests, plus additional time togenerate paper reports and deliver them to the operator,and still more time for the operator to read and interpretthem. Using the Q-DAS® solution, the operator gets vir-tually instant feedback from the CMM laboratory and isnever more than 3 mouse clicks away from resolvingany quality issues identified there.

The solution supplied to the Cologne engine plant uti-lized two standard Q-DAS® software products, a data-base manager – Q-DBM, and a management reportingtool – M-QIS. Both are designed to be easily modified toaccommodate specific customer data and reportingrequirements while maintaining complete interoperabil-ity.

Q-DAS® is an international software company specializ-ing in the computerization of statistical procedures forquality management applications with headquarters inWeinheim, Germany. The American subsidiary, foundedin 1988 by president Tom Stewart, is located inRochester Hills, MI. It provides services including soft-ware distribution, training, support hotline and on-siteinstallations for North American customers.

The entire Q-DAS® group is very proud of this award.

Tom Stewart, President of Q-DAS Incorporated

Global Powertrain Manu fac turing Engineering TMM ExcellenceAward der Ford Motor Company

Thomas Stewart, Q-DAS® Incorporated

The Q-DAS® software products procella® and qs-STAT® meet the require-ments of FDA’s 21CFR Part 11 standard for electronic data and systems. Dr.Bethold Lebert, freelance counselor for quality assurance in accordance withFDA and EU standards, conducted a comprehensive compliance analysisdetecting only three minor compliance gaps. Corrective action has subse-quently been defined. When Dr. Lebert checked the software again, he con-firmed the entire and correct realization of the standard which proves thatthe Q-DAS® software is able to fulfill the demands of FDA’s 21CFR Part 11standard. The extended range of functions will be available in the next recom-pilation of software version V10 (110914).In a regulated environment (pharmaceutics,medical engineering and relevant suppliers),software must be validated because of thestorage and usage of the data required for FDAapproval. Almost every relevant type of indus-try uses Q-DAS® software products today. Thenumber of our software applications used incompanies working in the regulated environ-ment rises steadily. Our software was thusextended continuously according to therequirements of these companies.Due to the declaration of conformity of the Q-DAS® software to FDA’s 21CFR Part 11 stan-dard we were able to confirm that the software is appropriate for the appli-cation in regulated environments. This fact provides safety and the basis forthe required validation of the software at the customer.

After the declaration of conformity confirmed that the Q-DAS® products qs-STAT® and procella® are suitable for the application in the regulated envi-ronment, Dr. Lebert also performed an analysis for our products solara® (measurement process capability) and destra® (Q-DAS® statisticspackage).

An analysis elaborated by Dr. Lebert confirmed that the software completelysatisfies the requirements of FDA’s 21CFR Part 11 standard. The requiredextended range of functions has been available since the compilation 110914of version 10. Thus all Q-DAS® standard products (statistical packages) aresuitable for the application in a regulated environment. In the regulated envi-ronment (pharmaceutics, medical engineering and relevant suppliers), thevalidation of the software is obligatory. The validation result is the docu-mented proof that the software works properly and that the results it pro-vides are correct and complete. The validation may be considered as an intro-

duction methodensuring the propermode of operationleading to correctand completeresults in thefuture. These factsare even guaran-teed then due tothe consequence ofthe validation rules.

Q-DAS® Software Complies with FDA’s21CFR Part 11 Standard


Q-DAS® Products

When purchasing products customers and suppliersspecify characteristics the supplier has to meet upondelivery. Generally, these characteristics are functionalcharacteristics with a defined range in which the char-acteristic has to lie. This is why we talk about specifica-tion limits or tolerance limits. Such characteristics canbe limited to both sides or even to one side in specificcases. There are two types of one-sided characteristics.The first type describes characteristics whose physicallimit is zero. Typical examples are shape dimensionsand positional dimensions. However, in case of charac-teristics like torque or coating thickness, a minimumlimit differing from zero is generally demanded andmust be met.

In principle, it is always a question of whether the spec-ified limit makes sense or not. In most cases the speci-fications are defined based on the respective design ofthe product. After completion, this design is responsiblefor the proper functioning of the product according tospecifications. As a consequence, you always tend todemand narrow limits for liability reasons. This require-ment serves your own protection – a quite comprehen-sible fact. These tolerances are referred to as “anxietytolerances” in practice.

On the contrary, you know that the production alwaysincludes a certain variation. No part is produced exactlyin the same way as another. As long as the characteris-tics of two-sided tolerances lie in the center of the tol-erance range, there normally will not be any problems.However, in case the characteristics are near the upperor lower specification limit the situation becomes criti-cal. Now the uncertainty of the measurement processcomes into play. Independent of the type of characteris-tic it is important to find out whether a delivered prod-uct stays within the demanded specification limits bymeans of suitable measurement processes. It is alsoclear that each measurement process incurs an uncer-tainty. There is no measurement process having anuncertainty of “zero”, not even the PTB (German nation-al metrology institute in Brunswick) is able to providesuch a process. If you are able to evaluate the charac-teristic quantitatively – thus you are able to measure it –there are two options available to determine the specifi-cation limits for the respective design.

You calculate the expanded measurement uncertainty ofthe applied measurement process exactly and considerit at the specification limits as demanded by ISO 14253.This approach certainly provides the neatest solution. In

case suppliers and cus-tomers know the mea-surement uncertainty oftheir measurementprocesses, there shouldnot be any debates aboutthe evaluation of productcharacteristics.

In case you do not thinkthe supplier or manufac-turer capable of knowingthe expanded measure-ment uncertainty of themeasurement process,you narrow the toleranceyourself (by an assumedmeasurement uncertain-ty).


Specification Limits and Costs

Figure 1: Far too high measurement uncertainty causes loss

An unnecessarily close distance between two specification limits can cause a considerable amount of addition-al costs. The measurement process should hardly contribute to the increase of costs which might lead to the factthat a higher quality and more expensive measuring instrument is required. However, the investment can beworth it as the following article shows.

Most people often do not think about the costs caused by narrowing the tolerance. Figure 1 shows a case study illus-trating the additional costs caused by an uncertainty that is far too high. Thus it is reasonable that the buying and salesdepartments have sufficient knowledge about the applied measurement processes. In this case they can consider thisknowledge in contractual agreements which might lead to the expanded specification limits.

31PIQ International

Q-DAS® Products

Do Limits Make Sense?

The first question that comes to mind is whether limitsmake sense. The clear answer is “yes” in case most ofthe measurement processes can be evaluated based onthese limits since these limits provide clear conditions.Many practitioners confirm that the widely used proce-dures of the measurement process capability analysiscan be applied to more than 50% of applications. This isa considerable amount. However, it is also clear thatmost measurement and test processes relevant in prac-tice cannot be evaluated one-to-one with this procedureand these limits. It depends on the characteristic to betested but also on the complexity of the measurementprocess under real conditions. In many cases you can-not even distinguish between a manufacturing and mea-surement process. For this reasons, it is important toevaluate the respective conditions in such cases and toestablish them according to the current state of the art.You individually decide on capability by means of tech-nological and economic aspects and by considering therespective risk.

Important Limits and their MeaningBy ignoring the procedure and calculation according toGUM (Guide to the Expression of Uncertainty inMeasurement) in terms of measurement process capa-bility, you may divide the evaluation procedures in threecategories:

• Company guidelines for the evaluation of mea-surement processes

• MSA (measurement systems analysis) of the AIAG(automotive industry action group)

• VDA 5 Capability of Measurement Processes orISO 22514-7 Measurement Process Capability

It is always the own guideline that is relevant to eachcompany at first. If such a guideline does not exist, supe-rior standards of technical associations or general stan-dards will be used in audits. You may divide companyguidelines in two categories. Some guidelines are basedon the MSA for historical reasons because it has alreadyexisted since 1990 and most Anglo-American groups,such as GM, Ford or Chrysler, prefer the MSA. Germancompanies, such as VW and their subsidiaries, Daimleror BMW, rather use the procedure described in the VDA

5 manual. The second edition of VDA Volume 5 pub-lished in 2010 is based on ISO 22514-7 (2012). For thisreason, these procedures are likely to become moreimportant on the international level in the next fewyears. Only time will tell if VDA 5 prevails against MSA.

However, this fact causes problems for the suppliers ofthe automotive industry. They might be obliged to eval-uate their measurement processes according to MSAand VDA 5. Main suppliers thus have to apply the limitsspecified in these two guidelines.

Relevant Statistics and their LimitsCompany GuidelinesCompany guidelines first evaluate the resolution bymeans of the %RE value. It must be less than 5% of thetolerance. In addition, the company guidelines normallycontain Type 1, Type 2 or Type 3 study. Type 1 studydetermines the Cg or Cgk value from repeated measure-ments on a reference part. This value has to exceed1,33. Type 2 study evaluates the measurement processunder real conditions. Thus the respective operator car-ries out the inspection at the operating location of themeasurement process and takes measurements on thereal test parts. The recorded data help to calculate thestatistic %GRR (gage repeatability and reproducibility).The same requirements apply to this statistic as men-tioned in the widely applied MSA:• %GRR ≤ 10% capable• 10 < %GRR < 30% conditionally capable• 30 ≤ %GRR not capable

Please note that the limit for a “capable“ processamounts to 20% instead of 10% in all OEM companyguidelines, such as GM, Ford or Chrysler. Suppliers cer-tified based on ISO/TS 16949 cannot avoid staying with-in the limits specified above, unless they are able toreach individual agreements with their customers.

These limits are quite an unfortunate choice since youdistinguish between “capable“ and “conditionally capa-ble”. Most measurement processes are rather condi-tionally capable than capable in practice. In this case youhave to find technological and economic aspects oradditional measures to give reasons for exceeding the10% limit.


Limits in Measurement Process Capability Analyses

Depending on the standard, standard of technical associations or company guidelines, there are different statis-tics to evaluate test or measurement processes in measurement process capability analyses. These statistics arecompared to specified limits in order to evaluate capability. The limits discussed in the following refer to quan-titative characteristics. It is also assumed that the procedures and the calculation of statistics are known.


Q-DAS® Products

Statistics in the MSA Manual

The MSA effectively includes three relevant limits. Ituses the ndc factor to evaluate whether the data cate-gories are sufficiently small. This factor must exceed 5.Please find notes about how to determine the ndc valuein the “Notes on MSA 4th Edition” PIQ article (pub-lished in PIQ 03/2010 on www.q-das.de). This statisticis comparable to the %RE value given in company guide-lines.

The MSA uses the t-test to evaluate the systematic mea-surement error. This test checks the bias. The calculatedstatistic has to lie inside the confidence limits, otherwisethe bias is significant and the measuring system is notcapable. In terms of equipment variation, the MSA 4th

edition just includes the general requirement that it hasto be small. It does not provide any specific limit. Youuse the limits listed above in the calculation. Former edi-tions of the MSA manual focused on the Average RangeMethod ARM in order to calculate %GRR. Since the 4th

edition published in 2010, the MSA manual prefers themethod of ANOVA for the calculation of %GRR.

Statistics of VDA 5 or ISO 22514-7The ISO standard and VDA Volume 5 distinguishbetween measuring system and further componentsleading to the measurement process (see Figure 1). Thisdistinction is based on the definition of VIM (Vocabularyof International Metrology) and the reason why thereare two limits. One limit refers to the measuring systemand the other to the measurement process. Both docu-ments calculate two capability ratios respectively, QMS

for the measuring system and QMP for the measurementprocess.Recommendation:

QMS ≤ QMS_max = 15%QMP ≤ QMP_max = 30%

Compared to company guidelines, the standard and thereference manual also assess the resolution in order toevaluate the measuring system. %RE must be less than5%.

Tolerance as a Reference Size

The reference size is of utmost importance for the cal-culation of the described statistics because it affects theresult considerably.

Since the MSA uses different reference sizes in order todetermine the %GRR value, VDA Volume 5, ISO 22514-7 and all company guidelines always apply the toleranceas a reference size. This is reasonable since it is a deci-sive figure valid in specifications and agreementsbetween customer and supplier.

This is the reason why it seems obvious to use the tol-erance as a reference size even in measurement processcapability analyses.

How Do We Achieve Limits?Nowadays, we may only speculate about how statisti-cians achieved the one or other limit. Only the personsinvolved in defining these specifications more than 20years ago know how they determined these limits.

Figure 1: Important influences on the uncertainty of measurement results

33PIQ International

Q-DAS® Products

Comments on single limits:

ResolutionThe evaluation of the resolution based on %RE hasproved to be particularly reasonable. This evaluation isquite easy on the one hand and very helpful on the oth-er hand. In case you do not meet this requirement, youmight effectively “classify” the measuring instrument inrepeated measurements taken on a reference part, i.e.the measuring instrument always shows the same mea-sured value. Actually the variation would be zero andthe measuring instrument would be more than suitablein this case; however, this conclusion might be wrongsince the resolution subject to the tolerance is too low.

Cg, Cgk ValueThe limit for the Cg or Cgk value as used in companyguidelines describing Type 1 study for the calculation ofthe equipment variation or the systematic error can beregarded based on the SPC procedure (statisticalprocess control) introduced in the middle of the 1990s.Back then, statisticians introduced the capability indicesCm, Cmk, Cp, Cpk and later Pp, Ppk. These statistics are cal-culated by comparing the variation or offset of theprocess to the specification limits.

The Cg and Cgk values are comparable to these capabili-ty indices but the permissible range is logically restrict-ed. Otherwise the entire variation within the tolerancewould be considered to be the variation of the measur-ing instrument. In German-speaking countries andEurope, these statistics were first mentioned in theBosch reference guide 10 and in the Ford guideline EU1880. The limit for Cg or Cgk amounts to 1,33 in theBosch reference guide and the tolerance as a referencesize is restricted to 20%, whereas Ford demands a Cg or

Cgk value of 1,0 and specifies a tolerance limit of 15%. Itseems like Bosch wanted to compare the Cg or Cgk val-ue to the requirement of 1,33 demanded in machinecapability analysis. On the contrary, Ford obviouslywanted to compare the Cg or Cgk value with the formerCp or Cpk value. At that time, these two values amountedto 1,0 instead of 1,33 as is customary today. However,please note that these explanations only reflect the inter-pretation or assumption of the author.

Note on Cg, Cgk

Since the Type 1 study for the calculation of Cg and Cgk

as mentioned before is quite easy to handle, thismethod was established in many international companyguidelines and not only in the MSA manual. Some exam-ples are GM, Fiat, Ford, etc.

%GRR ValueThe first edition of the MSA manual already included thefixed limits for the %GRR value. Whether these limits arereasonable - in particular the distinction between “capa-ble” and “conditionally capable” - is a favorite subject ofdebate among statisticians. However, the applied proce-dures in order to determine the %GRR have been usedfor more than two decades now. Thus these limits seemto be written in the sky, i.e. they are unlikely to changein the foreseeable future.

QMS and QMP

The reason why VDA Volume 5 and ISO 22514-7 definea QMS_max value of 15% and a QMP_max valueamounting to 30% is easy to explain. They wanted totake over the upper limit of 30% for the measurementprocess as specified in many company guidelines and inthe MSA manual. The first edition of VDA Volume 5 pub-

Figure 2: Comparison between %GRR and UMP

In case you do not exceed the limits in the evaluation based on the %GRR (MSA), you assume that the measurementprocess is okay. By estimating the expanded measurement uncertainty as demanded by ISO 14253, you may considerthis uncertainty at the specification limits.


Q-DAS® Products

Quality EvaluationDepending on the manufacturing or production process,selected quality characteristics are inspected in or afterthe different process steps. You may conduct a 100%inspection or an inspection based on a sample. You eval-uate the manufacturing or production quality graphical-ly by using various visualizations or numerically by cal-culating capability indices. The recorded measured val-ues are evaluated statistically and the required statisticsare calculated. These data are processed numericallyand, depending on the respective application and theresponsible user group, graphically, too. Only by suc-ceeding in communicating the results quickly specifical-ly to the respective task and user and in making themeasily accessible, these results are applied in order toevaluate and assess processes and certain issues. In thiscase they contribute to the quality evaluation.

UncertaintyThe results or issues include, amongst others, uncer-tainties as a result of:• measurement and test processes• the application of statistical procedures • erroneous data recording, transfer and manage-

ment• erroneous communication of resultsYou may solve the problems caused by the last twosources of error with organizational measures and ITsupport, e.g. by permanently checking the plausibility ofdata where relevant. The application of Q-DAS® prod-ucts helps you to describe processes by means of vali-dated statistical procedures specifying the confidenceintervals for the single statistics. The uncertainty causedby statistical procedures becomes assessable now.However, the uncertainties from the measurementprocesses remain and thus we will have a closer look atthem in the following.


Reducing the Uncertainty through Suitable Measurement Processes

lished in 2003 already contained the empirical assump-tion that the combined standard uncertainty of the mea-suring system amounted to about 50% of the combinedstandard uncertainty of the entire measurementprocess. It is more than obvious to define the limit of theQMS value as half of the QMP_max value so that itamounts to 15% in the evaluation of the measuring sys-tem.

International Meaning of these Limits

Particularly in a global, economic sense, manufacturersof measuring instruments or measuring systems have ahigh interest in the definition of standardized and bind-ing procedures and limits. This is also helpful for theexchange of goods between customer and supplier.There is no other way manufacturers of measuring sys-tems can be sure to meet the agreed specifications inselling and later acceptance of their products. The sameapplies to suppliers since they sign delivery contractsand agree to meet product characteristics. You may onlycheck and evaluate this demand in a reasonable way by

using a standardized measurement process capabilityanalysis and by being able to consider the expandedmeasurement uncertainty as correct and binding at thecustomer’s and the supplier’s.

Summary

Measurement process capability analyses for the calcu-lation of capability indices and ratios are important. Youdecide whether a measurement process is “capable“ or“not capable” by comparing capability indices and ratiosto specified limits. The better and the more frequentlyyou are able to apply these procedures, the easier it is toperform a capability analysis.

However, you should be aware that you cannot measureeverything by the same yardstick. You have to decide ineach individual case whether the standards discussed inthis article are applicable.

Q-DAS® offers a platform for the evaluation of thesespecial cases.

In industrial production, the applied measurement processes evaluate and assess the quality of manufacturingand production facilities as well as the produced parts, components and products. The results gained by the mea-surement processes and the statistical evaluation always include different uncertainties.

35PIQ International

Q-DAS® Products

Uncertainty Caused by the MeasurementProcess

The assertion “The accuracy of your measurementsdetermines the accuracy of your production!” is morethan true as the specifications become smaller. This isthe reason why a capability analysis must be conductedbefore applying a measurement process .

In case a measurement process determines a measuredvalue, this value will be invalid if you do not know theuncertainty of the measurement process. Only theexpanded measurement uncertainty UMP added to themeasured value xi leads to the measurement result yi =xi + UMP.

Figure 1 shows the measurement result relating to thespecification for a quality characteristic. It is easy tounderstand that the measurement uncertainty UMP mustbe low compared to the tolerance TOL. Otherwise,wrong decisions in the evaluation of the measured val-ue, particularly near the specification limit, will beinevitable. Such a decision mightbe responsible for the delivery ofdefective products to the customer.You avoid this problem by perform-ing capability analyses for yourmeasurement processes.

By comparing the capability ratio QMP to a specified lim-it you evaluate the capability of the measurementprocess. VDA 5 or the ISO 22514-7 standard propose alimit of 30% for QMP 30%. This is only a recommenda-tion that is not obligatory or binding. In some cases, thislimit has to be adapted to the respective measuring task.

Figure 2 displays how the capability index (also referredto as C-value) for the evaluation of machines, manufac-turing equipment and processes relates to the capabilityratio QMP of the measurement process. As the uncer-tainty of the measurement process rises (QMP rises) thisgraphic clearly shows that the difference between theobserved and actual capability index becomes greater. Incase of a QMP value of 40%, you will observe a Cg valueof 1,33 even though it actually amounts to 2,2 due tothe uncertainty. In order to calculate the capability ratioQMP, you have to determine the expanded measurementuncertainty UMP of the measurement process.

You may either use• GUM (Guide to the Expression of Uncertainty in

Measurement) for calibration laboratories or mea-suring rooms

• or ISO 22514-7 or VDA Volume 5 for measure-ment processes in manufacturing or production

in order to calculate the expanded measurement uncer-tainty.

QU

TOLMP

MP=⋅

⋅2

100%

Figure 1: Proof of conformance with the tolerance

1 Definition of measurement process as giv-en in VIM: Interaction of interrelated operatingresources, actions and influences creating ameasurement.Note: Operating resources can be men andmaterials .

Figure 2: Display of the actual C-value compared to the observed C-value sub-ject to QMP


Q-DAS® Products

Presentation

May I present myself? I work in the field of quality assur-ance. My company produces shafts.

Different machines produce these shafts. My shaft hasgot five points (characteristics) I have to inspect in orderto find out whether everything is alright.

There are different suppliers, different machine opera-tors and different appraisers for the material, but moreon that later.

I use a caliper in order to inspect my shaft. At the pushof a button the calipershows me the measuredvalue of the respectivecharacteristic and I write iton a piece of paper.

I know, this does notsound like the best practicefor quality assurance.

Is my Measuring Equipment Capable?Capability of measuring equipment -> solara.MPThis is the first time I encounter the world of Q-DAS®: IfI want to measure a certain quantity, I have to make surethat my measuring equip-ment works preciselyenough to do so.This is when it comes tothe capability of measur-ing equipment; and it isthis capability that I needto know.So I want to use the solara® software helping me tomake different analyses easily. Eventually I will receive areport providing an overview of all the important piecesof information I need to know.This report shows a green smiley at best. Always.Now I am ableto evaluate mym e a s u r i n gequipment. Iinspect it atregular inter-vals. And I canrely on the cal-culations madeby the soft-ware.

The results are correct and validated. So I have to take a closer look at my process now. And Ihave to deal with these Cp and Cpk values that everyoneis talking about…“I have to establish the capability of my measuring equip-ment continuously to perform the desired inspectiontask. solara shows me the capability in a clear and simpleway. “

Cm? But I Wanted to Calculate Cp!Machine capability -> qs-STAT®

Everybody talks about process capability…However, I have just checked whether my measuringequipment is able to measure correctly, so I should alsocheck whether mymachines are able to pro-duce my parts. The magic word ismachine capability analy-s i s . Q-DAS® refers to thistype of analysis as a sam-ple analysis. The capabili-ty index - a kindof “grade” for themachine- iscalled Cm/Cmk.I check whethermy machine iscapable in onego by using asingle sampleand as few influ-ence compo-nents as possible.I measure 50 parts that have been produced successive-ly. Then I enter the measured values in qs-STAT®. Now I tap<F10> and, lo and behold, the machine is capable. qs-STAT® selects the correct and best suitable calcula-tion method for my measured values automatically. I do not have to be a statistician in order to use this soft-ware but, if I want to, I may adjust everything in theminutest detail.“I use qs-STAT® in order to perform sample analyses andprocess analyses. I check whether my machine is capableof producing the desired parts. “

Thomas Schäfer, Q-DAS® GmbH & Co. KG

A Simple Introduction to the World of Q-DAS® and theCAMERA ConceptA not totally made-up report on my experience

Figure 1: My shaft

Figure 4: Machine capability50 values

Figure 2: My caliper

Figure 3: Evaluation results for measuringequipment

Figure 5: Machine capability analysis

37PIQ International

Q-DAS® Products

How Well Does My Process Work?Process capability -> qs-STAT®

Capable. But what does that mean anyway? We produce 100 shafts a day. I cannot inspect all ofthem in a measuring room. I only inspect five ofthem but I need softwarethat tells me the qualityof the remaining 95 partsbased on the measuredvalues of these fiveshafts. That is statistics.However, before thesefew shafts are able toprovide information about my current process, I have tobecome thoroughly acquainted with the process first.Three shifts measure five shafts for ten days. Based onthis “amount“ of values I am able to check whether theprocess is capable to produce the shafts.qs-STAT® does that for me. The software calculates e.g.the capability index “Cpk“ from my measured values.This index is like a school grade, the higher the better.And if this index even exceeds my demands/my targetvalue typically amounting to 1.33, my process will be“capable”.Short example: If I analyze my samples and this analysisleads to a Cpk of 1.67, statistics says that I have to expect0,63 erroneous parts out of one million parts produced. Result: Not even one part will be defective. I can livewith that. Once again: Even though I just measure few parts, qs-STAT® draws conclusions about all parts. The soft-ware reaches these conclusions automatically and, justas the Sample Analysis module does, selects the correctcalculation methods, but more on that later.By the way, the method used to calculate the capabilityindex and the reason why it refers to the number oferrors per million (parts per million, ppm) is a differentand complicated subject. The TEQ® offers trainings inthis topic area. They are the experts in training and con-sulting of the Q-DAS® Group. You may either participatein a certain training course or invite an expert to yourcompany. This seminar will then be adapted to yourneeds and requirements.“qs-STAT® helps me to understand my processes. I amable to control, monitor and analyze.“

I Need Measured Data, as Easily as Possible. Recording measuredvalues -> procella®/ O-QIS procella®

My system grows. Mymeasuring equipment iscapable. My machinesare capable. We produce.

I would like to check myproducts continuouslynow. Thus my appraisers shallinspect the shafts takenfrom the productionprocess.In this case, qs-STAT® isnot the ideal softwaresolution. I want my appraisers, who did not receive anytraining in using the software, to be able to enter mea-sured data easily. The data should refer to the respectivetest part and be entered in the correct sequence. It willbe perfect if my appraisers receive exact images, instruc-tions and information about where to measure the part. And this is procella® . First I create a test plan. Then I determine the measure-ment sequence. Now I define the characteristics and Iam geared to my drawing from the development depart-ment. These tasks are quickly to fulfill and procella® does allthe rest.In version 10, you may even stamp CAD drawings auto-matically, i.e. transfer measuring points / characteristicsfrom the development drawing automatically.The old piece of paper that I wrote the measured valueson has long vanished in the bin. Electronic data record-ing has got great advantages.Here I have got the option to save images and drawingsfor each characteristic. Even untrained operators areable to take the measurement correctly.In the course of time I realize that I would like to appendadditional information to the measured value. I havealready recorded date/time but I would like to add batchand machine. That is not a problem.The past is already forgotten: At the beginning weentered measured values manually. Now, one of thestrengths of procella® is the connection to measuringequipment. The latest list of interfaces includes morethan 200 current measuring instruments that can beconnected to procella® directly and the number still ris-es.In the meantime, we haveconnected several feelersand calipers, of course, toprocella®. Push the button and themeasured values arerecorded. procella® jumps to thenext characteristic to berecorded automatically.By the way, the feeler on the picture was not containedin the list of interfaces.

Figure 6: Process analysis

Figure 7: Drawing from devel-opment

Figure 8: procella input mask

Figure 9: Connected measure-ment system


Q-DAS® Products

In these cases, Q-DAS® creates a new connection. “procella® can visualize any measurement procedure,however complex it might seem. As an additional sup-port, I may integrate photos and drawings.“

I Want to Find Information More Easily!Files vs. database -> Q-DBM

We work with files all the time. This is reasonable whenperforming measuring system analyses (solara®).However, you will losetrack of your process datasomeday.Q-DAS® always installsan Access database pro-viding storage space forseveral millions of mea-sured values and it is stillfree of charge. The great advantage:Each program, no matter if it is solara®, qs-STAT® or pro-cella®, can load data directly from the database.I may define the display of test plans individually.So I will find my parts more quickly.If I am only interested in the production data of the lastmonth and only the values of machine 1 and batch 4 areimportant, I can filter the information.I have got the option to save these filters and even selec-tions which makes working faster.Someday, the Access database might not provideenough storage space anymore. Now, MS-SQL orOracle comes into play. And Q-DAS® offers Q-DBM, thedatabase module in order to connect a “real” database.And by the way: I was able to directly import my old Excel file that I stillhad from pre-Q-DAS® times by using an import tool. Icould also have done it by copy & paste but the importseemed more elegant to me.“Working with databases makes it easier to find data andprovides the option to work with advanced filters.“

My Measuring Machine Communicates withQ-DAS®: Let’s Pass the Data! Connecting measuring machines -> O-QIS CMM

The system continues togrow. Several measuringmachines are in the mea-suring room. Theirproviders are certified by Q-DAS®, i.e. themachines are able towrite the Q-DAS® transferformat.Files in the Q-DAS® trans-fer format are small textfiles containing the mea-

sured value, test plan information and any further infor-mation provided by the measuring machine.At the beginning, we sent these files to the database byusing the Upload Client, a little Q-DAS® tool that is freeof charge. This service just searches for the best suitablepigeonhole in the database and saves the measured val-ues there.However, in the course of time, you would have beenhappy if erroneous measurements had not been savedto the database at all or if additional information hadbeen appended to the measured values as procella®

does.This is when O-QIS CMM can help. (Even though this piece of information might be confus-ing: O-QIS is the main product and includes severalmodules, namely procella®, CMM, Alert Manager andMonitoring. However, procella® can also be purchasedas an own prod-uct.)We installed O-QIS CMMbetween theinterfaces of them e a s u r i n gmachine and theUpload.As soon as the coordinate measuring machine (CMM)takes a measurement, the measured values are immedi-ately displayed in O-QIS CMM. There you find two but-tons, a red one and a green one. Use these buttons toevaluate the measurement easily and quickly and decidewhether you want to upload the data into the database. Now we have O-QIS installed at every measuringmachine. A new task for the appraiser is now to confirm the mea-surement. This task leads to useful measured values in the data-base and it does not matter whether they are inside oroutside the tolerance.It is very easy to use the software. You may even activatethe input of additional data. Everything is possible.“Due to O-QIS CMM, measuring machines are integrat-ed perfectly into the system. In addition, the measureddata are evaluated.“

Who Is Allowed to Delete My MeasuredValues?Rights management in the world of Q-DAS® -> allproducts

Initially, we had local installations. This was sufficient.However, by using a central database we were alsosearching for an option to manage user rights centrally. In the user management you may define groups andassign specific rights to these groups. As an example, wehave the “appraiser” group whose members are actual-ly only allowed to measure.

Figure 10: Database and filter

Figure 11: O-QIS CMM

Figure 12: CMM data flow

39PIQ International

Q-DAS® Products

The members of the “process responsible” group havemuch more rights, e.g. they are allowed to create testplans.I do not have to create every single new user anymoresince I can use the “Use Windows registration” option.Due to this option new users are created automaticallywhen they sign in for the first time. Then I move the newusers to the respective group. Ready.

More and More People Want to Be Part of ItSingle & concurrent licenses -> all products

In order to save costs, we removed our single installa-tions (stand-alone) and decided to purchase the concur-rent version (floating) instead. So we use a server instal-lation now. This server installation knows that we own a certainnumber of “floating” licenses per product. Thus we mayinstall the client installation on any computer but thenumber of users working with the software at the sametime cannot exceed the number of licenses we pur-chased.In the meantime, we have got a qs-STAT® plant license.This means that we may use qs-STAT® as often as we likeat our site. If we require an additional license for a product, weorder it and receive a small license file that we can eas-ily load on a central server.

Everything under Control!Control loops in the world of Q-DAS® -> procella® &qs-STAT®

In addition to the recording of measured values and theoverall evaluation of our processes, which is alsoreferred to as a major control loop, the graphical userinterfaces of our recording programs were adapted in away that control charts and additional information aredisplayed to the operators while taking measurements.Thus we implement the minor control loop that helpsour operators to control the machines directly sincethey only have a short-time slot available. Action limits from differ-ent calculated controlcharts show operatorswhen a process changesand when they activelyneed to take correctiveaction in the process.You may calculate thecontrol charts automatically or specify them manually.This is also done on a central server since you would notwant to walk from workstation to workstation andadjust the respective settings there.In case the specification limits are violated, the machineoperator has to select an event, e.g. ”tool breakage“. Wenow gain more and more measured values including

additional information helping us to make evaluations.“O-QIS procella and CMM do not only suit the purposeof data recording, but also help us to evaluate and henceto control the machines directly.“

I Need Own Reports. And I Want to IncludeMy Corporate Identity.Report design adapted to your needs ->FormDesigner (Q-FD)The reports provided by the software have been ade-quate for a long time.Someday, customers and other departments had addi-tional requirements. The Form Designer givesus the option to createown reports or to changeexisting ones. You mayeven embed graphics anddrawings, e.g. anoverview of all measure-ment points including adrawing of the part show-ing the location of all themeasurement points.You configure individuallyhow the information con-tained in the report is dis-played. You may evencombine summaries of parts, data related to character-istics and single lists of values as requested.

More Communication between MeasuringRoom and Machines Required.Two-stage CMM ->O-QIS CMMAnother improvement. We introduce a new intermedi-ate step for the measuring machines. Do you stillremember? The measuring machines used to send theirdata to the O-QIS software in the measuring room sofar. There the data were receipted and transferred to thedatabase automatically by using the Upload. However, our problem was that the information fromthe measuring room hardly ever reached the machines.The operators delivered the part to be measured to themeasuring room. And what happened then?Now we have changed this procedure. We installed O-QIS workstations at the machines so that the whole pro-cedure has been modified: Depending on the machineproducing the part to be measured, the measuringmachines in the measur-ing room do not sendtheir measured values tothe Upload but sendthem to the second chainof O-QIS installationsfirst, directly to the corre-sponding machine.

Figure 13: Control chart

Figure 15: 2-stage CMM dataflow

Figure 14: Customized report


Q-DAS® Products

Hence, the operator, who has just delivered his part ofmachine 1 to the measuring room, promptly receivesthe measured value and the value chart of his part sincethey are displayed at his machine. Now he is able tomake adjustments, if required.The kind of evaluation the operator makes changes.Now, the inspector in the measuring room only evalu-ates the quality of the inspection. Measured correctly,correct test plan.And the operator at the machine may add additionalinformation, e.g. the reason why and how this measuredvalue was recorded and which corrective action hewants to take now. Tool breakage, machine readjusted,etc. The information is flowing…“QIS-CMM is applied flexibly. A better communicationbetween measuring room and machine raises the signif-icance of your measured values and leads to shorterresponse times.“

How Warm Was It? Process data from machines ->all productsThe free-of-charge Upload Client allows you to load dataavailable in the Q-DAS® transfer format automaticallyinto the database. We have managed that our in-housetester system is able to create these data. The data for-mat is described quite well and always disposable. Thuswe are able to combine machine data with measureddata. As an example, we record temperatures and othermachine parameters at the time of producing a part andsave this information to the database. We determinecorrelations when comparing these data to our mea-sured values. But don’t worry, qs-STAT® will do thatautomatically for you.However, it is important to have a clear allocation ofdata. In our case, this is a unique ID number identifyingthe machine and the measuring station. Now we gainsignificant information about each part.Q-DAS® or any Q-DAS® partner may also connect themachines to O-QIS. They write converters to be imple-mented into the Upload Client. After the implementa-tion is completed, your system can read any file format.And by the way, the program already includes an Exceland CSV converter.

Punctually at 8 A.M. the Report Is on theTable! Automated reporting system ->M-QIS ReportingMaybe “intermediate result” is the better heading. Ourdata stream flows all the time. Our information is valu-able. We regulate and control directly at the machinesand in a more comprehensive way in order to keep andimprove our quality standard.I got some new projects and I have only a short time leftto create reports on a daily basis, even though different

departments require them. Due to the growth of oursystem and the significance of the provided information,meanwhile, there are quite a lot of reports our plantrequires.M-QIS helps. This is another product that should be installed as aserver. It offers severalfunctions and the firstone I will use is theReporting System.I define reports that Ihave created with theForm Designer. Then Iallocate them to selec-tions in my database, i.e.a certain amount of data,no matter if they are related to parts, characteristics orvalues. I am able to provide “groups of recipients” witha certain report and data. These recipients are e-mailgroups, printers and PDF storage folders.The combination of report, data and “recipient” can becreated as a job in M-QIS so that this job will be doneautomatically at regular adjustable intervals.Would you like some examples?Every morning my shift supervisors receive an e-mailincluding a PDF file about the evaluation of the previousday. In addition, the head of our division gets a monthlyevaluation of the production. The QM department alsoreceives reports, but only about bad characteristics - aseparate report for each customer we support.Automatically, etc.“M-QIS is a powerful tool. Its included reporting functionautomates your reporting system and indicates corporateinformation paths.“

Our Group Wants to See Some Information.I Control It! Web-based intranet solutions ->M-QIS WebM-QIS was actually installed because of the automatedreporting system only. But I am curious and there is awebserver activated on the server M-QIS is installed on.Born as a small test system, it developed into a sourceof information for colleagues working in other plants.The principle is quite simple. Q-DAS® provides a kind ofqs-STAT® web version. It can be completely adapted toyour needs. Anyone whois acquainted with webapplications, is able tointegrate it into the ownintranet structure quickly.Now a Q-DAS® website ispart of our intranet.We actually use almostevery qs-STAT® function.Read from database.

Figure 16: M-QIS server

Figure 17: M-QIS web

41PIQ International

Q-DAS® Products

Display of measured data.Open reports etc.However, everything isglobally accessible. Ofcourse, only within ourgroup. Now the otherplants have got the optionto see current data, how-ever, only those data we release. By using VPN tunnel-ing I can also access these data at home.And how can we top this?Since lately, even suppliers are connected to our net-work. They have their own website in our Q-DAS®

intranet and just need a mouse click in order to loadtheir measured data into their own supplier databasefrom the distance.That is how we check the quality of the supplied parts,even though these parts have not even arrived at theplant yet.In the course of time, we reduce the burden of the peo-ple working in the goods receipt department. To someextent, the supplier now conducts the incoming goodsinspection.“The M-QIS web application may be integrated into cor-porate intranet systems. You can use quite a lot of qs-STAT functions there. Reports are generated dynami-cally.“

Even More Data. Do I Really Need Them All? Data compression & outsourcing ->M-QIS ReportingIn order to describe M-QIS comprehensively: There isalso the option to compress and outsource data. First,some information about outsourcing.Outsourcing helps to export certain datasets as filesautomatically at specified intervals. These datasets haveto meet certain regulations. After the export is complet-ed, the files are saved to an archive database or on anystorage medium, as required. We do not need that yet.But there will be the day when our databases exceed acertain storage capacity or contain a multitude of partsso that we would like to use this function. We will sep-arate our data. Then we will have a fast production data-base containing “few“ data and a big archive database.Data compression is another topic. I have got the optionto save describing statistics instead of measured valuesor in addition to these values. How do I apply it?We assume that I collect one million measured valueseach month for years. Actually, we do not need everymeasured value that we recorded five years ago in ourplant anymore. However, we require the capabilityindex, average, standard deviation, etc. calculated foreach single calendar week.This is what data compression does. These statistics arecalculated and saved to the 10000 measured values ofweek 14 automatically. Maybe they are even separated

by machine and batch. So, in the end, our 10000 piecesof information become about 20 in our example. Thedata are compressed. And in our case, this is sufficientfor the data recorded five years ago. The database iscompressed and I gain some storage capacity. And wesave the raw values on storage media.“M-QIS data compression and outsourcing helps to con-trol the amount of data. The compression and outsourc-ing is automated according to your own regulations.“

Other Plants Also Want the System – But theSame One, Please. Evaluation methods -> all productsJust a few more comments on the heart of the Q-DAS®

software which are the evaluation methods or the func-tionalities of the programs.Other plants of our group went along with us. However,it was important to meet the requirement that the cal-culation of statistics is the same for all plants. Q-DAS® software makesit possible by providing“evaluation methods“.They consist of a collec-tion of calculation rules,inspections / tests andrequirements.A huge number of groupand corporate strategiesexist. In most cases, thecompanies use the methods of their customers, if avail-able.The evaluation methods are another reason why you donot have to select the suitable distribution model or cal-culation method for the data yourself when workingwith Q-DAS® products. The program always determinesthem automatically and correctly. Our former Excel datarecording was only able to calculate normal distribu-tions, since Excel just offers functions for working withnormal distributions. However, the normal distributionrarely occurs. Q-DAS® programs analyze the data andselect the best suitable distribution model in order tomake a statement about the population. It does not mat-ter whether it is a Weibull, mixed, logarithmic normal orany other distribution, Q-DAS® products get it right. Supported by TEQ® (Q-DAS® Group) we defined an owncustomized evaluation method. This is our standardnow that can be imported and exported. We all use thesame calculation methods, company-wide.Moreover, you may not only exchange strategies butalso any other settings (graphics, catalogs, etc.) betweendifferent installations.“The evaluation methods help you to evaluate all thedata in the same way, company-wide. It is easy to selectspecific customer requirements or guidelines.“

Figure 19: Evaluation methods

Figure 18: M-QIS web


Q-DAS® Products

Camera Works.The CAMERA Concept -> all productsCollecting, Assessing, Managing, Evaluating, Reporting,Archiving. These are the single steps of the performancemeasuring system designed by the Q-DAS® company.We implemented this system by using all the compo-nents already described.We monitor our production. We know our products. Wehave an alarm system informing us about critical char-acteristics by e-mail. Our reports are distributed auto-matically. The customer receives less defective parts.We do not need any special trips to solve problems any-more. No loss of image. We optimize our productsdirectly at the machines. Ideal runtimes for tools. Exactmonitoring of cavities, batches and machines while pro-ducing parts. Improvement, optimization and finally:quality.A quote for a change: “The Q-DAS® CAMERA Conceptserves the purpose of efficiently introducing and design-ing a performance measurement system for qualityassessment in industrial production. “

My Conclusion:Wait! I really should stop here. Please keep in mind thatthis is just an experience report that I have made up. The author of this article, me, does not work in qualityassurance but is an employee in the system integrationdepartment of Q-DAS®. Thus he is definitely partisan,too. The employees of this department offer interna-tional workshops, support and installations for new andexisting customers. However, he did not invent all the experiences hedescribes in this report. Only the combination, i.e. thecompany and the product, does not agree with reality. This report consists of a wide range of experiences theauthor gained while completing his last installations.Maybe interested people and even long-time customerswill gather thoughts about new methods and solutionsafter reading this article.Of course, not every company requires all these com-ponents. Some of them will not be as reasonable asdescribed here. And sometimes it is not even advisableto introduce all required components immediately.A system has to grow. Users shall accept it. And itshould be flexible and provide a lot of different options.Maybe, with this in mind, we will meet soon…?

Why Validate Software?

People generally assume that software systems alwaysprovide correct results. However, these programs arehighly complex and complexity causes errors. You cannever exclude errors having a dramatic impact on theresults.

It is not possible to imagine companies that do not applysoftware today. Nowadays, software is kind of the “cen-tral nervous system“ of a company. Thus the correctnessof such systems is of utmost importance, especially incase of calculated quality indices major corporate deci-sion processes are based on. People became more andmore aware of this problem and paid more attention tosoftware validation. Software validation is even obligato-ry for officially regulated companies (pharmaceuticaland medical engineering).

What is Software Validation?

If you want to put it simply: By validating an installedsoftware system you want to furnish proof that the sys-tem delivers the performance it is supposed to deliver.

In accordance with ISO 9000:2005, validation is the“confirmation through the provision of objective evi-dence that the requirements for a specific intended useor application have been fulfilled“. You may provide thisevidence based on conformity assessments by observa-tion or evaluation and, in case of conformance, validatethe software by means of measurements, tests and com-parisons.

Software ValidationDr.-Ing. Edgar Dietrich, Stefan Weber, GmbH® & Co. KG

Dr. Berthold Lebert, Independent Consultant

43PIQ International

Q-DAS® Products

Where Are Software Validation RequirementsSpecified?The certification of a quality management system alsoincludes the auditing of the applied software system,especially systems processing quality information. ISO9001:2008 or ISO TS 16949:2009 demands in clause7.6:

”When used in the monitoring and measurement ofspecified requirements, the ability of computer soft-ware to satisfy the intended application shall be con-firmed. This shall be undertaken prior to initial use andreconfirmed as necessary.“

Since the Q-DAS® software processes measurementresults of measurement processes, the requirements ofISO 10012:2003 clause 6.2.2 also apply:

“Software used in the measurement processes and cal-culations of results shall be documented to ensure suit-ability for continued use. Software, and any revisions toit, shall be tested and/or validated prior to initial use,approved for use and archived.“

How Can You Meet the Requirements ofSoftware Validation?

You cannot “touch“ software systems like you can touchother products and thus is it not possible to evaluatethem based on common test processes. In order to val-idate the software you first need to install the softwareprogram in the environment it will be applied to. Afterimplementing the software successfully and completely,the software must be tested according to the definitionmentioned before. Release the software after complet-ing the test successfully. This test is based on scenariotesting tailored to the important factors included in therequirements specification and, if possible, covering allrelevant applications and potentials of risk.

Q-DAS® verify their software products based on stan-dard scenarios and release the products before deliver-ing them to the customer. Since the exact application atthe customer‘s is un-known to Q-DAS®, the company isnot able to guarantee an absolute freedom from error.However, customers can act on the assumption thatmost applications and especially com-puter operationswork properly and provide correct results. Customersnow face considerably less effort to test the softwaresince they only require inspections ensuring that thesoftware verified by Q-DAS® also works properly intheir system environment (computer hardware, op-erat-ing system, network, database, interfaces, etc.) and fortheir specific application(s).

It is recommended to prepare an own scenario testingthat will be applied every time the Q-DAS® software is

installed. This procedure helps to meet official validationrequirements and to release the software for applica-tion. If customers change the validated system, thechange management has to guarantee that the validatedstate of the software remains.

How Can You Validate Software?

Referring to the Q-DAS® software, there are two differ-ent test phases in general:

• Testing the calculation of statistics based on definedreference datasets (see “Eligibility statement” on theQ-DAS® website www.q-das.de/en/software/eligibility-statement).

• Testing the input of data and the correct processingbased on a specified data flow and the general pro-gram control. The user has to define the main testscenarios.

A well-structured validation process is based on areleased process description. The main components arethe description of the requirements specification, of therisk analysis and of scenario testing. You record the sin-gle qualification steps and release the software in casethere are not any relevant deviations.

Note:

Especially for the application in a regulated environ-ment, the Q-DAS® standard software products qs-STAT®,solara.MP, procella® and destra® were subject to a con-formity test in accordance with the requirements of theFDA standard 21CFR Part 11. The test result proved thatsoftware works properly and that the results it providesare correct and complete. Q-DAS® provides the reporton the conformity assessment issued by Dr. BertholdLebert, freelance counselor for quality assurance inaccordance with FDA quality standards, on request. Wewill be pleased to contact Dr. Lebert for you in case youneed further consulting services or support in validatingsoftware.


Q-DAS® Products

The MINI – who does not know this car? The car hasbeen an eye-catcher on the streets for years and sinceMINI’s cooperation with BMW, it has “grown up”. Thisis really a success story and nowadays almost 7 millionBMW MINIs drive on our streets all over the world.Meeting high quality standards is BMW-MINI’s top pri-ority. So the implementation of a SPC project based on Q-DAS® software products was quite obvious.

In the past, measurement and test data was recordedmanually by using pen and paper in the assemblydepartment. Data were collected once a week in orderto conduct a rudimentary process analysis. A real statis-tical evaluation including a “drill-down functionality”was not feasible because of the data structure. The firstQ-DAS® licenses were installed in 2003. These licenseswere single-workstation solutions, i.e. there was no sys-tem network. However, the single-workstation solutionsprovided the basis for implementing the Q-DAS® soft-ware in the entire final vehicle assembly.

BMW-MINI defined the following project aims togetherwith our English partner Measurement Solutions:● Introduction of real-time SPC in the entire vehicle

assembly● Automated data recording (instead of hand-written

on paper)


BMW-MINI Project in England

“Whenever we have a problem, we no longer simply measure it …We Q-DAS® it!”This is a clear statement issued by the manufacturing manager of the BMW-MINI plant in Oxford, England!

Figure 1: The Mini - then and now

Figure 2: System layout

45PIQ International

Q-DAS® Products

● Provide evaluation results to the managementpromptly

● Generate reports “on demand“● Analysis of the processes based on the “drill-down”

functionality● Application of the intranet for visualization purpos-

es and for the reporting system.

The project was implemented in multiple steps. In thefirst step, BMW-MINI and Measurement Solutions gen-erally specified definitions for the project requirementstogether. The result was the basic configuration of theQ-DAS® software. Based on this configuration, the firsttraining courses were provided in the following steps inorder to acquaint employees with the software and tolearn more about the options the software offers. Nowthey were able to consider the software and its functionsin the definition of further project requirements. Newdefinitions formed the basis for the installation and con-figuration of the Q-DAS® systems in the BMW-MINIplant in Oxford (procella for data recording, Q-DAS®

Web for evaluation and qs-STAT® for detailed analyses).

After some training and completing the installation, thefirst data were recorded by means of procella® in theassembly. At the same time, BMW-MINI started to spec-ify the layout of reports and websites the managementonce would access for statistical evaluations.

An easy access to the data via intranet and the integra-tion of the functionality to click through different levelsof detail - from an overall summary to single depart-

ments of the plant to individual characteristics - were oftop priority. Even this stage of the project was complet-ed successfully. Thanks to intuitive website navigationsusers only required a short introduction into the han-dling of the software. Statistical trainings were also pro-vided.

Overall the customer considered the project to be verysuccessful and beneficial and they summarized it as fol-lows:● The Q-DAS® Web application helps us to visualize

data for each employee in the entire plant● Employees do not have to be experts in statistics to

operate the system● Configuring websites and adapting them to user

requirements reduces the effort of additional train-ing

● Focus on the most important process issues● The management is always up to date about the

quality status in the processes.

„The Q-DAS® solution enables us to focus on processissues when they occur and be pro-active rather than re-active towards problems. Whenever we have a problem,we no longer simply measure it …We Q-DAS® it!“ (orig-inal quote by a BMW-MINI manager). Even after con-cluding the project, the customer and Q-DAS creatednew ideas of how to develop corporate Q-DAS® websolution.

Mini…. a success story to be continued!

Figure 3: Websites for evaluations via intranet


Q-DAS® Products

In most cases of online or offline data recording a cen-tral data management helps to perform a statistical eval-uation in qs-STAT®. qs-STAT® analyzes processes basedon current standards and guidelines. In order to elabo-rate on the applications of Q-DAS® WEB, we start toexplain the OCB WEB application first. The applicationof OCB is generally the integration of Q-DAS® WEB andqs-STAT®. This is an easy way of providing an onlineprocess evaluation and process monitoring to the user.The measured data is recorded as usual. You either entermeasured values manually or the values are recorded bymeasuring equipment and transferred via serial inter-face. However, data can also be created in the ASCII orSPS format automatically. M-QIS-S processes these dataand makes them available in the WEB application. Bymeans of hyperlinks you may display this information ina global network or, as shown in the example, on com-puters installed in production divisions.

The concept explained in the following forms the basisof the WEB application. Data of measuring instrumentsare loaded and provided in a buffer database in order toedit them in M-QIS-S and visualize them as a web appli-cation by using IIS (Microsoft Internet InformationServices). Of course, qs-STAT® is able to access thesedata, too, in case any manual evaluations are required.

Q-DAS® products are designed to provide informationabout the process capability of characteristics, e.g.diameter, coating thickness, lengths, etc. In the OCBapplication example, however, production machines andtheir tools are analyzed and thus they become the deci-sive factors in the evaluation.

This is the reason why Q-DAS® developed Q-DAS® WEBin a way that it allocates measured values to the corre-sponding machines and tools. This allocation helps youto find out whether capability is established or correc-tive action is required.

The following section shows screenshots of a real cus-tomer environment. They illustrate four stages of thetop-down navigation. Additionally, it shows what a bestfit move looks like and that an active intervention can betaken at the machines or the tools subsequently. At first,the characteristics of all machines are summarized byusing catalog records (050A1, 050A2, …) and evaluated

based on a parts evaluation. The relative number ofmeasured values outside the specification leads to thered proportion. Values having a tolerance capacity of atleast 90% are displayed in yellow and measured valuesfalling below 90% of the tolerance are shown in green.You have got the option to adjust the number of rangesand the colors in order to even make more distinct clas-sifications.

By clicking on a bar the program shows the machine theoperator wants to examine. The list of all the corre-sponding tools is displayed on a following page. Theproportion of bad characteristics per tool is also dis-played here.

You select the tool you want to observe more closelyfrom this list. The software visualizes the characteristicsand their capabilities subsequently. The individualgraphics and statistics help you to find the correctiveaction to be taken at the machines.


OCB: Have a Different Look at Your Measured Data

OCB is short for operator control board and implies an optimized application of Q-DAS® WEB (M-QIS), one ofour products that many of our customers have already used for years. The variety of functions the WEB applica-tion provides is almost unlimited.

Figure 1: Maschines in one work cycle

Figure 2: Tools in one machine

47PIQ International

Q-DAS® Products

Best Fit Move: Another Helpful Method

This function is applied to positional tolerances, e.g.required for bores. In case of bores, the permissibledeviation from the target location is defined by a toler-ance circle or a tolerance ellipse. A single correction inlocation is relatively easy to realize. But what about sev-eral bores to be analyzed simultaneously?

In many cases a machine produces several bores atonce, e.g. when manufacturing engine blocks. As soonas the location of the bores is not satisfying, the ques-tion of whether to correct the location arises. By con-sidering that setting the tools requires a relatively higheffort, operators search for a “quick“ option to correctthe location without having to adjust thetools again. The numerical output of cor-rection values includes an offset in x-direction and y-direction and a recom-mended rotation angle.

In addition, users may display the resultsof statistical analyses. The results areeither shown directly on the website or ina PDF document. Users can change thelevel of detail anytime. In order to displaymore details or less, they either use thedisplayed top-down navigation or theyinclude different results such as capabilityindices or summary graphics.

Q-DAS® Web is a flexible, adaptableapplication. In this case, Q-DAS attachedparticular importance to data compres-sion based on additional information,

such as machines and tools. Only a com-pression helps to provides highly trans-parent processes that would not be possi-ble in a mere calculation of capabilities atthe characteristics level. In these calcula-tions it would be hard to find the relationbetween single machines and tools affect-ing the process considerably.

The different display options offered byQ-DAS® Web help operators to detectdeviations and to immediately correctiveaction immediately, e.g. by changing atool just in time. In addition, the webfunctionality has the advantage that oper-ators only require a web browser in orderto gain the necessary information. Thesoftware does not have to be installed atthe workstation and the employees do notneed any specific software training sincethey only use predefined websites. The

navigation on the Q-DAS® Web website is similar tocommon websites which makes it easy to handle.

Flexible SystemSince Q-DAS® Web is configurable and easy to use, itbecomes a tool that is implemented by an increasingnumber of companies in order to display process infor-mation at different levels in the form of web portals. Q-DAS® would be pleased to support you in realizing asystem adapted to your individual requirements.

Figure 3: Characteristics summary in x-y-Plot

Figure 4: Best Fit Move


Q-DAS® Products

Approaches to the Integration of both SystemsOur customers apply the Q-DAS® software products asintegrated solutions (CAMERA Concept) but also as indi-vidual installations. Since our products are widely used,they often have to interact with MES solutions of differ-ent providers. This situation demands flexibility andopenness of the Q-DAS® company in order to offer thebest solution to common customers. The AQDEF dataformat (Q-DAS® ASCII transfer format) and the Q-DAS®

Statistical Server already provided the basis for the inter-action of both systems in the most flexible way. The following section presents different approaches.Various requirements and tasks distinguish theseapproaches. Does the MES solution already include therecording of quality data? And would you like the sys-tem to evaluate quality data and process data?

Recording and Evaluation of Quality Data byUsing Q-DAS®

Q-DAS® products plan tests for the recording of qualitydata in this approach. The MES system always contains

the current status of each machine, i.e. the order themachine or the entire operation is just producing partsfor and the required quantity to be produced. When acompany has to conduct a quality inspection for n partsproduced, the MES opens procella® and passes theheader data specific to this order to the Q-DAS® prod-uct. procella® now runs automatically in the foreground.Based on the passed header data, the program loads therequired test plan and takes over additional data for datarecording. procella® carries out the inspection (includingthe respective alarm monitoring, additional display ofhistorical values for trend identification, etc.). After com-pleting the inspection, procella® provides the MES withthe current status (alarms) and the manufacturing exe-cution system runs in the foreground again. The MES isable to react to the alarm message accordingly, e.g. itsorts out a defective part.Quality data are available permanently in the Q-DAS®

environment and can be evaluated in qs-STAT®.


Interaction of Q-DAS® Software Products andMES Solutions

The market requirements for software products are changing. The current trend shows that companies demanda uniform and universal solution leading to transparent products and processes. Companies may reach this tar-get by providing operators in the shop floor with real-time visualizations including the required details and offer-ing compact and clear displays to the management. The contents to be processed are quality-related statisticsand graphics as well as information on the status, records and workload of the processes or production facilities.

Q-DAS® focuses their software products on the display and evaluation of quality information in industrial pro-duction and processes. A MES (manufacturing execution system) offers solutions for transparent and flexibleillustration and planning of manufacturing processes reaching from the order to the finished product. Integratingboth systems provides each company a universal and value-added solution.

49PIQ International

Q-DAS® Products

Recording Process and Quality Data in the MES(Version A)The MES may also record quality data. However, theMES requires the corresponding test planning. Now theMES transfers quality data together with process data tothe Q-DAS® products. This approach does not transferstatistical evaluation results back to the MES. The MESis responsible for the entire data storage and data man-agement. It feeds data into the Q-DAS® software prod-ucts for detailed evaluation.This approach does not transfer statistical evaluationresults back to the MES. The MES is responsible for theentire data storage and data management. It feeds datainto the Q-DAS® software products for detailed evaluation.

Recording Process and Quality Data in the MES(Version B)Even in case of this version, the MES is responsible forthe planning and data storage. The MES automatically

provides the data to Q-DAS® software products for eval-uation.After the transfer of data is completed, Q-DAS® prod-ucts evaluate these data automatically. Depending onthe desired evaluation, the MES creates data packetsand transfers them, e.g. a packet for each order, productor period. After completing the evaluation, Q-DAS®

products provide the statistics to the MES for furtherdata processing. Meanwhile, the Q-DAS® softwareproducts are running in the background as a “blackbox“. The MES triggers them and they transfer theirevaluation results to the MES, e.g. numerical results,reports or even graphics. The MES always saves theseresults.These are only three approaches that only require mini-mum effort and are quick and easy to realize.Depending on customer requirements and the scope ofthe desired solution, there are further options to realizethe interaction between Q-DAS® products and MESsolutions.


Q-DAS® Products

New Graphical User InterfaceBeing able to use many functions and being flexible toadapt the program are advantages of the Q-DAS® soft-ware. However, too many settings and options leadquickly to a confusing graphical user. In order to avoidthis problem, we changed the layout of the softwarecompletely.

For a better illustration of the differences, this article isdivided into two columns. We only use a few examplesto show the differences. The left-hand column shows thecurrent version 10 and the right-hand column displaysthe differences of version 11.

Version 10Start of ProgramAfter starting the program, the program surface opens bydefault. You choose the “File” menu in order to selectwhether you want to open files or data from the data-base.

Version 11Start of ProgramAfter starting the program, the submenus “Open” and“Read from database” are displayed immediately. Inaddition, the program shows a list of the last openeddata sets.

Big icons indicate further basic functions or functionalgroups, such as “Database…“.

Open Data SetDefault graphics open, value chart by default.

Open Data SetThe new structured and well-arranged workspace opens.

Markus Pfirsching, Q-DAS® GmbH & Co. KG

Version 11

When starting the new version 11 of Q-DAS® software products, the difference compared to the previous versionis obvious at first glance. Q-DAS® completely changed the program surface. The new surface has the target tolead users quickly to the desired function. Now, the entire structure is less overloaded with icons and menus andhas a more transparent design. The layout is similar to MS Office products since it also categorizes differentapplication areas. When changing the previous version, we wanted new customers to be able to handle the soft-ware quickly and we focused on software that can be operated intuitively. Even people switching to the new ver-sion should not have problems in using the software. This is the reason why we kept all the terms and conceptsfrom previous versions and integrated an intelligent search function into the programs.

Start screen version 10

Loaded data version 10 Loaded data version 11

Start screen version 11

51PIQ International

Q-DAS® Products

Version 10Graphical SettingsYou edit graphics by selecting “Graphical settings“ fromthe context menu. There are different options availableincluding several subitems. Changes take effect onlyafter closing the dialog box.

Version 11Graphical SettingsYou may change graphical settings directly by using theicons available under “Graphical settings”. Changes takeeffect immediately. All options available for the respec-tive graphic are displayed automatically.

There are various changes compared to version 10 sothat we first describe some of them in detail.In the header of the software you switch the applicationarea. There are different menus available, such as “File“,“Start“, “Graphics“ etc. Depending on the menu youselected, the program displays the corresponding func-tions in the form of buttons or icons. We selected the“Graphics” menu in the following example.

Since the functionalities are categorized, users find andopen them quickly. The example below shows the his-togram of averages:

All the graphics are categorized, i.e. there are differentgraphics available under “Graphics“ and “Results“ buteach graphic has an own button including all its relatedoptions.

New graphics such as the pie chart provide a new kindof display. The following example shows the respectivepercentages of characteristics referring to the fourdefined classes of capability indices.

You may include buttons in any graphic dialog box. Thesebuttons help you to open further graphics easily. Theexample below shows the summary of C-values. Usersclick the buttons in order to open the value chart, QCCor histogram without selecting the respective menu item.Any number of buttons and any graphic can be includedand users may adjust the buttons for each single graphicdialog box individually. This option helps users to openrelated graphics quickly and directly.

Ribbon version 11

Graphic selection version 11

Graphical settings version 10

Graphical settings version 11

Buttons opening other related graphics

Pie chart – classified capability indices


Q-DAS® Products

These examples show that version 11 offers a better easeof use. The entire layout has become more intuitive andstructured. In addition, the program provides the “searchfor function“ functionality showing you the position ofthe respective option live in the program.

32bit or 64bit?64bit systems are gaining ground and will establish asthe standard in the long run. In version 11, you maychoose whether you want to install the 32bit or the 64bitversion of the Q-DAS® software.

Since the software has always been provided with an MSAccess database, it is important to differentiate both ver-sions. There are not any technical changes for the 32bitversion; however, the 64bit version installs and applies aMS SQL Express database instead of a MS Access data-base. When starting the Q-DAS® setup, the softwareinstalls this database automatically in case you selectedthe 64bit version.

The database connection to Oracle or MS SQL databas-es is still available as an option for each version, ofcourse.

UnicodeYou may display the program surface or reports in otherlanguages just by switching the language. However,there is something new about it. Now, you can also savethe input of operators, e.g. characteristics descriptions ornotes, in Unicode to the Q-DAS® software. This optionleads to the correct display of any character in the pro-gram and in reports – worldwide.

M-QIS Dashboard

There is a clear trend away from the single softwarelicense and toward the web solution providing variousadvantages. It particularly leads to less technical effortfor the IT staff since the user needs nothing but an inter-net browser in order to use the functionalities. Even theapplication of a web solution is intuitive since users arenormally acquainted with the handling of websites.

We thus extended the web functionalities of version 11.The new M-QIS Dashboard helps to process statistics aswell as graphics and displays significant information.

Planning of the DashboardIf you want to display information, e.g. statistics, onbrowser pages, you have to plan them at first. There isanother new aspect about version 11. First of all, thesoftware tries to illustrate the corporate structure or thestructure of the manufacturing process. This structure isdisplayed in a folder hierarchy. You may add further fold-ers (knots) individually to the hierarchy.

Based on this structure, the software creates the web-sites automatically.This example shows several graphics. In this case, youselect the tab of the process being of interest to you. The

tabs refer to the respective operations displayed in thetree structure.

We select “Line 1“ in this example.

You select the period for the results you want to observein the upper right-hand corner. Days, weeks and monthscan be chosen. We adjusted 4 days in our example.

The graphics on the web page thus refer to 4 days. Forthis reason, the software creates 4 bars automatically,one for each day. These bars show compressed informa-tion. In this case, the bars display frequencies of classi-fied process capability indices of all characteristics. Youmay select your classes individually, e.g. red correspondsto a capability index < 1 etc.

The planning of the structure shows that the parts pro-duced in line 1 may pass 4 operations that you may alsodisplay in the benchmark graphic. You will quickly noticethat there are hardly any problems in operations AG1and AG5. However, AG2 and AG2.1 include a high pro-portion of characteristics whose process capability indexdoes not meet the requirements.

Planning of the structure

Operations displayed in tabs

Selection of period

Web display – dashboard

53PIQ International

Q-DAS® Products

Benchmark graphics are interactive. A click on the “AG2“bar in the benchmark graphic opens a new page. Itshows the production machines corresponding to“AG2”. There are two machines (102 and 104) having

their own benchmarks. You may thus continue to getmore detailed information until reaching information andindividual characteristic graphics being less compressed.

You may display garphics like summary graphics or indi-vidual characteristic graphics here. Of course, you selectthe desired graphic interactively. In this case, the soft-ware already offers a small selection of graphics.

The web page creates links to previous levels automati-cally in order to trace back the hierarchy. In case you adda new knot when planning the structure, the respectivepages are added automatically. You do not have to adjustthe pages manually.

The M-QIS Dashboard combines an easy and intuitiveplanning of the web pages with the navigation and thesimple handling of results on web pages. It is easy toadapt structures and contents to your requirements.

Selection of alternative graphics

Classified capability indices of all characteristics referring to 4days

Benchmark by machine

Benchmark by operation


Q-DAS® Products

Display of characteristic

Considering that the prevailing Quality Standards(means, methods and traditional Statistics) are orientedtowards the automotive industry and its traditional man-ufacturing systems, it is surprising that even within theAutomotive Industry there is significant disagreementand in some cases outright confusion about the defini-tion of the major Quality Metrics. This has led manycompanies to author their own company guidelines inorder to clarify the Statistical Assessment of measure-ment systems, manufacturing tools and processCapabilities. These company Guidelines mainly deal withthe blind alleys left open to interpretation in the prevail-ing guidelines, bringing definition and acknowledging theneed for engineering judgment based on pragmaticanalysis.

Lost in the discussion are several key points that areindustry specific.

As it relates to the Aircraft industry, the fact that thesemachining production systems will produce fewer partswith more frequent pauses in production means that thetraditional Automotive Quality Metrics, especially thosethat assess process stability, may not be sufficient toaccurately describe the process or provide the informa-tion necessary to improve a process. Also unique toAircraft production are the exotic materials and theextensive use of general purpose tooling and machines,such as flexible Milling, Turning, Boring and Drillingmachines, which lend themselves to the manufacturingof multiple part characteristics by a single tool, wherenot every characteristic has the same level of criticality.Advances in Carbon fiber manufacturing materials andprocesses still continue at a fast pace and these process-es are unique enough to warrant special attention ontheir own as it pertains to descriptive Statistics.


Quality in ManufacturingTwo industries - Same problem

55PIQ International

Q-DAS® Products

As it relates to the Automotive industry, Governmentalenvironmental regulation and standards have drivenpowertrain manufacturing away from Steel and Irontowards the successful development of exotic materialsfor manufacturing lighter and more fuel efficient power-trains. For Automotive Body and Assembly the nowextensive use of Polymers, Die cast and thin wallstamped body parts means that they too have made sig-nificant advances in manufacturing technology.

Ironically, Automotive now has more in common withthe Aircraft industry than at any point in history and bothstill have the same significant gap when it comes to deal-ing with data collection and analysis (Quality Metrics) forpurchased components.

Regardless of the industry, the communication of QualityStandards and Requirements to part component suppli-ers are generally the responsibility of a large companiesPurchasing Supplier Quality Team and not the directresponsibility of Manufacturing or Quality Engineeringgroups. There is a tendency to rely upon the traditionalSPC methods, Quality Metrics and techniques that havemost generally been taught and used for years. Since thepredominant SPC training leans towards bending (trans-forming) the data until it fits the kind of analysis methodsthat can be calculated using simple spreadsheet and typ-ical SPC tabulation software, the resulting QualityMetrics are frequently misleading at best. At its worst,these are the same SPC methods, Quality Metrics andtechniques that are so frequently misunderstood evenwithin the same company. The communication ofrequirements to external Parts Component Supplierstherefore becomes a new challenge with every indepen-dent supplier.

Most companies will still simply rely on the traditionalSPC methods for Part Component Suppliers becausethey are in fact widely accepted as the Best Practice, eventhough the OEM companies own internal Manufacturingand Quality Engineering teams have already discoveredthe issues and developed the internal companyGuidelines that support an advanced Statistical AnalysisSoftware capabilities. This results in an unspoken phe-nomenon… where some companies might have differentQuality Standards applied for internal production sys-tems and external purchased components, while believ-ing that everyone works to the same standards.

The solution to this dilemma is the universal deploymentof Q-DAS® products across OEM's and their Suppliers,effectively merging the independent Quality Systemsinto one common system that still allows the organiza-tions to function independently.

Q-DAS® has developed and implemented systems meet-ing the Global Quality and Individual Company stan-dards as a “standard off the shelf" product that is bothcost effective and value added back to the process. TheQ-DAS® methodology is a systematic approach to theassessment of measurement, test, manufacturing and

assembly systems while providing the real-time produc-tion monitoring with exceptions reporting capability thatenables a true vision into the process. The Q-DAS®

Statistical Engine can automatically evaluate Non-Normaldata, assessing process stability according to preset busi-ness rules that understand the difference between simi-lar features that have a different relative importance(Characteristic Class), Natural Bounded and PositionalTolerance Characteristics (GD&T Features) andRelational features for example; in a Best Fit Move. Thispowerful and flexible Data Analysis Engine is coupledwith a modular products design that together supportevery phase of Production Quality:

• Collection and Recording• Analysis• Management• Evaluation• Reporting • Archiving

Collectively known as the CAMERA Concept.

Q-DAS® Data Analysis capability is unique in the worldtoday and provides sound rational methods for analysisof Normal and Non-Normal data at the same time. Thebuilt in capability to adapt the method for analysis(Configuration of Evaluation) to suit the Reason for Testmakes it possible for the product to fulfill the individualneeds of every OEM plant and supplier, while also mak-ing it possible to have a true Global System at the sametime.

These same software tools are flexible enough to supportboth High and Low production rates, traditional andexotic materials and traditional or agile manufacturingtechnologies.

Currently there are more than 85 suppliers ofMeasurement, Test and Assembly systems that areCertified to support the Q-DAS® ASCII Data Format andthat means Customers can purchase the best suitableequipment and expect that it will be compliant with a Q-DAS CAMERA® system.

Q-DAS® & TEQ Training insure that the confusion sur-rounding the Quality Metric, Methods and Techniques iseliminated and promote common understanding of prag-matic methods. These teams also support the prepara-tion of Company Guidelines that can virtually eliminatemiscommunication of Requirements and the attainmentof Quality Objectives for both OEM and Supplier alike.

In summary, regardless of your production system tech-nology, Q-DAS® has the right products and the rightexperience to provide both local and global solutionsthat can link OEM's together with their suppliers for aCommon and Effective Quality System. Contact yourlocal Q-DAS® Team to find out more about how we cantake the mystery out of your data recording and analysis.


Q-DAS® Products

Evaluation of Individuals Based on the Extended ToleranceClassification by Using Q-DAS® Products Michael Wagner, Q-DAS® GmbH & Co. KG

The evaluation of part measurements included in Q-DAS® products is implemented based on the summa-rized evaluation results of individuals.

The following criteria serve as a basis for evaluation:● Evaluation inside/outside the specification

● Evaluation based on the definition of ranges withdefinable classifications

● Evaluation inside/outside the specification consider-ing the measurement uncertainty.

The awarding of points and the definition of the colorscan be adapted to each single range individually. In addi-tion, you may weigh the single characteristics classes inthe evaluation.

The evaluation based on the definition of ranges withdefinable classifications has been realized successfullyin a customer project; however, after the process hadreached a certain condition, the classification was nolonger sensitive enough to allow for precise evaluationsand reactions. In this case, it is important to consider theaspect of evaluating measured values by means of savedclassifications (comparable to the meaning or severity ofthe consequences of error from the customer’s point of

view in an FMEA environment) together with the toler-ance capacity. A classification of deviations should befeasible for each element of the created matrix. Hence,the definition of the parts evaluation available in theadjustable evaluation methods was extended in a waythat these requirements can be indicated:

In this very example we created 7 sections for the toler-ance utilization and 10 FMEA classes including 3 evalu-ation levels. You may save an individual points ratingand the corresponding color to each level.

The company guidelines specify different reactions forthe respective levels C-M-S (C = critical failure, M =major failure, S = standard failure). Each employeeinvolved in the manufacturing process knows thesereactions.

Moreover, a new K-field (K2190) was defined allowingfor the categorization of a characteristic according to theFMEA classes (0-10). Since you want to continue usingthe existing datasets, this classification must be added tothe characteristics data successively. In order that youdo not have to adapt all the characteristics of a certain

57PIQ International

Q-DAS® Products

type of part, special configurations have been saved tothe “Characteristics statistics” graphic. Now it showsonly the characteristics whose K-field 2190 is still emp-ty. In addition, the characteristics are sorted by the num-ber of tolerance violations. Thus all the missing infor-mation can be supplemented very efficiently by meansof the table of characteristics. As soon as all the missinginformation is added, you may evaluate the data direct-ly by using qs-STAT® or O-QIS.

The following graphics are available for evaluation:● Characteristics-related summary of measurements

displaying the characteristics indicating the mostCMS violations first. In addition, the graphic shows the CMS classes cor-responding to the respective characteristics. Thisgraphic makes it quite easy to identify the charac-teristics’ main errors in case there are already sev-eral part measurements available.

● Examining the table of individuals for all characteris-tics.You show the classification of each measured valuefor all characteristics.

● Points rating referring to single measurements: You display the results referring to the respectivemeasured part. Once again, the correspondingclasses are easy to identify.

● Display in the parts protocol:The rating can also be displayed in the parts proto-col and thus e.g. it may help to evaluate the lastmeasurement in the CMM-Reporting.


Q-DAS® Products

AQDEF® standardizes the recording and cross-systemexchange of quality information and is thus ground-breaking in metrology. As experience in practice impres-sively shows, the data exchange format helps to reachthe consistently demanded simplification, cost cuttingand acceleration of data management.

AQDEF® is based on the Q-DAS® ASCII transfer formatthat has been established in the automotive industry fora long time and became a requirement for the suppliersof measuring equipment that has been specified in cor-porate guidelines at an early stage. Companies like Audi,BMW, Bosch and General Motors were pioneers of thispractice. By consistently demanding a certified datainterface in the Q-DAS® ASCII transfer format, the cor-rect provision of data was ensured for statistical evalua-tion through standard tools like qs-STAT®. The advan-tages of a standardized data format for the company-wide recording and processing of quality informationquickly became obvious. The compact, simple andtransparent structure, its high flexibility and the specifi-cation revealed right from the start also contributed tothe increasing acceptance and use of this format.

Due to its successfulness, many different corporate andplant-specific individual solutions of the data interfacehave been created over the years. They all have a simi-lar approach in common; however, these solutionsrequire customized modifications in each individualcase affecting costs and the introduction effort. In orderto establish a company-wide standard, users of the datainterface from the automotive and supplier industry cre-ated a work group developing a uniform and adaptedspecification. Representatives of renowned companieslike Adam Opel AG, BMW Group, Daimler AG, FordWerke GmbH, General Motors Company, GetragCorporate Group, Robert Bosch GmbH and VolkswagenAG specified AQDEF® and first published it in 2006.Back then this data format was called AutomotiveQuality Data Exchange Format. Since the format is usedby a rising number of companies even outside the auto-motive industry, the format was finally renamed asAdvanced Quality Exchange Format in 2011. MeanwhileAQDEF® has become a cross-industry interface stan-dard the work group evaluates and develops at regularintervals.

Centerpiece of the AQDEF® specification is a list of datafields an interface has to support and that can be acti-vated or deactivated. This amount of fields also provides

the basis for a uniform and comprehensive certificationof the interface that does not have to be made specifi-cally in each company anymore. Different certificationcategories consider, where necessary, the limitedamount of fields for different measuring tasks and appli-cations. A list of all interface certificates that havealready been issued is available on the Q-DAS® websiteunder www.q-das.de. There you will find leading manu-facturers of measuring instruments such as Carl ZeissGmbH, Hommel-Etamic GmbH, Klingelnberg GmbH,Marposs and promess GmbH. Q-DAS® is also responsi-ble for the certification process in accordance withrequirements of the AQDEF® specification.

Based on the feedback the work group received, theexperiences of the members of the work group and theseminars provided to manufacturers of measuringinstruments at regular intervals, we determined the fol-lowing aspects to characterize the benefits and advan-tages of AQDEF®: ● Uniform exchange format for any kind of industry● Binding amount of fields transferred via interface● Adaptable to the respective measuring task due to

fields that can be activated optionally● Reduced effort in implementing interfaces● Complete and structurally correct data for an imme-

diate visualization and evaluation with Q-DAS®

products● Certification of the data interface provides safety● Worldwide application as a specification in the pur-

chase of measuring instruments● Groundbreaking standardization of data exchange

in the future● Meeting a high number of requirements based on a

single interface definition● No double data maintenance and converter solu-

tions required anymore● Cost savings and minimization of transmission

errors.

In addition to the AQDEF® specification, the work groupalso created a user specification providing helpful infor-mation to the developers of data interfaces. We collect-ed many requirements and examples regarding the han-dling and user friendliness of the interface in this userspecification since they are an integral part of therespective data logging software.

A Vision Becomes a Reality – Global Industry Standard for the Transferof Measured Data

AQDEF® – Advanced Quality Data Exchange FormatStephan Niemczyk, Opel AG, Rüsselsheim

Stefan Weber, Q-DAS® GmbH & Co. KG

59PIQ International

Q-DAS® Products

History of AQDEF®

Additional InformationThe specification and documents of the AQDEF® work group are published on the Q-DAS® website under www.q-das.de. The same applies to the documents about the certification of interfaces and already issued certificates.

ContactQ-DAS® GmbH & Co. KG- AQDEF -Eisleber Straße 2

Figure 1: AQDEF® Basic structure

1996 Q-DAS® and Ford developed together the Q-DAS® ASCII transfer format.

2005 Initiation of a data format work group with the aim to define a standard for the certification of data fieldsand interfaces.

2006 Release of the first version of the ”Automotive Quality Data Exchange Format“ specification.

2008 Version 2.0 was released and supplemented by the user specification.

2010 Work group published the advanced version 3.0 of the AQDEF® specification.

2010 AQDEF® became a registered trademark of Q-DAS® GmbH & Co. KG. It represented a standardized dataexchange between measuring instruments and Q-DAS® software products.

2011 Since there has been an increasing demand for this format even outside the automotive industry, it wasrenamed as “Advanced Quality Data Exchange Format“.

69469 WeinheimPhone: +49 6201 3941-0Mail: [email protected]

Most companies are concerned with different matters ofquality assurance and deal with them by using variousmethods. There is the planning and organizational leveland the level of operation.

Planning and Organizational LevelBefore producing products, these products are oftensubject of an extensive planning phase. Companiesapply various methods and tools in this planning phase,such as FMEA, control plan and initial sampling. Duringseries production or afterwards neither complaints northe inspection equipment management play a role. iqsoffers the tools required on this level.

Level of OperationThe specifications of the planning level are realized onthe level of operation, e.g. by conducting measurementsystem, machine and process capability analyses.

There is a close connection between the planning phaseand the real implementation prior to, during and afterseries production. This is the reason why Q-DAS® coop-erates with iqs.

Merging Both LevelsThe aim of the cooperation between Q-DAS® and iqs isto make both levels permanently available to usersregardless of whether they apply the iqs or Q-DAS®

software. As an example, users do not have to create aninspection plan in both systems; it is only available onceand users can access it in both systems for respectiveapplications. This option reduces organizational effortson the one hand and offers a wider application range ofcentralized data on the other hand. In the end, process-es become more transparent since users are able toaccess all information on a product. Operators get asummary of any information quickly starting with thecreation of the inspection plan based on CAD data, toFMEA, initial sampling, data recording and capabilityanalysis, right through to complaints management.

Inspection Plan and Recording of MeasuredDataiqs and Q-DAS® synchronize data via WebService inter-face. In order to do so, iqs fills a cache memory withinspection plan information after the iqs systemreleased the corresponding inspection plan. Only theseinspection plans will be available in procella® for record-ing data. Moreover, this functionality defines any furtherinformation to be transferred, e.g. which characteristicsshall be included in a SPC measurement and whetheryou want to transfer information from drawings (draw-ing details).

procella® accesses this information directly. For this rea-son you may use the inspection plan immediately forrecording data without having to create it in procella®

first. Now procella® conducts the inspections and savesthe measured data in the Q-DAS® database. You mayuse them later for statistical analysis.

Since procella® guides operators through the measure-ment process, it is quite useful that CAD drawingsalready saved in the iqs system are also transferred anddisplayed in procella®. Thus operators display the draw-ing of each characteristic including the position of thecharacteristic (stamp).

Current Inspection Plans Always AvailableIn case the iqs inspection plan changes, these changesare automatically transferred to procella®. Every timethe operator opens an iqs inspection plan in procella®,the program checks whether there are changes in theinspection plan. Changes are e.g. modified tolerancelimits or new characteristics to be inspected. This check

Stephan Sprink, Markus Pfirsching, Q-DAS® GmbH & Co. KG

Interaction of iqs and Q-DAS® Software Products

In order to benefit from the advantages of the iqs CAQ software and to use the strength of the statistical analy-sis of Q-DAS®, both companies formed a partnership and now provide their software products as one package.Both companies started this development by combining the iqs Inspection Plan with the Q-DAS® SPC datarecording software procella® and by connecting the iqs Inspection Equipment Management to the Q-DAS® soft-ware solara.MP for measurement process capability analysis.

Stamp iqs drawing and select characteristics


Partner

61PIQ International

Partner

Frank Graether, Director QMBehr GmbH & Co. KG, Pforzheim

“Through use of

iqs CAQ our processes are significantly more efficient and transparent.”

CAQ Software from iqs measurably

improves your quality, effectively

reduces your costs, helps to prevent

errors and creates a continuous

quality control loop making these

factors permanent.

l Find out more about the CAQ Software for your business:

Phone: +49 7223 28148-0

What can you expect from your CAQ Software?

www.iqs.de

Quality reaps rewards.

ensures that the program alwaystransfers and analyzes the correctinformation.

The implementation of this conceptonly requires the installation of theiqs software and procella®. Any nec-essary settings are adjusted in procel-la®. However, these settings are adjust-ed automatically when installing procel-la®. The operator only has to select the“iqs Inspection Plan” function in orderto load the current data from the sharedmemory.

In the future, procella® will also trans -fer measurement results to the iqsCAQ system after the measurement iscompleted. Especially in case of devi-ations, it is reasonable to process thiskind of information. The feedbackprovided by procella® can be appliedfor further use in the complaints ma -nage ment, action management, spe-cial releases and FMEA.

Inspection Equip -ment Ma nage mentand Mea sure mentProcess CapabilityThere is still an interactionbetween the iqs moduleInspection Equip ment Ma -nage ment and the Q-DAS®

software solara.MP formeasurement process ca -pa bility. The iqs InspectionEquip ment Ma nage mentorganizes inspection equip -ment. Measuring equip-

ment or measurement system capa-bility analyses at regular intervals arepart of this organization. It seemsobvious to make these capabilityanalyses by using the Q-DAS® soft-ware solara.MP intended for this use.iqs transfers the required informationto solara.MP. The program now opensautomatically and starts the respec-tive statistical calculations. solara.MPis able to create reports on a capabil-ity analysis as a PDF document or toprovide defined statistics (evaluationresults). The iqs system imports theseresults and allocates them to the cor-responding inspection equipment. Inthe end, users gain a system that isresponsible for the management - theiqs Inspection Equipment Manage -ment - and a system performingdetailed analyses - solara.MP. Due tothe interaction between these twocomponents, users are able to benefitfrom the advantages of both softwareproducts.

CAD drawing from iqs displayed in procella®


Partner

Q-DAS® and QUINDOS® - two experts, who over theyears have established, formed and expanded the fieldof dimensional metrology. Since the eighties, both com-panies have remained true to their philosophy of stan-dardization and have grown as a consequence of theirexperience in quality management and analysis.

1992 came the first crossing of the ways. The dataexchange format DFT/DFI was implemented in QUIN-DOS. The subsequent development of the Q-DAS®

ASCII transfer format DFQ was in turn also implement-ed in QUINDOS. From 2006, the Data format AQDEF(Automotive Quality Data Exchange Format) has been aconstant QUINDOS companion and was the first soft-ware to be certified by Q-DAS® in 2006.

The statistics module from QUINDOS has grown into areliable analysis package for the automotive, aviationand general engineering industries for decades and isrecognized as a reliable process and measurement sys-tem analysis tool local to the machine.

The consequent standardization of QUINDOS (QualityInspection of Dimensional Object and Sizes) since itsbirth in 1985 has produced a measuring software withconsistence and compatibility for various technologies.

The hitherto most extensive approach to standardizationis the I + + DME (Dimensional Measuring Equipment)Interface, which can be applied to coordinate measuringmachines, form testers and other types of length mea-suring instruments, with tactile, optical as well as multi-sensor systems.

The latest software product from Hexagon MetrologyPowerTrain Solutions GmbH, the I + + Simulator is alsobased on the internationally recognized interface of thesame name. The roll out was at ‘Q-DAS® – aktuell’ in2007. From day 1, offline programming was here tostay.

Programs can be created efficiently in a virtual measur-ing room environment.

The software package allows the simulation of a com-plete measurement process in connection with any I ++ compliant measurement and analysis software.

Machine, sensors, tools, tool changers, parts and fix-tures are integrated into a realistic scenario by the pro-grammer. The resulting measuring program is collision-free and produces realistic measurement values.

Equipment and programs can be prepared off line andare 100% ready to go at production start.

Using this concept, Hexagon Metrology PTS GmbH hastaken offline programming to a new level, away fromjust the visualization of the measurement environmentand process. The 4th Release version of the standaloneI + + Simulator is now available.

For more than 20 years the requirements and lessonslearnt from customer powertrain projects have beenimplemented in the software packages developed byHexagon Metrology PTS GmbH, formerly MesstechnikWetzlar GmbH.

Hexagon Metrology PTS

Coordinate Metrology, Statistic and Quality Management through the ages - QUINDOS® congratulates Q-DAS®

QUINDOS through the ages

63PIQ International

Partner

MEASURE -

AND ANALYSE

OFFLINE!

The new QUINDOS Reshaper software option

allows you to measure and analyse

parts completely offl ine. You supply the

point cloud, QUINDOS Reshaper does the rest

- independent of hardware availability. A digital

representation of the real part is generated

and measured against the nominal data, making

full use of QUINDOS - the most comprehensive

measurement software on the market!

Learn more about Hexagon Metrology and the

advantages of offl ine measurement:

http://hex.ag/ppAVk

This experience has been used to develop a software plat-form which can be used to a wide range of tasks, frommachine runoffs and monitoring, through complex analy-sis of individual features to blisk measurement and evalu-ation.

QUINDOS is a modularized measurement and evaluationsoftware. The multitude of options allows the software togrow in parallel with the competence and experience ofthe user. Here the user can select from an options pool inorder to expand the basic package as and when required.

The range of powertrain modules includes, among others,CMM inspection and acceptance, gears and gear cuttingtools and other powertrain-related special geometries.

Plus a variety of other modules for curves, wear measure-ment or the automated measurement of parts on pallets.

All of the above options were developed in the context ofpractical customer requirements.

Through our participation in various committees we canimplement the latest technological requirements of themachine manufacturers as well as national and interna-tional standards and make them available in QUINDOS forfield testing within a very short timeframe.

Visit us at www.QUINDOS.com and let yourself beinspired by our expertise.

QUINDOS user interfaces


TEQ® Training and Consulting

Risks of an Intuitive Definition of Control Limits for a Shewhart Control Chart

Companies maintain quality control charts a thousandtimes a day in statistical process control (SPC) and usethe provided results to gain capable and stable process-es. It is of particular importance to check the averagelocation of the main product characteristics for nominalvalue and tolerance center. You usually use a Shewhartaverage chart for this task. Walter Andrew Shewhartwas the first to introduce quality control charts in 1924and published them in his book ”Economic Control ofQuality of Manufactured Product“ in 1931.

The design of an average chart focuses on a specificquestion - how do you choose control limits?

The upper control limit is referred to as UCL whereasLCL stands for lower control limit. Normal distribution issupposed to be a suitable model in order to describe themonitored characteristic and thus the calculation of theShewhart average chart is based on this distributionmodel.

1. Intuitive Definition of Control Limits Relating tothe Tolerance

People sometimes use average charts in practice whosecontrol limits are not based on statistical relations. Theyjust select them intuitively, even though they claim theselimits to be pragmatic. As an example, they define thecontrol limits in a way that the distance between UCLand LCL amounts to 70% of the tolerance and the twolimits are symmetric with respect to the tolerance cen-ter.

2. Definition of Control Limits Based on a StatisticalApproach

The classical approach is to estimate the parameters ofthe normal distribution, the expectation µ and the stan-dard deviation σ in the lead time. The accuracy of the

estimation depends on the number of units inspected inthis lead time. In general, many standards demand aminimum number of 125 units e.g. divided into sub-samples of n=5.

The distribution of individuals is known from the normaldistribution parameters determined in the lead time. Youuse the known distribution in order to calculate theexpected distribution of averages. And based on the dis-tribution of averages you calculate control limits. Thedistribution of averages only differs from the distributionof individuals by a slight standard deviation. This leadsto:

where standard deviation of averages

σ standard deviation of individualsn sample size of average chart

You define the control limits of the average chart youwant to maintain based on the distribution of averages(Figure 1).

It is common practice to use the 99% or 99,73% randomvariation range for the calculation of control limits. Thefollowing example uses the 99,73% random variationrange, i.e. the control limits correspond to the dddddddlimits of the distribution of averages. This leads to thefollowing formula for the calculation of the control lim-its applied in the average chart:

You may either replace µ0 by ● the tolerance center (nominal value) or● the estimate of the expectation from the lead

time. However, the tolerance center is gen-erally preferred. By using the toler-ance center, the average chart con-trols the required nominal value inthe long run. In process control, thisis more reasonable than controllingan average estimated from the leadtime data that is also subject to a cer-tain degree of uncertainty.

For σ you use the estimate calcu-lated in the lead time.

Dipl.-Ing. Frank Stockhaus, TEQ® Training & Consulting GmbH

Figure 1: Defining the control limits of the average chart (principle)

UCLLCL n

= ± ⋅μ σ0 3

65PIQ International


3. Comparison

We now compare both approaches for an average chartwith the sample size n=5 based on a specific processsituation. This situation meets the minimum require-ments of most automotive company guidelines. Theyrequire a process capability index of at leastCp=Cpk=1,33.

In this case, the standard deviation amounts toand the expectation is exactly in the middle ofthe tolerance center Tm.

Insert these values and n=5 into the calculation formu-la for the control limits of the Shewhart average chart inorder to gain the following equation.

I.e. the distance between the control limits amounts toabout 33,5% of the tolerance.

Figure 2 shows the control limits of the “real“ Shewhartaverage chart (UCL, LCL) and the limits of the 70% chart(UCL*, LCL*) without statistical basis for comparison. Inorder to understand the context easily, we calculated thecharts for a characteristic includingthe specification of (0±1)mm.

First you notice the considerable dif-ference between the control limitsin Figure 2. The distance betweenthe control limits of the intuitiveapproach is much wider. In addition,you see that the sample size n andthe standard deviation σ influencethe distance between UCL and LCLconsiderably in the “real” Shewhartaverage chart. The intuitiveapproach completely ignores thesetwo parameters.

In case the process location moves away from the nom-inal value µ0, the control chart will indicate this shift inlocation through averages outside the control limits.How likely is the average chart to detect such a processfailure? Figure 3 already shows that this control proba-bility (1-Pa) depends on the shift of the expectation.

The graphical display of the control probability subjectto the shift in average Dµ compared to the nominal val-ue leads to the operation characteristic. Figure 4 showsthe operation characteristics of both charts. The shift inaverage plotted on the abscissa is standardized withrespect to the tolerance.

Meaning of the symbols:

1-Pa control probability of the control chartΔµ deviation of the current expectation from the

nominal value µ0 (Δµ=|µ*- µ0|)T tolerance

You notice that the 70% chart indicates the shift of theexpectation much later than the real Shewhart averagechart.

Figure 3: Calculating the control probability of the average chart (principle)

Figure 2: Comparison between the “real“ Shewhart averagechart and the 70% chart

8

T� �

UCLLCL n

T Tm= ± ⋅ = ± ⋅μ σ0 3 0 1677,

Figure 4: Control line for the Shewhart average chart and the70% chart in case of n=5



The 70% “control chart“ is not able to indicate relevantchanges in the process caused by a sudden or steadyshift or trend of the expectation in time. This chart onlytriggers a possible alarm after the process has alreadyproduced a considerable share of rejects!

The following example illustrates this fact. In case thetrue expectation of our characteristic used in Figure 2increases from the nominal value of µ0=0 mm toµ*=0,7 mm, it refers to Δµ=|µ*- µ0|=0,7 mm.Referring to the tolerance T=USL-LSL=1-(-1)=2, thedeviation amounts to Δµ/T=0,7/2=0,35.

Figure 4 thus shows a control probability of 50% for the70% chart, i.e. the chance to detect this considerableshift in process location only amounts to 50% whenusing the 70% chart. The inability of the intuitive 70%chart to detect changes in the process becomes evenclearer when calculating the expected share of rejectsfor the process failure observed.

You gain the result shown in Figure 5 with the help ofthe qs-STAT functionality Extras|Probability distribution.This leads to an expected share of rejects amounting to11,5%.

Compared to the 70% chart, the control probability ofthe “real” Shewhart average chart amounts to about100% for the shift in process location as mentionedbefore!

Summary

As a result, the calculation of Shewhart average chartsshould always include the definition of control limitsbased on statistics as described in section 2.

In case you cannot make any lead time inspections, youmay calculate the standard deviation σ from the mini-mum requirement of Cp in exceptional cases:

However, please consider that the true standard devia-tion of the process may differ from the calculated devi-ation. For this reason, it makes sense to determine thestandard deviation empirically after recording a certainnumber of values in the control chart. Now you mayeven calculate control limits again.

If possible, try to avoid control limits for average chartsthat are defined intuitively based on certain %-regions ofthe tolerance. Otherwise, it is advisable to at least chal-lenge your limits in order to find out whether they areable to detect process failures.

Figure 5: Yield and share of rejects in case of a shift in location of Δµ/T=0,35

6p

T

C� �

�

67PIQ International


While developing the ISO 22514 series of standards, itis about time to think about the meaning and purpose ofa capability index once again.

You often face capability indices for the first time whenevaluating the number of defective parts. You can findcountless tables in technical literature specifying frac-tions nonconforming expected for each specific capabil-ity index. This mainly superseded calculation methodbased on fractions nonconforming is still contained inISO/TR 22514-4; however, only with reasonable restric-tions. Almost 15 years ago, this method was even astandard in some parts of the automotive industry.

But why using quite an abstract capability index insteadof “ppm”, many people asked themselves soon though.In case of capability indices exceeding 1,67 consider-ably, estimating fractions nonconforming is rather com-parable to gazing into a crystal ball. The benefit washighly doubtful indeed. This capability index only pro-vides information about the process output but does notindicate anything about the properties of the process orthe process behavior. Moreover, completely differentprocess qualities and process structures might lead toabsolutely identical results (see figure 1). Consequently,even this capability index did not help to discover thefactors to be optimized in order to reach full potential.

Other calculation methods calculate the capability indexfrom distribution parameters and provide considerablymore information.

All the methods ISO 21747 already summarized as“general geometric method“ in 2006/2007 offer clearinformation about the process variation and processlocation relative to the specified characteristics toler-ance. The gathered information is based on Cp and Cpk.Today, ISO 22514-2 is mainly about these methodsincluding the classical ”6•σ“ method and the percentileor quantile method. Provided that the determined distri-bution model is correct, you may also derive any poten-

tial for improvement from these capability indicesimmediately. In case of two-sided characteristics, Cp

describes the variation behavior and indicates the maxi-mum capability for an optimal location whereas Cpk

describes the quality that was actually achieved for thecurrent process location. In case Cp is insufficient, youhave to aim at optimizing the process. If Cp is sufficientbut Cpk is not, you need nothing but a correction of loca-tion in most cases. The amount of information gatheredby using this calculation method is considerably higherthan the amount of information obtained with themethod of excess fractions (see figure 2).

Dipl.Ing. Stephan Conrad, TEQ® Training & Consulting GmbH

Which Information Do Capability Indices Provide?

Example

Principle of Maximum Information

Figure 1: The Cpk of both processes amounts to 1,30 and was calculated from the fractions nonconforming. This capability index isthus not able to describe the differences between processes.



However, there are phenomena that even the “generalgeometric method” is not able to describe sufficiently.We need new technology in order to record the higheramount of information, e.g. for hole patterns or balanc-ing machines. The evaluation of a position according toDIN only considers the radius or the circle diameter the

center of the borehole lies on. You may easily calculatecapability indices in this situation by using the “generalgeometric method”, however, there is a catch. The fol-lowing two hole patterns have the same calculated capa-bility indices Cp/Cpk even though they demonstrate total-ly different qualities (see figure 3).

Example

Figure 3: Both hole patterns have the same (one-dimensional) capability indices Cp/Cpk since the deviation from the radius is the same

Figure 2: The capability indices calculated according to the percentile method better describe the real process behavior and provideadditional information about possible process optimization.

(Cm=1.64)

Cmk=1.98

(Cm=1.64)

Cmk=1.98


In the first case the average is quite in the center but thevariation is too high. The second case shows a processthat is not in the center but includes considerably lessvariation. Since we only observe radiuses without anyspecified angles, we are not able to use these additionalpieces of information.

ISO 22514-6 offers the solution to the problem. Basedon the two-dimensional normal distribution you calcu-late the capability indices Po/Pok since they can describedifferent qualities analogously to Cp/Cpk (see figure 4).Of course, you may estimate fractions nonconforming

and calculate capability indices even based on thesehole patterns and the two-dimensional normal distribu-tion – but you will lose all the gathered informationabout location and variation.

Which conclusions can you draw now? You shouldalways calculate capability indices in a way that yougather the most possible amount of information, i.e.:

• only estimate capability indices from fractions non-conforming if there is nothing but information on theproportions of defects available. Consequently, youonly use the calculation method based on fractions

Example

Figure 4: The two-dimensional capability indices Po/Pok are able to describe the different process qualities

(Po=1.68)

Pok=1.56

(Po=2.88)

Pok=2.18



nonconforming in the evaluation of discrete charac-teristics. You should do without these methods assoon as measured values are available to you.

• calculate capability indices based on the “general geo-metric method” in case of any one-dimensional char-acteristics. One-dimensional characteristics are char-acteristics that only need a single measured value inorder to be displayed comprehensively, e.g. length,weight, hardness, etc.

• always calculate two-dimensional capability indices incase there are two pieces of information required inorder to describe the characteristic comprehensively,e.g. x/y-coordinate, radius and angle, mass momentand angle (imbalance).

And now the whole issue gets even more complicated.Two-dimensional and multidimensional characteristicsare also characteristics that are linked e.g. by the maxi-mum material principle. There are also various mea-sures of shape and measures of location describing spa-tial phenomena (e.g. cylindrical shape). In case there isan interaction between more than two characteristics -e.g. in welding or injection molding where you try toassure the product quality by controlling the processcharacteristics – you may even calculate multidimen-sional (multivariate or multicriterial) capabilities. Thesecapabilities exceed our three dimensions of space; thusthey are beyond our spatial sense. The mathematicsbehind it now becomes quite complex. Fortunately, ISO22514-6 provides a solution based on “assessmentfunctions“. However, select them with caution in ordernot to reduce a multidimensional problem to a simplecalculation method based on fractions nonconformingdue to an oversimplified assessment function.

Even in the simple daily application you may reflectabout the principle of “maximum information”. In caseof discretized characteristics, measured values are “sim-plified” and reduced to error proportions. Some infor-mation about these characteristics gets lost. This factalso applies to automated test systems actually measur-ing parts but only counting defective parts. Anotherexample is a one-sided limit having a second “natural”boundary. In most of these cases, only the capabilityindex Cpk is calculated even though Cp could also be cal-culated and would offer some more information. PIQ2/2012 [5] described this issue in detail. However, manycompanies do not calculate this capability index for verypragmatic reasons – and thus they do without the addi-tional knowledge of their processes.

The target of the new ISO 22514 series of standards isthe acquisition of information. It is about reaching anunderstanding of processes, identifying optimizationpotentials and gaining confidence in your own process-es and the processes of suppliers. In the end it is a ques-tion of gathering and maximizing information. In orderto succeed in doing so, always select the optimal calcu-lation methods depending on the respective situation.

Literature

[1] ISO/DIS 22514-2: 2011-03Statistical methods in process management - Capabilityand performance - Part 2: Process capability and perfor-mance of time-dependent process models.Statistische Verfahren im Prozessmanagement - Fähig -keit und Leistung - Teil 2: Prozessleistungs- und Prozess -fähigkeitskenngrößen.

[2] ISO/TR 22514-4:2007-12 (E)Statistical methods in process management - Capabilityand performance - Part 4: Process capability estimatesand performance measures.Statistische Verfahren im Prozessmanagement - Fähig -keit und Leistung - Teil 4: Prozessfähig keitsschätzer undProzessleistungsmaße.

[3] ISO 22514-6: 2013-02 Statistical methods in process management - Capabilityand performance - Part 6: Process capability statisticsfor characteristics following a multivariate normal distri-bution.

Statistische Verfahren im Prozessmanagement - Fähig -keit und Leistung - Teil 6: Prozessfähigkeits kennwertefür mehrdimensional normalverteilte Merkmale.

[4] DIN ISO 21747:2007-03Statistische Verfahren - Prozessleistungs- und Prozess -fähigkeitskenngrößen für kontinuierliche Qualitäts -merkmale. Statistical methods - Process performance and capabili-ty statistics for measured quality characteristics (ISO21747:2006-07).

[5] Conrad, StephanEinseitige und natürliche Toleranzen in der Prozess -fähigkeitPIQ® Partner Info Qualität, 2. Ausgabe 2012Q-DAS® GmbH & Co. KG, Weinheim

71PIQ International


This essay explains how to create and evaluate a regres-sion study based on process data. The aim is to find anempirical model y=f(x1, x2, x3…) for our process dataexplaining their impact on the response.

The injection molding process of a thermoplastic pro-duced components causing problems during the assem-bly due to high shrinkage. In order to find out about anymethod to reduce the percentage shrinkage, data aboutthe injection temperature, injection speed and hold-ing pressure were collected. The general model equa-tion leads to percentage shrinkage = f (injection tem-perature, injection speed, holding pressure).

Which one of the three influencing factors has the majorimpact on the percentage shrinkage of the injectionmolding process? And how do you have to adjust theinfluencing factors in order to minimize the percentageshrinkage of the injection molded parts as far as possi-ble?

The following reliable data about the process have beengained:

The average percentage shrinkage amounts to 0,886%.

1 Selection of Regression Method – MultipleLinear Regression and Model Structure

The software destra® supports users in creating andevaluating process data, field data and experimental dataobtained through systematic planning. destra® helps toanswer questions about the main factors affecting theprocess (their absolute value) and their effective direc-tion (their algebraic sign), about interactions and abouthow to adjust factors optimally.

The collected process data are inspected in a multiplelinear regression analysis.

There are two different approaches available. You mayselect and compare the multiple linear regression andthe multiple quasilinear regression.

Click the “next“ icon to start the evaluation.

2 Evaluation of a Multiple Linear Regression

The evaluation result provides an overview of importantstatistics describing the quality of the regression model.

The coefficient of determination (R²) = 77,68% is quite high. (R²) indicates the degree to which the vari-ance of the response (percentage shrinkage) can beexplained by the three influencing factors or howstrongly the three selected influencing factors affect thevariation of the shrinkage values.

The adjusted coefficient of determination (R*²) is afair instrument to compare alternative models, e.g. thelinear and the quadratic model. It considers the numberof regression parameters and the number of measuredvalues.

The variation of the percentage shrinkage that cannot beexplained by the selected influencing factors is given inthe residual variance (s²) or the residual standarddeviation (s).

An Approach to Process ImprovementMultiple Linear Regression Dr. rer. nat. Thomas Pfeilsticker, TEQ® Training & Consulting GmbH



3 Estimation of Coefficients (Estimation ofParameters)

In order to calculate the regression model you estimatethe minimum sum of square deviations (squared resid-uals). There is no regression model including a smallersum of square deviations.

After describing the responses and factors, you specifythe regression coefficients (bi) and their confidenceintervals (95%, confidence level) bi (…). In order toevaluate the regression coefficients the software pro-vides the respective standard deviations sbi (bi) and thet-test statistic calculated by dividing the regressioncoefficient by its standard deviation. In addition, thereare three options available to evaluate the significance.The program highlights the t-test statistic with a super-script star, offers a bar chart with the three usual signif-icance levels marked by red lines and provides the P-val-ue as an alternative.

In case the α = 5% limit is deceeded, the respectiveinfluencing factor is considered to be significant and thecorresponding bar is displayed in yellow. If even the α =1% limit is deceeded, the corresponding bar exceeds thesecond red line, the software will show a red bar. The P-value must be less than 5% (0,05). This value indicatesthe residual risk to detect a significant value that is actu-ally insignificant. Injection temperature and holdingpressure prove to be significant but the injection speedis not.

The VIF values (variance inflation factor) show whetherthere is interdependence between the single influencingfactors.

Regression models are often reduced, i.e. they areadjusted for insignificant influencing factors and theirinteractions accordingly. In this case the software recal-culates the model and evaluates it again.

Injection temperature and holding pressure have aparticularly strong impact on the percentage shrinkage.The reduced model shows this fact even more clearly.

By raising the injection temperature by 1°C the percent-age shrinkage increases by about 0,026% and raising theholding pressure by 1MPa means reducing the percent-age shrinkage by about 0,018%.

4 Model Evaluation

4.1 Model Significance

The F-test analyzes whether the influencing factors(coefficients) have a joint significant impact on theresponse and do not affect it individually as explained insection 3 about the t-test of coefficients. In case the nullhypothesis is rejected (red), at least one factor is signif-icant.

4.2 Model Fit – Linearity F-test (LoF)

In order to evaluate a model, you use a linearity F-testbased on the analysis of variance in addition to the coef-ficient of determination and the significance t-test. Thelinearity F-test is also referred to as lack-of-fit test (LoF)and helps to analyze the fit of the model. However, thetest requires several observations of identical processadjustments and thus it is particularly important forplanned experimental data.

5 Overview through Response Surface Plot andContour Plot

73PIQ International


The graphics show the impact of two influencing factorson the response in case the conditions of the third fac-tor remain constant.

One of these graphics shows the 3-dimensional displayof the variable before reducing the model for the quasi-linear approach (interactions included). The other oneprojects the percentage shrinkage onto the planespanned by the influencing factors.

Reducing the linear model leads to the following graph-ic:

6 Single-factor Plots Overview: The single-factor plots overview shows the effect ofeach selected influencing factor and displays the predic-tion value and prediction interval to be expected in caseof the currently specified settings. The software fore-casts a percentage shrinkage of 0,85% for the adjust-ment values highlighted by red lines.

7 Analysis of Residuals

Residuals (lat. rest, singular residuum) are the differ-ences between the observed values (yi) and the values ofthe response estimated by the model . As a result,residuals are the deviations that are not explained by themodel.

The graphical analysis of residuals gives informationabout whether the model assumptions are fulfilled.

The upper graphic plots the residuals on the value num-ber. Ideally, the residuals show a random distributionwithout any specific structure.



The lower left-hand graphic displays residuals on aprobability plot. They are tested for normality. Theassumption that residuals are normally distributed is arequirement for some tests within the scope of regres-sion analysis.

The lower right-hand graphic compares the estimatedvalues (fitted values) to the residuals. The values shouldbe randomly distributed around the zero line from theleft to the right in this graphic.

8 Leverage, Cook:

These diagnostics help to check whether single values(datasets) have a supererogatory impact on the entireregression model.

The “Leverage“ graphic analyzes whether single sets ofvalues of the influencing factors can be interpreted asoutliers in x-direction.

The Cook’s distance graphic helps to recognize an influ-ential value that also affects the entire model (its ownestimate in the model) considerably.

9 VIF and Red%

The variance inflation factor (VIF) indicates the mutualdependence between the single influencing factors. TheVIF should not exceed 10. In case there are no interac-tions, the VIF amounts, not exceeds, to VIF = 1. If theVIF does not rise, the estimations of the coefficientsbecome more imprecise. This might affect the signifi-cance test of the coefficients.

Red% indicates the percentage reduction of the coeffi-cient of determination in case a specific influencing fac-tor is removed from the model (reduced model).

Red% is a variation complement corresponding to thecoefficients t-statistic analyzing the expectation of theresponse, i.e. how much it can be raised.

Compared to the t-statistic, Red% inspects the impact ofsingle influencing factors on the variation of theresponse.

23,13% means that the coefficient of determinationamounts to 77,68% - 23,13% = 54,55%without injection temperature.

Even after reducingthe variation, theinjection tempera-ture is still moreimportant than theholding pressure.

10 Optimization

Before optimizing the process, the average percentageshrinkage amounted to 0,886%.

After finding the optimal adjustment of parameters, thepercentage shrinkage is reduced to 0,825%. The controltest proved that the process has improved.

75PIQ International


Roles and Responsibilities in Six Sigma Projects

Six Sigma – regardless of whether “classical“ or lean - has become one of the most important strategies forprocess improvement. Green Belts and Black Belts bear the brunt of project work. Becoming a Green or BlackBelt requires challenging training. Moreover, in order to make the work of Green and Black Belts successful, theentire organization must be geared to the application of these methods, you need to assign the respective rolesand responsibilities to specific persons and they have to assume their roles in a qualified way.

The roles and responsibilities in implementing a Six Sigma project are explained in the following. They also applyto Lean Six Sigma and Design for Six Sigma.

Prof. Dr.-Ing. habil. Claus Morgenstern, TEQ® Training & Consulting GmbH

The Leader is in charge of defining visions, motivatingemployees to participate in Six Sigma projects, settingthe stage for success, implementing the requirementsdemanded for success, evaluating the results and effect-ing the required changes.

Compared to the Leader, the Champion is responsiblefor one or several Six Sigma projects. Champions areusually experienced executivesbeing well-acquainted with SixSigma and convinced that this initia-tive is going to be successful. Theyensure the success of this initiativeand the corresponding project tasksby checking progress of the projectat regular intervals. Moreover,Champions always communicatewith the Leader.

Master Black Belts are responsiblefor the training of Black and GreensBelts and offer them methodologicalsupport. They have profound knowl-edge of the required statistical meth-ods and techniques. All of themmust have management experiencein a company, organization, etc. Themain task of Black Belts is findingsolutions immediately and managingdevelopment projects. As team lead-ers of the Green Belts and projectteams they also manage their work.

Experts in the respective subject areaof the project and Green Belts areusually the members of the projectteams. Green Belts support BlackBelts in realizing projects or they areeven Six Sigma project managers atthe lower process level. They man-age the respective Six Sigma projectfrom the concept to its conclusion.During this time Black Belts orMaster Black Belts guide them.

Yellow Belts are employees having basic knowledge ofSix Sigma methods and applying them every day toimprove processes; however, not within the scope of SixSigma projects. They sometimes participate in Six Sigmaprojects but only as team members under the directionof Black or Green Belts.



Since the third edition of the AIAG Core Tool MSA 2002manual has been published in 2002, the subject of ndckeeps haunting the world. It almost seemed like this sta-tistic will continue to have a ghostlike, unremarkablepresence and might even disappear slowly but surely.But then came the fourth edition of this manual in 2010and the ndc gained in importance once again.

By now, people have had enough time to gather suffi-cient experience to apply this statistic; however, most ofthem have rather had a bad experience. This is the rea-son why we will have a closer look at whether the”number of distinct categories (ndc)“ makes sense ornot.

TV, PV and GRRFirst of all, it is important to clarify one of the mainissues of the AIAG MSA manual. How do total variation(TV), part variation (PV) and gauge repeatability andreproducibility (GRR) relate to one another? This ques-tion is actually about the fact that the observed processvariation is always a (quadratic) combination of the actu-al process variation and the gauge repeatability andreproducibility. In case of normally distributed values,this fact leads to

σ²observed process variation

= σ²actual process variation + σ²gauge repeatability and reproducibility

The AIAG MSA manual puts it as follows:

σ²Total = σ²Process + σ²MSA orTV² = PV² + GRR²

This relationship provides the essential basis for under-standing the AIAG MSA manual perfectly.

ndc according to AIAG MSAThe AIAG MSA manual defines the ndc as the number ofcategories of measured values that can be reliably dis-tinguished. To put it simply, you may count how manytimes the gauge repeatability and reproducibility GRRfits into the actual process variation.

You always truncate the ndc, unless it is less than 1. If itis less than 1, you have to round it up. The factor of 1,41(=√2) does not refer to the 97% confidence interval as

described in the MSA manual. It is calculated from thevariation ratios given in the ISO plot.

The AIAG MSA manual says the ndc should be greaterthan or equal to 5. The thought behind this specificationis that process control only makes sense in case you areable to divide the process into at least 5 distinct cate-gories of measured values based on the ndc.

The AIAG MSA manual lists the ndc under ”additionalwidth error metric“. However, the actual and main eval-uation of the measurement system is based on the GRRvalue. The manual demands a GRR value being less thanor equal to 10% of the reference value.

Together with the relationship between total variation,part variation and GRR you obtain the following threeequations.

These equations clearly illustrate that ndc and %GRRinterdepend. So you may convert ndc directly to %GRRand vice versa.

With the help of %GRR, you may find the ndc directly inthe diagram.

So the ndc amounts to 14 in case %GRR = 10% and thendc is 4 when %GRR = 30%. Unfortunately, the limitndc = 5 as found for %GRR = 27,2% does not matchthe limits for %GRR.

Dipl-Ing. Stephan Conrad, TEQ® Training & Consulting GmbH

The Crux of the ndc

% %GRRGRR

TV= ⋅100

ndcPV

GRR= ⋅2

% %GRRGRR

TV= ⋅100

TV PV GRR2 2 2= +

ndcGRR

= ⋅ −⎛⎝⎜

⎞⎠⎟

21

12%

77PIQ International


By using the tolerance as a reference, the following for-mulas apply to TV and PV according to AIAG MSAChapter III - Section B (p. 122).

The part variation PV calculated here, of course, doesnot concern the real process variation at all. It is nothingbut an auxiliary quantity in order to calculate the ndc.With the tolerance as a reference, the ndc is calculatedfrom the following formula.

Roughly speaking, • the %GRR specifies how many times the GRR fits

into the tolerance• the ndc specifies how many times the GRR fits

into the tolerance reduced by (the square of) GRR.

You notice that the information is redundant. The calcu-lation of two statistics is not required and, when con-sidering the different limits of %GRR and ndc, it is evencounterproductive.

Common Practice and the Meaning of ndc

In order to make sense of these methods, you generallyuse an approach different from the AIAG MSA manual.%GRR is related to the tolerance (TV=T/6) whereas ndcrefers to the actual part variation (PV) gained from themeasured parts. Now you realize quickly that evaluatingthe measurement system based on the ndc does notmake sense. The %GRR shows that a measurement sys-tem is applicable with respect to the corresponding tol-erance whereas the ndc only indicates that the partsused in the inspection can be divided into a minimumnumber of categories. The ndc thus depends directly on

the variation of the 10 parts and rather provides infor-mation about the applied parts than the measurementsystem.

You may also question the meaning of the ndc whenthinking of SPC as an issue usually dealing with qualitycontrol charts and sampling. The larger the sample size,the more precise the control chart. How is it possible todecide whether the measurement system is appropriatefor process control based on the ndc alone?

What Does the Author of the ndc Say?

The doubts about the benefits of the ndc continue toarise when reading the AIAG MSA manual more careful-ly. The third edition of the manual still referred to theorigin of the ndc in Chapter III - Section B, i.e. it men-tioned the book ”Evaluating the Measurement Process“(1984) by Donald J. Wheeler and Richard Lyday. Thispiece of information is missing completely in the fourthedition. So the question arises why the authors and thereference were deleted from the footer.

The background of this strange fact becomes clear whenreading Wheeler’s blog on Quality Digest from March2011. Please find the article online on the websitehttp://www.qualitydigest.com/inside/quality-insider-col-umn/problems-gauge-rr-studies.html. Wheeler makes itvery clear that the formula the AIAG MSA manual usesto calculate the ndc does not determine the ”number ofdistinct categories“ at all. He says “… nowhere in thattext did we ever suggest that this ratio would define thenumber of distinct categories.” But what is it this for-mula defines? Wheeler: “Unfortunately, as I has discov-ered after much effort, there is no simple interpretationfor the classification ratio in practice.” Or even moreclearly: „The number of distinct categories value (…)does not represent anything that can be expressed inpractical terms.” This value is of no relevance in prac-tice!

How shall we deal with the ndc given in the AIAG man-ual now? Wheeler’s answer: "So even though I may bethe author of this ratio, it is useless in practice. I per-sonally quit using it back in the 1980s. I suggest that youdo the same, starting immediately.”

TVUSL LSL T= − =

6 6

PV TV GRR T GRR= − = − ⋅( )2 2 16

2 26

ndcPV

GRR

T GRR

GRR= ⋅ = ⋅

− ⋅( )⋅

2 26

6

2 2

Q-DAS® GmbH & Co. KG

Eisleber Str. 2

D-69469 Weinheim

Tel.: +49 6201 3941-0

E-Mail: [email protected]

Internet: http://www.q-das.de

TEQ® Training & Consulting GmbH

Eisleber Str. 2

D-69469 Weinheim

Tel.: +49 6201 3941-15

E-Mail: [email protected]

Internet: http://www.teq.de


AustraliaElectronic Gauging SolutionsMelbourne, AustraliaMr. John Ciriacos E-Mail [email protected]

AustriaWanzel GmbH01220 Wien, Austria Mr. Horst Hickl E-Mail [email protected]

BrazilDAS Quality Ltda.Araçoiaba da Serra – SP, BrazilMr. Alberto K. Saiki E-Mail [email protected]

Denmark / SwedenMetrologic ApS2970 Hørsholm, DenmarkMr. Jørgen Meinertz E-Mail [email protected]

Great BritainMeasurement Solutions Ltd.Peterborough PE7 8GX, Great BritainMr. Iain Caville E-Mail [email protected]

HungaryT & T Quality Engineering KFT1038 Budapest, HungaryMr. Tibor Tóbiás, Mr. László Hanthy E-Mail [email protected]

JapanCKB CorporationTokyo 150-0002, JapanMr. Nagamine Kiyoyuki E-Mail [email protected]

KoreaKIMEAS.Co.,Ltd.Seoul, 152-848, KoreaMr. Mel Kim E-Mail [email protected]

MexicoIrapuato, Gto. CP 36670Mr. Hector PuenteE-Mail [email protected]

PolandOBERON 3D metrologia40-750 Katowice, PolandMr. Robert Seehafer E-Mail [email protected]

Notika System02202 Warszawa, PolandMr. Wojtas E-Mail [email protected]

PortugalDRILCO Portugal, Lda1885 Moscavide (Lisboa), PortugalMr. Inácio Marques E-Mail [email protected]

RomaniaT&T Quality Engineering RO S.R.L.525400 Targu Secuiesc, RomaniaMr. Zoltan Janosy E-Mail [email protected]

RussiaTechnopolice GroupRussia, 117218, Moscow,Mr. Dipl.-Ing. Alexander Loktev, MBAE-Mail [email protected]

SloweniaiNOVA MS, d.o.o.1290 GrosupljeMr. Tadej SirceljE-Mail [email protected]

SpainDRILCO s.a.San Sebastian de los Reyes, SpainMr. Alfonso Rueda E-Mail [email protected]

SwitzerlandBrigel AG8708 Männedorf, SwitzerlandMr. Urs GrütterE-Mail [email protected]

TurkeyMESTEK LTD STI16180 BURSAMr. Muammer Kiran E-Mail [email protected]

USADeSimone Quality InternationalSan Pedro, CA 90731, USAMr. Joe DeSimone E-MAIL [email protected]

Distributors:GERMANYQ-DAS® GmbH & Co. KG69469 WeinheimMr. Edgar Dietrich E-mail: [email protected]

CHINAQ-DAS® Software Technology(Shanghai) Co. Ltd.201203 Pudong, ShanghaiMr. David Sun E-mail: [email protected]

CZECH REPUBLICQ-DAS® spol. s.r.o.266 01 Beroun 2, Česká RepublikaMrs. L. Fuskova E-mail: [email protected]

FRANCEQ-DAS® France SARL95130 FranconvilleMr. Emmanuel Marie E-mail: [email protected]

INDIAQ-DAS Software Private LimitedPune - 411006Mr. Natarajan R. Iyer E-Mail: [email protected]

ITALYQ-DAS® s.r.l.25025 Manerbio (BS), ItalyMr. Giovanni Mabizanetti E-mail: [email protected]

KoreaQ-DAS Ltd.E-mail: [email protected] Korea

USAQ-DAS® IncorporatedMichigan 48309, USAMr. Tom Stewart E-mail: [email protected]

Q-DAS® Subsidiaries:

200 000 users 8 000 corporate clients 55 countries 21 languages

Documents