cualificacion metrologos

IAEA Regional Training Course under the Programme "Licensing Fuel and Fuel Modelling Codes" Bratislava (Slo-
vak Republic), 22-23 June 1999
QUALIFICATION OF TECHNOLOGICAL PROCESSES AND
TECHNOLOGICAL AND METROLOGICAL INSTRUMENTS
USED AT OAO "MSZ"
Lecturer: Sergey Balabanov SK00K0021
(Chief of bureau, Quality Department, OAO "MSZ", Electrostal)
Introduction.
OAO "MSZ" is one of the biggest European facilities manufacturing nuclear
fuel for nuclear power plants. The traditional selling markets of our plant are the
countries of the Eastern Europe and Finland. The quality system during the fuel as-
semblies manufacturing corresponding to the international standard ISO 9002 re-
quirements is valid at OAO "MSZ".
For the recent few years we have started to assimilate the markets of the
Western Europe which are new for us. In particular thanks to our cooperation with
the "Siemens" firm (Germany) the fabrication as well as the delivery of the fuel for
such customers as the "Obrigheim" NPP (Germany) and "Goesgen" NPP (Swit-
zerland) with the PWR type reactors have become possible for us.
Two elements of the OAO "MSZ" quality system which we are going to
consider during the lecture are the process management to assure quality during
manufacturing and check of the test equipment to assure the feasibility of the
measuring instruments. Specifically we will consider some aspects connected with
the technological equipment qualification and the measuring systems qualification
which are important for understanding the principles to be guided with when ar-
ranging such work at the plant. It is necessary to mark that the data on the techno-
logical equipment qualification as well as the measuring systems metrological
qualification can be used for the advertisement of the OAO "MSZ" products and
also as a quotation at the negotiations with the potential consumers of the products
or as a side letter at making deals (agreements, contracts).
PART 1
. METROLOGY AT THE QUALITY MANAGEMENT
1. LEGAL BASIS OF THE LEGAL METROLOGY
1.1. Act of the Russian Federation on the "Measurements uniformity
assurance".
The Act of the Russian Federation on the "Measurements uniformity assur-
ance" was adopted by the Supreme Soviet of RF in April 1993. The measurements
uniformity assurance in Russia before transition to the market economy was per-
formed by means of the centralized management. All the measuring instruments
were under the state surveillance. It determined a rather high level of the measure-
ments uniformity maintenance in the country but also some additional costs were
also required.
The adoption of the Act on the "Measurements uniformity assurance" was
caused by the necessity in revision of the legal, organizational and economical

bases of the metrological activity in accordance with the market economy transi-
tion conditions.
The mainpurpose of the Act is theprotection of the rights and legal interests
of the citizens, established law order and the RF economyfrom the negative con-
sequences of the measurements unauthentic results. Such a wide trend of the Act —
the economy protection — is notpeculiar to the legislation of theforeign countries.
However nowadays in the conditions of maintaining the significant share of
the Russian economy national sector the full rejection of the state regulation of the
metrology issues in the industrial sphere would have been premature. The meas-
urements uniformity assurance has always been and is the most important national
function. The analysis of the experience of foreign countries indicates that the
measurements uniformity assurance in them is covered by the sphere of the na-
tional management.
In accordance with the Act on the "Measurements uniformity assurance" the
national management of the activity related to the measurements uniformity assur-
ance in RF is carried out by the Committee of the Russian Federation on standardi-
zation, metrology and qualification (Gosstandart of Russia).
In compliance with the Act the metrological service is being established to
perform the work on the measurements uniformity assurance and for carrying out
the metrological inspection and surveillance. The National metrological service,
the metrological services of the national management bodies and the metrological
services of the legal persons are distinguished.
The National metrological service includes the scientific investigation me-
trological institutes, liable for creation, keeping and application of the national ref-
erences, and the national metrological service bodies (former territorial authorities
of the Gosstandart) performing the national metrological inspection and the sur-
veillance for the assigned territory.
To implement the Act on the "Measurement uniformity assurance" into
force the development and the adoption of the complex of the standard acts and
the Gosstandard documents were required. For instance, by the adoption of decree
N 100 of the RF Government dated 12.02.94 the following documents were ap-
proved: "Statement on the national scientific metrological centers", "Order of ap-
proving the statements on the metrological services of the federal bodies of the
executive power and legal persons", "Order of accreditation of metrological serv-
ices of the legal persons as to the measuring instruments verification rights",
"Statement on the Russian Federation defence metrological assurance" and other
normative documents.
1.2. Analysis of the legislation in the field of metrology.
By the Act on the "Measurements uniformity assurance" the following types
of the national metrological inspection are established:
• Measuring instruments type approval;
• Verification of the measuring instruments including the references;
• Licensing of the legal and physical persons as to the right of the
measuring instruments manufacturing, repair, selling and renting.

At this point the national metrological inspection and supervision are per-
formed only in the spheres stipulated by the Act.
The international metrological surveillance covers the following spheres:
health; veterinary; environment protection; labor safety assurance; commercial ac-
tivities and payments between the seller and the buyer; international accounting
procedures; national defence assurance; geophysical and hydro-meteorological
work; fabrication of products delivered for the national needs as per the contracts;
tests and inspection of the product quality to determine the compliance with the
obligatory requirements of the international standards; obligatory certification of
the products and services as well as other fields stipulated by article 13 of the RF
Act on the "Measurements uniformity assurance"
That means, that all measuring instruments can be divided into two groups
as far as a scope of their application is concerned.
All the measuring instruments designated for application in the international
metrological inspection and surveillance spheres belong to the first group. Such
measuring instruments are subject to the tests and the type approval as well as to
the primary and periodical verification.
The second group - the measuring instruments which are not designated for
the application and are not used in the spheres covered by the national metrological
inspection. The state (Gosstandart of Russia) doesn't conduct the surveillance of
these spheres.
The owners of the second group of the measuring instruments establish the
way of maintenance of the measuring instruments in the operating condition them-
selves (by means of their calibration).
(For information) At OAO "MSZ" all measuring instruments belong to the
first group, i.e. they subject to obligatory verifications.
Thus the difference between the old normative base and a new law actually
consists in the difference of the application scopes of the national metrological in-
spection and surveillance.
There is one more essential difference. The old normative base was sup-
ported by the Government Decrees and normative documents Of Gosstandart. In-
troduction of the new Law permitted transition of the metrological activities on the
juridical base at the Federation level.
1.3. Licensing of the activity related to the measuring instruments manu-
facturing, repair, selling and renting.
In accordance with the Act the activity on manufacturing, repair, selling and
renting of the measuring instruments covered by the national metrological surveil-
lance sphere is subject to licensing by the national metrological service [1].
Licensing is one of the types of the national metrological inspection together
with the approval of the measuring instruments type and measuring instruments
verification. The order of licensing is determined by metrology rules 50.2.005-94
'TCH. The order of licensing of the activity on the measuring instruments manu-
facturing, repair, selling and renting" approved by the Statement of Gosstandart of

Russia dated 08.02.1994 N 8 and registered under N 741 09.12.1994 in the Minis-
try of Justice of Russia.
License - authorization given by the national metrological service body
to the legal or the physical person (licensee) for the performance of the activ-
ity on the measuring instruments manufacturing, repair, selling and renting
on the territory assigned to it.
The license is valid on the whole territory of the Russian Federation.
The persons intending to obtain a license for the measuring instruments
fabrication should have a certificate of approval of a measuring intrument
type.
1.4. Tests and approval of type of measuring instruments.
Tests and approval of measuring instrument type relate only to the
measuring instruments used in the field of the national metrological control
and surveillance.
Normative base to perform the above activities includes the following
documents:
IIP 50.2.009-94 «rCH. Tests and approval of measuring
instrument type»;
IIP 50.2.010-94 «rCH. Requirements for the national test
centers for measuring instruments and
their accreditation procedure));
IIP 50.2.011-94 «rCH. Maintenance of the National Register
of measuring instruments)).
Document EDP 50.2.009-94 was developed to be used in the Russian
Federation instead of GOST 8.001, GOST 8.383, GOST 8.326 and specifies
general requirements to an organization and sequence of work to be performed
in the framework of Tests and approval system for a measuring instrument
type.
The main difference of this document from the basic documents of the
former System of the State tests of measuring instruments (GOST 8.001,
GOST 8.383) is a fact that it covers only the measuring instruments to be
used in the field of the national metrological control and surveillance
independent of a manufacture volume, i.e. it covers a piece-work (isolated)
production.
The System of tests and approval of a measuring instrument type
includes:
• Testing of measuring instruments for the purpose of approval of a
particular type;
• Decision making on a type approval, national registration of the type and
issuing a certificate of approval of the type;
• Testing of measuring instruments for the conformity to the type approved.
• Recognition of the approval of a type or test results of measuring instru-
ments type, performed by the Competent Authorities of the foreign coun-
tries;
4

• Information service of consumers of measuring equipment, inspection and
surveillance bodies and government bodies.
The measuring instruments for which the certificates of type approval have
been issued are subject to the state registration in the National Register of measur-
ing instruments.
GOST 8.326 "Metrological qualification of measuring instruments" was
cancelled in September 1997. Instead of it revision Nl of IIP 50.2.009-94 was in-
troduced. This revision establishes a procedure for testing single copies of meas-
uring instruments as well as measuring complexes. The right to test single stan-
dards is only granted to the national test centers for measuring instruments.
7.5. Verification of measurement devices.
Verification of measurement devices is a set of operations performed by the
Bodies of National Metrological Services or other accredited organizations with
the purpose to determine and confirm the compliance of the measurement devices
with the established technical requirements.
According to the Law of the RF "Assurance of measurements uniformity"
the measurement devices which are subject to National metrological inspection and
surveillance are to be verified after fabrication and repair, when imported and in
process of operation.
To develop the Law Gosstandart of Russia approved a number of documents
regulating different aspects of verification activities:
IIP 50.2.006-94 " Qualification of measurement devices. Arrangement and
performance";
EDP50.2.012-94 " Qualification of verifiers of measurement devices";
IIP 50.2.007-94 " Verification stamps".
In conformity with nP 50.2.006-96 measurement devices are subjected to ini-
tial, periodical, extraordinary, inspection and expert verification. Measurement de-
vices of the approved types are subject to initial verification after fabrication, re-
pair and when imported. The initial verification can be deleted for those measure-
ment devices during import on the basis of the agreements concluded by Gosstan-
dart of Russia about the acknowledgement of the verification results performed in
foreign countries.
Periodical verification is arranged for the measurement devices in operation
or in storage in specified intervals between verifications. The owners of measure-
ment devices are responsible for the lists of measurement devices intended for
verification. The results of periodical verification are valid during the intervals
between verifications. The first verification interval is specified when the type is
approved. Correction of intervals between verifications is carried out by the Bodies
of National Metrological service with the approval of the legal person's me-
trological service.
When correcting the intervals between verifications it is advisable to use as a
methodological document MH 1872-88 'TCH.Verification frequency of standard
measurement devices. Determination and correction procedure", and MH 2187-92
5

'TCH. Measurement devices calibration and verification frequency. Procedure for
determination".
1.6. Qualification of procedures toperform measurements.
On the first of July 1997 new national standard GOST P8.563-96 'TCH.
Procedures to perform measurements" was introduced specifying the requirements
to their design, qualification, standardization and metrological surveillance.
The procedures to perform measurements constitute a set of operations and
rules, their fulfillment providing for results with a definite error.
Obtaining measurement results with a definite error or with an error not ex-
ceeding the limits allowed (measurement precision norms) is one of the most im-
portant conditions assuring unity of measurements.
Thus, such procedures meeting up-to-date requirements are of major impor-
tance to assure uniformity of measurements.
These procedures qualification is the process to establish and confirm the
compliance with the metrological requirements specified.
Qualification is obligatory for all procedures used in spheres covered by na-
tional metrological inspection and surveillance as well as for the inspection of
status of complex technical systems which GOST P22.2.04-94 extends to.
In case measurement procedures are used in the fields not covered by na-
tional metrological control and surveillance, they are qualified according to the or-
der established at the department or at the enterprise.
If the metrological service qualifies measurement procedure used at other
enterprises, this metrological service should be accredited for the right to qualify
measurement procedures. Metrological services of enterprises and organizations
are accredited in conformity with the Rules relating to metrology FTP 50.2.013-94
FCH. Accrediting of metrological services of legal persons".
Measurement procedure qualification is carried out by means of metrologi-
cal expertise of documents, theoretical and experimental investigations of meas-
urement procedures. The choice of a qualification method is defined by the meas-
urement procedure complication and by the experience gained when qualifying
similar measurement procedures.
During the measurement procedure document expertise it is reasonable to
study the object of measurements with the objective to assess whether the
measurement procedure mission and the value measured correspond to the task of
the object control. The expertise of the measurement procedure document includes
the assessment of comprehension and precision of requirements to conditions of
measurements. During the document expertise for the measurement procedure
which will be used in the spheres covered by national metrological control and
surveillance, it is necessary to check the approval of measurement devices' types.
As to measurement devices the comprehension and justification of require-
ments to metrological characteristics are studied. In case the calculations or results
of experimental evaluation of measurement errors are introduced, these materials
are studied with the purpose to consider all the factors affecting measurement error
and to determine the methods correctness.

The measurement procedures applied in the areas of national metrological
control and surveillance are to be under national metrological control.
The metrological surveillance over the qualified measurement procedures is
carried out by the metrological services of legal persons.
During the performance of the national metrological surveillance by the
bodies of the National metrological service or of the metrological surveillance car-
ried out by the metrological services of legal persons the following parameters are
checked:
- availability of the document regulating measurement procedures and qualifi-
cation certificate;
- compliance with the requirements in the document for the measurement pro-
cedure of applied measurement devices and other technical means, measure-
ment conditions, preparation and fulfillment of measurements, processing and
drawing up the results of measurements;
- fulfillment of requirements to the error control procedure of measurement re-
sults for measurement procedure, if such procedure has been regulated;
- operators' professional skill to perform measurements in conformity with the
measurement procedure.
1.7. National metrological surveillance.
We should keep in mind as in any other sphere the effectiveness of the Law
is defined by the effectiveness of surveillance over its execution. It is exactly the
function of surveillance over the Law execution which is the duty of the National
metrological surveillance, and the achievement of the basic aim of the law, that is
protection of interests of citizens and of the state at large against negative conse-
quencies caused by wrong measurement results.
National metrological service is responsible for the performance of the na-
tional metrological surveillance applying three normative documents. One of these
documents is IIP 50.2.002-94 FCH. National metrological surveillance over fabri-
cation, measurement devices status and usage, qualified procedures to fulfil meas-
urements, references and observance of metrological norms and rules".
This document is the most traditional one, it does not require significant recon-
struction in the work of both state inspectors and of enterprises themselves.
Basic changes in this activity are brought to the following: inspection over
the status and usage of measurement devices extends only to those measurement
devices which relate to the sphere of national metrological control and inspection.
Due to that the first priority task of each enterprise is to prepare a list of measure-
ment devices relating to this qualification group, in other words, subjected to veri-
fication. The list is prepared according to MH 2273-93 'TCH. Fields of measure-
ment devices usage subjected to verification".
The procedures dealing with measuring devices included into this list are the
objects of national metrological inspection over the qualified procedures of meas-
urement performance.

1.8. Calibration of measurement devices.
The new term "calibration of measurement devices" was introduced by the
Law of the Russian Federation "Assurance of measurements uniformity" (further
the Law) and the old terms "department verification" and "metrological qualifica-
tion" became out of use.
According to the Law calibration of measurement devices is a set of op-
erations carried out with purpose to determine and confirm actual values of me-
trological characteristics and/or usability of measuring devices for use not subject
to national metrological control and inspection.
Difference between calibration and verification lies in:
-field of application
Only those devices are calibrated which are not subjected to national me-
trological control and inspection, that is verification;
-verification is mainly carried out by the bodies of national measurement
service, and calibration is done by any other metrological service or physical per-
son provided there are conditions to fulfil such work there[l].
In case verification is an obligatory operation inspected by the bodies of the
National metrological service, calibration is a voluntary function performed either
by a metrological service of a company or under request by any other organization
capable to do the work.
However a voluntary calibration does not relieve the metrological service of
a company from a necessity to observe specific requirements, in particular: trace-
ability, i.e. an obligatory "connection" of a working measuring instrument with the
national reference.
The Russian calibration system is based on the following principles:
• voluntary introduction;
• obligatory transfer of unit dimensions from the national references to the
working measuring instruments;
• technical competence;
• self-repayment.
The introduction of the metrological service to the system by the way of
accreditation is done on a voluntary basis. A major motivation of introduction is a
reinforcement of a consumer's trust to the product quality levels inspected by
measurements that increases product competitiveness. Besides, a product
certification process which is currently developed in the country according to the
international standards ISO-9000 and European Standard EN 45000 brings
forward a mandatory requirement to accredit test and calibration laboratories that is
a condition of acknowledgment of the quality level achieved.
An obligatory transfer of unit dimensions from the national references by
means of working references available in the National metrological service to the
company standards and further to the working measuring instruments is the main
condition as it was mentioned before of assurance of the true measurement results.
System self-repayment is stipulated by the requirement of the market econ-
omy.

The subjects of the Russian calibration system are:
• metrological services accredited to calibrate measuring instruments;
• national scientific metrology centers and bodies of the National me-
trological service registered as accreditation organizations having the right
to accredit metrological services of legal persons for calibrations of meas-
urement devices;
• Gosstandart of Russia being the central body of the system that coordi-
nates the activities of all other subjects;
• VNII (All-Russian Research and Development Institute) of metrological
service responsible for the organizational, methodical and informational
support of the system activities;
• Council of the System.
1.9. Certification of measurementdevices.
The problems of keeping measurements uniformity in the fields that are not
subject to the national metrological control and surveillance, justified the necessity
of establishing a voluntary system to certify measurement devices for the confor-
mity to the metrological norms and rules.
The methodological basis of the certification system for the measurement
devices is normative documents of the International Organization for Standardiza-
tion (ISO), International Electrotechnical Commission (IEC), International Con-
ference on the Accreditation of Test Laboratories, GOST R certification system
and certificate system of the International Organization for Legislative Metrology.
Measurement devices certification provides for:
• voluntary certification of the devices for the conformity to the me-
trological norms and rules valid for any type of measurements;
• development and introduction of the normative documents specifying
metrological norms and rules for the measurement devices;
• development and implementation of the standard test programs for the
purpose of measurement devices certification;
• testing and approval of the procedures for measurement device calibra-
tions during a certification process and preparation of the proposals to
determine calibration frequencies;
• qualification of the procedures to perform measurements using certified
measurement devices;
• implementation of the extended network of certification bodies for the
measurement devices that are accredited for various measurements and
test laboratories for particular groups of the measurement devices;
• cooperation with the national metrological agencies of other countries
with regard to the mutual recognition of the accredited bodies, laborato-
ries, conformity certificates, compliance marks as well as the results of
measurement devices certification.
The certification system ensures confidentiality of the information being a
commercial secret.

Measurement devices are certified according to the ISO classification pat-
terns: III, IV and V. If a particular measurement device has any specific features
other patterns may be used.
2. MAIN SYSTEM PROVISIONS FOR ASSURANCE OF MEAS-
UREMENTS UNIFORMITY.
2.1. Measurement quality.
There is a direct connection between the quality of measurements and that of
products. When the quality of measurements does not meet the requirements of the
technological process, the high quality level of products can not be expected.
A problem of assurance of the high quality products is to a great extent the
problem of measuring quality parameters of the materials and components, of pro-
viding given process modes, i.e. measuring the parameters of the technological
processes and these measurements are used to control the process.
Measurement quality - a totality of measurement status properties provid-
ing measurement results with the required accuracy, in the required form and dates.
The main properties of the measurement status cover:
• measurement accuracy;
• precision of measurements;
• measurement reproducibility;
• fast rate of obtaining results;
• measurement uniformity.
The above definitions correspond to MH 2247-93 'TCH. Metrology. Basic
terms and definitions".
While solving the task related to the measurement quality assurance the most
important part belongs to metrology - science of measurements, techniques and
means to assure measurements unity and required accuracy. The solution is
achieved by creation of the national references, their connections to the measure-
ments performed and by establishing various rules and norms for measurements
and measurement devices. If the measurement unity is not observed even the finest
measurements performed with the help of the adequately selected measurement
devices will not give the required results.
Measurements uniformity - measurement status characterized by the re-
sults expressed in terms of legalized units which dimensions (in the prescribed
limits) equal to the unit dimensions reproduced by the primary references and
measurement errors are known and do not deviate from the prescribed limits with a
given probability.
2.2. Choice of measurement instruments.
Measurement quality depends on the proper choice of the measurement de-
vices. While choosing the measurement devices the following factors should be
taken into consideration:
• physical value to be measured;
• measuring method;
10

• range and error of the measurement devices;
• measurement conditions;
• allowable measurement error;
• measurement device price;
• easy operation;
• service life;
• losses due to measurement errors (rejection I and II).
The solution of the task is more difficult because there is no single factor
based on which measurement devices could be compared. Thus the choice of the
measurement devices depends on a measuring task being solved, when preference
is given to some factors while the others are neglected.
Measurement errors are basic features of the measurement devices. They
substantially affect the quality of measurements, that is why they are given a prior-
ity when the choice of the measurement devices is considered.
There are three main principles of choosing measurement devices:
1) economical approach (the most optimal because it takes into account
almost all the factors). It is necessary to bear in mind the following:
• higher measurement accuracy allows a better control over the techno-
logical process;
• more accurate measurements allow to reduce an item's tolerance;
• higher measurement accuracy (that means lower measurement errors)
leads to reducing unfound and false rejects.
Measurement error growth results in the growth of losses, while the
costs of measurements go down. As a rule, one of these dependences is not
linear, thus their sum, i.e. total production costs depending on measurement accu-
racy has an extremum.
Cost-effective measurement accuracy of a process parameter corresponds to
the minimum sum of losses due to the measurement error and costs including the
costs of metrological maintenance of measurement devices. The optimum meas-
urement accuracy corresponds to the standard deviation 50trr.
2) probability approach is choice of measurement device accuracy ac-
cording to a given tolerance of the parameter inspected and given values of the in-
spection rejects I and II (unfound and false rejects).
If measurements were performed by absolutely accurate measurement de-
vices all the items being within the tolerance zone would be accepted and the items
with a measured parameter exceeding a tolerance would be rejected. Due to a
measurement error a part of the defectives being measured will be accepted (in-
spection reject II) and a part of acceptable items will be rejected (inspection reject
I). The inspection rejects are influenced by the dispersion of actual values of a pa-
rameter measured, its prescribed tolerance, distribution law for measurement errors
and dispersion of actual value of the parameter measured. Let's assume that the
plots have been constructed of an inspection reject probability as functions of tech-
nological dispersion of the parameter measured, of a measurement error, of a
measured parameter tolerance. With the help of these plots when there are pre-
n

scribed values of the inspection reject probability, standard deviation of dispersion
of the actual values of the parameter measured and its tolerance it is possible to as-
sess measurement error boundaries and required accuracy of measurement devices
[2]-
3) directive approach allows to determine a relationship between a meas-
ured parameter tolerance and measurement error limit.
3. SOME ASPECTS OF METROLOGICAL QUALIFICATION OF
MEASURING SYSTEMS AT OAO "MSZ".
The development of nuclear power, higher requirements for nuclear fuel
quality and efficiency of the fuel usage at NPPs resulted in a radical change of the
measurement requirements. One of the major aspects of these requirements is a
possible provision of reliable enough estimate of a measurement error.
An incorrect estimate of the measurement error is fraught with large eco-
nomic losses- An understated measurement error estimate leads to the increase of
product rejections, non-economical consumption of material resources, erroneous
solutions made during developments and testing of specimens of new products. An
overstated estimate of the measurement error leads to the erroneous conclusion that
more accurate measurement devices are required and causes extra costs for the de-
velopment, manufacture and operation of measurement devices.
A measurement error is conditioned in general by a number of factors. It de-
pends on the properties of measurement devices used, methods of their usage
(measurement procedures), adequate calibration and verification of the measure-
ment devices, conditions under which the measurements are performed, an error
caused by an operator and some other factors.
That is why during preparation for measurements, design of various proc-
esses (for example, technological processes, quality control processes) where
measurements are used, it is important to pay attention both to a choice of meas-
urement devices and random error associated with them and some other factors af-
fecting a measurement error.
At OAO "MSZ" during product quality control measuring systems are
widely used together with actual measures, measuring apparatus and equipment.
According to the standard [3] a measuring system is a set of measuring instruments
and auxiliary devices interconnected by communication channels; the system is de-
signed to generate measurement information signals in the form suitable for auto-
matic processing, transmission and (or) use in the automated control systems.
Metrological qualification means an examination of measuring system for
the purpose of determining its metrological characteristics and conformity of the
measuring system and measuring channel inputs to the specified technical re-
quirements.
The following tasks should be listed for the metrological qualification of the
measuring systems:
1) definitions of the nomenclature of metrological characteristic estimates
and their assessment;
12

2) determination of the conformity of the metrological characteristcs to the
specified technical requirements;
3) determination of the nomenclature of metrological characteristics, subject
to periodical inspection.
The main steps of the merrological qualification of measuring
systems at OAO "MSZ" are:
1) review of technical documentation;
2) determination of the inspection scope for the measurement channels in-
corporated into the measuring system;
3) determination of the quantity of points to be examined on the basis of
measurement range, of the approximation method of measurement results.
4) determination of the quantity of observation in each control point;
5) generation of initial data and conditions to determine an error of the
measurement channels;
6) analytical representation of the measuring system errors;
7) estimate of the measuring system errors under real operating conditions;
8) review of metrological provision of the measuring system based on the
qualification results.
When technical operating documentation is reviewed it is assessed from
the point of view of its suitability for a user. Informational sufficiency of the
operating documentation means a possibility to get familiarized with an operation
procedure of equipment and its maintenance. The main steps of metrological
qualification of measuring systems are specified in document MH 2002-89
"Organization and performance of metrological qualification of measuring sys-
tems".
At OAO "MSZ" special attention is paid to the quality of measurements
to assure trustworthiness of the measurement results obtained. After the Russian
Federation Decree N4871-1 "Assurance of measurements uniformity" dated
April,27,1993 was passed our company by the order of the Russian Gostandart
received an accreditation to perform verifications of the measuring instruments,
qualifications of measurement procedures and standards designed to determine
true value of a given characteristic during calibration of a particular measuring
system.
The operation of any measuring system can be assessed by a high or low
quality levels of the measurements made. If the measurement results are close to a
true value of the characteristic measured the quality of data is considered high. If
a few or all the results are far from the true value the quality of data is consid-
ered low. Statistical characteristics the most widely used to determine data qual-
ity are a shift and variance. A characteristic named "shift" describes a shift of re-
sults in relation to the true value, while "variance" is a characteristic that describes
a spread in results.
One of the typical reasons of poor quality of data is a large variability of
data. If a data quality level is not acceptable it must be improved. To achieve it is
not data that should be improved but a measuring system. In cases a meas-
uring system can not be enhanced to prevent from a shift of measurement re-
13

suits with regard to true value, i.e. from a systematic error, there are used ra-
tioning methods of measurement systematic, random error, and in some
cases of an additional measurement error caused by so called "effect" function.
By an effect function we assume a dependence of a measurement error of a
particular measuring instrument on the deviations of affecting physical values
from their rated values which affect a measurement result, though are not
measured by the measuring instrument.
14

4. DEVELOPMENT OF PROCEDURES FOR QUALIFICATIONS
OF MEASURING SYSTEMS.
4.1. General
The first step in the preparation for the metrological qualification of a
measuring system is to confirm that a proper value is measured. If the latter
is improper there is no sense to discuss accuracy and thoroughness of the
measuring system because it wastes its resources in vain. At OAO "MSZ" a pro-
cedure [4] has been developed to use the company's standards (COI1) for
graduation and calibration of measuring instruments, for qualification and
check of accuracy of measurement procedures and some other purposes. Stan-
dards (COribi) are a measure for performing measurement. They can be refer-
ence and working. Reference standards are used for verifications of measuring
instruments, for qualification of a measurement procedure. Working standards
are used for checking the adjustment of equipment. Each standard is to be subject
to obligatory qualification at the plant when quantitative metrological character-
istics of a standard are determined, then a certificate for the standard is issued.
The next step in preparation for the metrological qualification is to de-
termine what characteristics a measuring system should have to be accept-
able (i.e. it is necessary to prepare a metrological qualification program). For
this purpose it is important to know how the data will be used otherwise the
proper statistical characteristics can not be determined. At the plant there is a
procedure to carry out metrological expertise of the technical documentation [5],
according to this procedure the documentation for a measurement performance
procedure while using a particular measuring instrument is assessed.
After statistical characteristics have been determined a measuring system
should be qualified for availability of the above properties. In metrological terms
such statistical characteristics are often named metrological characteristics of a
measuring instrument. At the plant there is a special procedure to perform me-
trological qualifications during development and implementation in the pro-
duction of non-standardized control and measurement devices [6] and infor-
mation-and-measuring systems [7].
While determining characteristics to be rationed that reflect a basic error of
each individual measuring instrument of a particular type it is necessary to choose
a mathematical model. The model parameters should represent the rationed me-
trological characteristics reflecting, on the one hand, properties, on the other
hand - requirements for a basic error of a measuring instrument. At the plant
there is an arrangement to perform metrological qualifications of measurement
procedures [S].
Further we will consider a case when a operator is included into the meas-
uring system. In this case during metrological qualification of a measurement
procedure a check of precision and reproducibility of measurement results pro-
duced by the apparatus may be prescribed.
Precision is a variability of measurement results produces by one measur-
ing apparatus used several times by the same operator for measurements of an
identical characteristic of one part.
15

IAEA Regional Training Course under the Programme "Licensing Fuel and Fiiei Modelling Codes" Bratislava (Slo-
Reproducibility is a variability of mean value of measurements made by
different operators using the same apparatus for measuring identical character-
istic of one part.
4.2. Analysis of variance under ANO VA method is one of a
possible procedures for investigation of a measuring apparatus.
The investigation of a measuring apparatus with determination of its me-
trological characteristics is carried out according to the metrological qualification
program. The above mentioned investigation can be performed by several differ-
ent procedures that fill the content of the metrological qualification program. In
this lecture I would like to dwell on the method of analysis of variance (ANOVA).
Analysis of variance (ANOVA) is a standard statistical procedure used to
analyze measurement errors and other causes (reasons) of data variability while
studying measuring systems. For analysis purpose variability can be expanded into
four categories: parts, operators, interaction between parts and operators and an er-
ror of repeated measurements caused by equipment. Due to the fact, that some of
you might be unfamiliar with this method I permit myself to dwell on a few simple
examples illustrating the basic ideas of ANOVA method.
4.2.1.Basis ideasof analysisof variance(ANOVA).
In general, the purpose of analysis of variance (ANOVA) is to test for sig-
nificant differences between means. Why the name analysis of variance? It may
seem odd to you that a procedure that compares means is called analysis of vari-
ance. However, this name is derived from the fact that in order to test for statistical
significance between means, we are actually comparing (i.e., analyzing) variances.
At the heart of ANOVA is the fact that variances can be divided up, that is,
partitioned. Remember that the variance is computed as the sums of squares of de-
viations from the overall mean, divided by n-1 (sample size minus one). Thus,
given a certain n, the variance is a function of the sums of (deviation) squares, or
SS for short. Partitioning of variance works as follows. Consider the following data
set (example 1):
EXAMPLE 1
Observation 1
Observation 2
Observation 3
Mean
Sum of Squares (SS)
Overall mean
Total Sums of Squares
Group 1
2
3
1
2
2
4
28
Group 2
6
7
5
6
2
From Example 1 it is seen that the means for the two groups are quite differ-
ent (2 and 6, respectively). The sums of squares within each group are equal to 2.
16

Adding them together, we get 4. If we now repeat these computations, ignoring
group membership, that is, if we compute the total SS based on the overall mean,
we get the number 28. In other words, computing the variance (sums of squares)
based on the within-group variability yields a much smaller estimate of variance
than computing it based on the total variability (the overall mean). The reason for
this in the above example is of course that there is a large difference between
means, and it is this, difference that accounts for the difference in the SS. In fact, if
we were to perform an ANOVA on the above data, we would get the following re-
sult (see Table 1):
Table 1
Statistical data
Effect SSb
Error SSW
Main Effect
SS
24,0
4,0
df
1
4
MS
24,0
1,0
F
24,0
As you can see, in the above table the total SS (28) was partitioned into the
SS due to within-group variability (2+2=4; see the second row of the Scrollsheet)
and variability due to differences between means (28-(2+2)=24; see the first row of
the Scrollsheet).
The within-group variability (SS,), is usually referred to as Error variance.
This term denotes the fact that we cannot readily explain or account for it in the
current design. However, the SSAEffect we can explain. Namely, it is due to the dif-
ferences in means between the groups. Put another way, membership explains this
variability because we know that it is due to the differences in means.
In Table there is given a computation of F-criterion (F=24,0), one of the sta-
tistical criteria used for an estimate of significance of mean differences. F-criterion
represents a ratio of explained to unexplained variability. F-criterion indicates
whether the ratio of the two variance estimates is significantly greater than 1 and
allows to test the null hypothesis that there are no mean differences between-
groups in the population. In our example above, this criterion is highly significant,
and we would in fact conclude that the means for the two groups are significantly
different from each other. To summarize the discussion up to this point, the pur-
pose of analysis of variance is to test differences in means (for groups or variables)
for statistical significance. This is accomplished by analyzing the variance, that is,
by partitioning the total variance into the component that is due to true random er-
ror (i.e., within-group SS) and the components that are due to differences between
means. These latter variance components are then tested for statistical signifi-
cance, and, if significant, we reject the null hypothesis that the means (in the
population) are mfferent from each other.
The variables that are measured (e.g., a test score) are called dependent vari-
ables. The variables that are used to divide observations into groups that are com-
pared are called factors or independent variables.
17

4.2.2. Multi-Factor ANOVA.
In the simple example above, it may have occurred to you that we could
have simply computed t-criterion for independent samples [9] to arrive at the same
conclusion. And, indeed, we would get the identical result if we were to compare
the two groups using this criterion. However, ANOVA is a much more flexible and
powerful technique that can be applied to much more complex research issues.
The world is complex and multivariate in nature, and instances when a sin-
gle variable completely explains a phenomenon are rare. Thus in a typical experi-
ment many factors are taken into account. One important reason for using ANOVA
methods rather than multiple two-group studies analyzed via t-criterion is that the
former method is more efficient, and with fewer observations we can gain more
information. Let us expand on this statement.
Suppose that in the above two-group example we introduce another group-
ing factor, for example, Gender. Imagine that in each group we have 3 males and 3
females. We could summarize this design in a 2 by 2 table (see ex.2):
EXAMPLE 2 Experimental Experimental
group 1 group 2
Males 2 6
3 7
1 5
Mean
Females 4 8
5 9
3 7
Mean
Before performing any computations, it appears that we can partition the to-
tal variance into at least 3 sources: (1) within-group variability (error), (2) variabil-
ity due to experimental group membership, and (3) variability due to gender.
(Note that there is an additional source-interaction - that we will discuss shortly).
What would have happened had we not included gender as a factor in the study but
rather computed a simple t-criterion? If you compute the SS ignoring the gender
factor (use the within-group means ignoring the gender; the result is
SS=10+10=20), you will see that the resulting within-group SS is larger than it is
when we include gender (use the within-group, within-gender means to compute
those SS; they will be equal to 2 in each group; thus the combined SS-within is
equal to 2+2+2+2=8). This difference is due to the fact that the means for males
are systematically lower than those for females, and this difference in means adds
variability if we ignore this factor. Controlling for error variances increases the
sensitivity (power) of a criterion. This example demonstrates another principal of
ANOVA that makes it preferable over simple two-group t-criterion studies: in
ANOVA we can test each factor while controlling for all others; this is actually the
18

reason why ANOVA is more statistically powerful (i.e., we need fewer observa-
tions to find a significant effect) than the simple t-criterion.
4.2.3. Interaction effects.
There is another advantage of ANOVA over simple t-criterion: ANOVA al-
lows us to detect interaction effects between variables, and therefore, to test more
complex hypotheses about reality. Let us consider another example to illustrate this
point.
Imagine that we have a sample of highly achievement-oriented students and
another of achievement "avoiders". We now create two random halves in each
sample and give one half of each sample a challenging test, the other an easy test.
We measure how hard the students work on the test. The means of this (fictitious)
study are as follows (see ex.3):
EXAMPLE 3 Achievement-Oriented Achievement
Students Avoiders
Challenging test 10 5
Easy test 5 10
How can we summarize these results? Is it appropriate to conclude that (1)
challenging tests make students work harder, (2) achievement-oriented students
work harder than achievement-avoiders? None of these statements captures the es-
sence of this clearly systematic pattern of means. The appropriate way to summa-
rize the result would be to say that challenging tests make only achievement-
oriented students work harder, while easy tests make only achievement-avoiders
work harder.
Example 3 — an example of a two-way interaction between achievement ori-
entation and test difficulty. While the previous two-way interaction can be put into
words relatively easily, higher order interactions are increasingly difficult to ver-
balize. Imagine that we had included factor Gender in the achievement study
above, and we had obtained the following pattern of means (see ex.4):
EXAMPLE 4
Females
Challenging test
Easy test
Males
Challenging test
Easy test
Achievement-Oriented
Students
10
5
Achievement-Oriented
Students
1
6
Achievement
Avoiders
5
10
Achievement
Avoiders
6
1
19

The pattern shown in the table above represents a three-way interaction
between factors. Thus we may summarize this pattern by saying that for females
there is a two-way interaction between achievement-orientation type and test diffi-
culty, i.e. achievement-oriented females work harder on challenging tests than on
easy tests, achievement-avoiding females work harder on easy tests than on diffi-
cult tests. For males, this interaction is reversed
A general way to express all interactions is to say that an effect is modified
(qualified) by another effect. For example, for the three-way interaction in the pre-
vious paragraph (example 4), we may summarize that the two-way interaction
between test difficulty and achievement orientation is modified (qualified) by gen-
der.
4.3. Mathematical model to investigate variability of a measuring appara-
tus with the help ofANOVA.
4.3.1. A few words about the model.
For the production process of parts manufacture there is a basic characteris-
tic of each part [10]. It can be represented as follows:
1) Reference value = Mean for all parts + Deviation of one part.
In practice there is an error of the measuring system that could be written
as follows:
2) Value observed = Basic value + Measurement error.
Usage of the below ratio is a simple computation of the measurement error:
3) Measurement error = Shift + Observer influence + Repeatition error.
Shift is a value representing total systematic error caused by an instrument.
Additional errors may be contributed by different observers. A repeatition error
shows differences during various measurements of one and the same part repeated
by the same instrument and the same observer. Combination of items 1), 2) and 3)
results in the following model:
4) Value observed = (Mean for all parts + Shift) + Deviation of one part +
Observer influence + Repeatition error.
Mean for all the parts and a shift are constant values. They can not be esti-
mated separately without a reference instrument. Effect (influence) of a part, effect
of an observer and repeatition error are random variables. A part effect reflects
variability of the production process. An observer effect depends on the observer
difference and is sometimes called "reproducibility". A repeatition error repre-
sents a variability of the measurements repeated by the same observer and in-
strument and is sometimes called "precision". Mean for all parts and part devia-
tion are properties of the production process and do not depend on the instru-
ment. The other members of the model reflect an instrumental error (measuring
system error).
20

4.3.2. Mathematical representation of the model.
Let's designate m-measurement of one part performed by j-observer as Yjjm.
Each part has a basic value, let's say Xj, that is impossible to measure in reality.
Then designate an error added to Xj in practice as Sjjm and the following equation
can be written:
Yjjm = Xj + Sjjm (1)
or
observed value = basic value + error.
We assume that basic values Xj are distributed under the normal law with
u.- mean and a2
- variance, where u. is a mean value of the parts. The equation
(1) can be rewritten as follows:
Yjjm = u + a i + eijm (2)
where: aj= Xj- u. - a part deviation with zero mean and a -variance.
An error may be considered as a systematic overall shift available due to
the instrument and observer, an additional shift due to the influence of an ob-
server and measurement repeatition error. Thus, in the formula (2) model pa-
rameter 8jjm can be replaced with a sum of the parameters associated with the in-
strument shift (b), influence of observers (Pj) and repeated measurements (£jjm):
eijm = b + Pj + e,jn, (3)
or
error = instrument shift + observer effect + repeatition error.
Combining (2) and (3) the model can be written as follows:
Yijrn = (n + b)+ <xi+ pj+ei j m (4)
or
observed value = (mean for all the parts + instrument shift) + part influence
+ observer influence + repeatition error.
Model parameters a;, fy and Sjjm are normally independently distributed
values with zero mean and variances a2
, co2
, x2
, where x2
measures the instru-
ment precision, co2
variance measures variability of observers'influence, i.e.
reproducibility and a variance measures variability associated with a part.
In an additive model where interaction between an observer and
part is not significant variance of the measurement process equals to:
VAR (YUm) = a2
+ co2
+ T2
(5)
21

In a non-additive model where interaction between an observer and part is
significant, in the formula (4) there appears an additional member A
,
yrepresenting
a specific interaction between j-observer and i-part:
Yijm = (n + b)+ ai+ pj + Xij + eUm (6)
If A
-
y is normally distributed with zero mean and y variance, then total
variance for the expression (6) is determined as follows:
VAR (Yijm) = a2
+ (co2
+ y2
) + T
2
(7)
4.4. An example of investigation under metrological qualification pro-
gram of measurement procedure including a given instrument,
using ANOVA technique.
Let's assume that under the metrological qualification program it is re-
quired to assess suitability of a measuring system consisting of a measuring in-
strument (thickness gauge) intended for measuring a part's thickness, measure-
ment procedure for this dimension and operators to perform measurements. Deci-
sion on the suitability of the measuring system should be made in accordance
with the criteria established by the International Guidelines [10], the criteria are
given in Table 2 thereto.
Table 2. Criteria for suitability estimate of a measuring system based on pre-
cision and reproducibility (P&R),%
Estimate
(P&R),%
Recommen-
dations for decision-
making about
measuring system
suitability
Less than
10%
Measuring
system is ac-
ceptable
from 10% to
30%
May be ac-
cepted depending on
application impor-
tance, instrument
price, etc.
More than 30%
System needs
to be improved. Take
every effort to find a
cause and solve the
problem.
In order to obtain reliable initial data to the further processing by ANOVA
technique, an experiment for the purpose of receiving measurement has been thor-
oughly planned. A simple way to provide for a balanced system consisting of n
part, k observers and r attempts is randomization. If the data have not been col-
lected on a random basis it may result in value shift. One of the usual ways of ran-
domization is to write on pieces of paper designations from Ai to An for the meas-
urements of n parts to be performed by the 1-st observer, then from Bi to Bn - for
the second observer, from C to Cn - for the third observer, etc. When kn combi-
nations have been written, the pieces of paper are put into a hat or bag. They are
selected one by one and determine a measurement sequence for the instrument in-
vestigation. When all the kn combinations have been selected they are again put
into the hat or bag and the procedure repeats. It will be done until an experiment
sequence has been determined for each repeatition. But it is better to use the Table
of random numbers (see. for example, the Table in Appendix 8 [11]). In our case
22

IAEA Regional Training Course under the Programme "Licensing Fuel andFuel Modelling Codes" Bratislava(Slo-
there were 10parts and3 observers, the experiment wasperformed in a random
order two times foreach combination ofa part and observer (operator). Themeas-
urement results aregiven inTable3.
A quantitative analysis ofvariance by ANOVA isperformed according to the
formulas given in thebook [12],theanalysis results are shown in Table 4. This
Table contains six columns. A source column describes thecauses of variability.
DF column is degrees of freedom associated with the source. SS column or the
sums of the squares is a deviation ofthe source from theaverage. MScolumn or
mean squares is thesumof squares divided bydegrees offreedom. Thecolumn of
F ratio is computed only for the interaction, it is determined by the interaction
mean square divided by the error mean square. EMS column or estimated mean
squares determines a combination of variance compounds for each mean square
MS. ANOVA permits toexpand theoverall variability into four compounds: parts,
observers, interaction between observers andparts aswell as repeatition errordue
to the instrument.
Table 3. Control list of data for measuring system analysis.
Nfi
Operators
and
attempts
Control units
1
2
3
0,65
0,6
0,85
0,8
0,85
0,95
0,55
0,45
0,95
0,95
4 |
1
2
3
0,55
0,55
1,05
0,95
0,8
0,75
0,85
0,85
0,4
0,4
1
1,05
0,95
0,9
1
2
3
0,5
0,55
1,05
1
0,8
0,8
0,8
0,8
0,45
0,5
1
1,05
0,95
0,95
13 Unit
average
0,567 1,008 0,8 0,85 0,458 1,017 0,942 0,783
Based on themeasurement results given in Table 3 we compute a sum of
squares SSP from variability related to theparts (control units), a sum of squares
SS0 from variability related to the observers (operators), a sumof squares SSop
from variability dueto theinteraction between an operator andpart, as well asa
sum of squares SSe from variability dueto therepeated measurements (attempts)
by the same observer using the same part. Wehave:
23

Part (n=10) => SSP = kr E ( x / ~X ) =
3-2[(0,567-0,81)2
+ (1,008-0,81f +
(0,8-0,81)2
+ (0,85-0,81)2 + (0,458-0,81f + (1,017-0,81)2
+ (0,942-0,81f + (0,783-0,81 f
+ (1,008-0,81)2
+ (0,667-0,81)2
] = 6- 0,344512 = 2,067
k
(- - V
Observer fk=3) => SSO = nr J ] x . - x = 10-2[ (0,8275-0,81f + (0,775-
0,81)2+ (0,8275-0,81)2] = 0,0367
Interaction between an observer and a part
x t
SSop
M I"
2- [( 0,625-0,567-0,8275+0,81)2 + (1,0-1,008-0,8275+0,81;
),81)2
+ (0,9-0,85-0,8275+0,81)2
+ (0,5-0,458-0,8275+0,81;
H (0,825-0,8-
0,8275+0,81)^ + (0,9-0,85-0,8275+0,81)' + (0,5-0,458-0,8275+0,81)^ + (1,0-1,017-
0,8275+0,81)2 + ^0,95-0,942-0,8275+0,81)2 + (0,825-0,783-0,8275+0,81/ + (1,0-1,008-
0,8275+0,81)2 + (0,65-0,667-0,8275+0,81/ + (0,55-0,567-0,775+0,81/ + (1,0-1,008-
0,775+0,81f + ( 0,775-0,8-0,775+0,81/ + (0,85-0,85-0,775+0,81 f + (0,4-0,458-
0,775+0,81r + (1,025-1,017-0,775+0,81 r + (0,925-0,942-0,775+0,81 f + (0,725-0,783-
0,775+0,81)2 + (0,975-1,008-0,775+0,81)2 + (0,525-0,667-0,775+0,81 )2
+ (0,525-0,567-
0,8275+0,81)2 + (1,025-1,008-0,8275+0,81)2
+ (0,8-0,8-0,8275+0,81 f + (0,8-0,85-
0,8275+0,81)2
+ (0,475-0,458-0,8275+0,81)2
+ (1,025-1,017-0,8275+0,81)2
+ (0,95-
0,942-0,8275+0,81)2
+ (0,8-0,783-0,8275+0,81 f + (1,05-1,008-0,8275+0,81f + (0,825-
0,667-0,8275+0,81)2] = 0,1074
Remaining (error) =>SSe =
,=l 7=1
~r
X,•= & 0,7825-2(0,625)2
] + [2,0 -
2-(1,0)2
] + [1,3625-2(0,825)2
] + [1,625-2(0,9)2
] + [0,505-2-(0,5)2
] + [2-2] + 0 + [1,3625 -
2-(0,825)2
] + 0 + [0,85 - 2(0,65)2
] + 0 + [2,005 - 2(1,0)2
] + [1,2025 -2(0,775)2
] + 0 + 0 +
[2,1025 - 2(1,025)2
] + [1,7125-2(0,925)2
] + [1,0525 - 2(0,725)2
] + [1,9025 - 2(0,975)2
]
+ [0,5525 -2-(0,525)2
] + [0,5525 -2-(0,525)2
] + [2,1025 - 2(1,025)2
] + 0 + 0 + [0,4525 -
2-(0,475)2
] + [2,1025 - 2-(1,025)2
] + 0 + 0 + 0 +[1,3625 - 2(0,825)2
] = 0,0375
Overall variance ^ S S T = 41,615 - 60-(0,81)2
= 41,615 - 39,366 = 2,249
Table 4
Source
Observer
Parts
Observer x part
Instrument error
Overall variance
DF
2
9
18
30
59
SS
0,0367
2,067
0,1074
0,0375
2,249
MS
0,01838
0,22967
0,00597
0,00125
F
4,77(
*)
EMS
T2
+ 2y2
+ 2Oco2
T2
+ 2y2
+ 6a2
T2
+ 2y2
T2
Note (*): interaction «observer x part» is significant at significance level
a<5%, because F(3, 10) = 3,71 (see. Table [11]).
The estimates of variance compounds for each source were computed ac-
cording to the below formulas:
24

Instrument ^apparatus)
Interaction
Variance estimate
T2
= MS.
2_(MSOP-MS(
Observer
MS0-MSOI
nr
Part
Table 5
MSP-MSni
kr
Variability estimate
Precision x2
= 0,00125
Observer m2
= 0,00062
Interaction y2
= 0,002359
Precision and reproducibility (P and R)
(P&R = T2
+ co2
+ y2
)
Part a2
= 0,03728
Standard
deviation
a
0,0354
0,0249
0,0486
0,00422
0,193
Contribution, %
3,0
1,5
5,7
10,2
89,8
It is seen from Table 5 that variability value due to precision and reproduci-
bility makes 10,2%. Thus, using the criteria for suitability estimate of a measuring
system based on the precision and reproducibility values given in Table 2, a con-
clusion can be made that the given instrument is in order and suitable for meas-
urements.
PART 2. CONTROL OF PROCESSES FOR PURPOSE OF QUALITY
ASSURANCE IN PRODUCTION.
1. ORGANIZATIONAL BASIS FOR QUALIFICATIONS OF TECH-
NOLOGICAL PROCESSES.
1.1. Control of processes. What does it mean?
At the plant there should be developed a long-term strategy of management
that assures the achievement of a maximum effect taking into consideration func-
tional organization of a company and orients all kinds of activities to the final
maximum effect for the company. The control of processes is such a strategy [13].
25

The activities of the employees and production workers can be considered as
processes that are controlled similar to the production processes. Most types of
employees' activities such as design, product sales, data processing, etc. do not
yield to the production processes in complexity. In the past primary attention was
only paid to the control of the production processes. At present an efficiency level
is ensured by reliable management of the production process and its close interac-
tion with all key activities of a company.
A modern model of a company management is based on the following:
meeting a customer's and employees' demands as well as effect on the society is
achieved through quality leadership regulating a policy and strategy, staff, re-
sources and processes, finally resulting in perfection of the activity results.
The basic motivation to improve company management model nowadays is
a strong competitiveness among the producers of nuclear fuel in our traditional
markets in the Eastern Europe and Finland. A greater impact on the improvement
of the quality system valid at the plant has the requirements of our customers too.
It should be noted that there is a strong demand of the plant itself to work
with higher profits and labor productivity, because an improvement of economical
efficiency of the company's activities (for example, due to decrease of poor quality
costs) allows to meet customers' needs and assures a better reimbursement of in-
vestments made in production.
The following definition of a "process" is proposed according to the Quality
Management and Quality Assurance Vocabulary (ISO 8402):
a set of inter-related resources and activities which transforms inputs into
outputs.
It is worth mentioning that a series of operations (activities) carried out with
the source material (process input) increases its value and yields in a certain result
(process output). The value of the source material is being increased by usage of
the qualified labor and knowledge.
1.2. A role the manager of a process. Process improvement group.
Executive's responsibilities are to ensure functioning of the whole process
being interconnected with all organization departments, to increase its efficiency
and to support modernization positively affecting the whole process. Why a man-
ager of the process is required? If nobody manages the process it will never be im-
proved.
A priority task of the process manager is a strict determination of the process
boundaries taking into account initial input by a supplier of the resources needed
for the process and a final step stipulating the transfer of the results to a consumer.
When the process boundaries have been determined the process improvement
group may be formed.
The process manager selects candidates for the creation of the process im-
provement group comprising representatives of all the departments interconnected
with the process. Each member of the group is appointed by the head of the de-
partment concerned and is a representative in the group that is responsible for ac-
26

tivities providing the process improvement. The main responsibilities of the group
are:
1) review of the process flow chart available;
2) determination of control areas and feed back;
3) process qualification;
4) development and implementation of the plans for the process im-
provement;
5) preparation of the reports on the process quality, efficiency and proc-
ess changes;
6) development and implementation of resource control system to assure
a timely operation of the process;
7) evaluation of the process results and development of proposals aimed
at the process improvement.
With regard to the theme of our lecture I would like to dwell at length
on the process qualification procedure.
1.3. Process qualification.
Process qualification is a set of actions for the purpose of inspection of the
process functioning in accordance with the requirements prescribed.
A process is considered prepared for the qualification when the following is
available: necessary procedures, documentation, training programs and programs
for staff preparation, measurement procedures, documents designed for the control
of inspection and measuring equipment as well as programs for a smooth func-
tioning of the production process being a guarantee of the high quality levels of
products even under strenuous work.
A difficulty in the qualification performance depends on the complexity of
the process itself because while checking the conformity of the parameters of a
complex process to the specified requirements it is required to inspect more pa-
rameters than for the qualification of a simple process.
A system of technological process qualifications established at OAO "MSZ"
is based on testing of products. The initial stage of the qualification is performed
during preliminary and acceptance tests of new or modified products or during the
qualification tests of the products manufactured before by other suppliers. The
further qualifications are carried out during periodic tests of the products. The
normative documents regulating preliminary, acceptance and periodic tests of the
products have been developed at the plant. Based on the test results the coefficients
of stability and technological process adjustments Cp, Cpkmin and CPkmax are deter-
mined. A decision on the conformity of the technological process to the require-
ment specified in the design documentation is made depending on the coefficients.
By the demand of individual customers a specific qualification program is
developed for the processes to be performed under a particular order.
A qualification may be performed of a separate production operation or
equipment or of the whole technological process. A company standard for the
qualifications of technological instrumentation and technological process as a
whole is currently developed at the plant.
27

The purpose of the qualification of the production process is to create a con-
trol condition of the process prior to the beginning of mass production of products
for consumers.
1.4. Use ofreproducibility indexes for technological process qualification.
It is a standard task for the technological processes to estimate a process be-
havior, its stability.
When a process output index (quality index) is measured according to the
quantitative attribute and the upper and lower limits (tolerance boundaries) of the
index are specified, process quality can be estimated with the help of reproduci-
bility indexes. These indexes show to what extent a process is satisfactory,
whether there is a confidence that the tolerance requirements will be fulfilled. The
reproducibility indexes play a special part in the quality systems of companies. In
modern technological processes it is required to assure a very high confidence of
the quality index being within the specified tolerance. If a process is controllable,
i.e. there is a possibility of purposeful change of the process parameters, then the
reproducibility indexes show how the process parameters should be changed to as-
sure a higher confidence of the fulfillment of the tolerance requirements.
There are indexes characterizing a process at a relatively short interval of
time (Cp and CPk indexes) and indexes that characterize a long-term process be-
havior (PpandPPk indexes), see Fig.1. To determine indexes properly it is im-
portant to know a variability nature of a particular process, i.e. a quality index be-
havior at the process output.
Process reproducibility indexes
Indexes of short-time
processbehavior
Indexes of long-time
process behavior
Index of
process
capabilities
Index of
process
centralization
pk
Index of
process
suitability
Index of
process serv-
iceability
Momen-
tary
cP
One mo-
mentary
sample
Average
cP
A few
momen-
tary sam-
ples
Momen-
tary
cP
One mo-
mentary
sample
Average
cP
A few
momen-
tary sam-
ples
Fig.l. Qualification of process reproducibility indexes.
1.4.1. Quality index behavior atprocess output (process variability).
Values of the quality index at the technological process output measured in
accordance with the quantitative attribute change in time, i.e. from one item to an-
other one. These changes are called a process variability.
28

A basic characteristic of process variability is a spread of the quality index
values characterized by standard deviation a or variance D = a2
of the process.
a characterizes an average value of deviation of random values from
their total true average value i.
For the majority of processes u. and a parameters change in time and
depend on the duration of observance. A variability of the most processes can be
described by means of two models shown in Fig.2 andFig.3.
x
1-st period
of statistical
stability
2-nd period
of statistical
stability
k-lh period
of statistical
stability
Fig.2. Periods of statistical stability of process behavior (model 1).
It is characteristic of model 1 (Fig.2) to have periods of statistical stability
during that time a process stores (keeps) the average values J
L
X and standard devia-
tionCT.Such behavior of a technological process is typical, for example, for ma-
chining processes. Parameters p.and a can be estimated during one period of sta-
tistical stability by a corresponding momentary sample. The term "momentary"
sample indicates that a sample is drawn for rather short period of time when the
parameters |u and o do not change.
At certain (known) moments there are changes in the technological proc-
ess such as adjustments, replacements of tools and others which by assumption
can change parameter (j, and possibly, parameter a too. Meanwhile until the next
considerable known change there occurs a transition of the technological process
to another statistically stable condition with other values of parameters p. and
a. They also can be estimated by a new momentary sample. Technological proc-
ess behavior under this model is given in Fig 2 in the form of the graph of data
change at the technological process output (according to one quality index x).
29

X •
Fig.3. Smooth change of p. and cr process parameters (model 2).
A smooth change of parameters |i and a is supposed for model 2 (Fig.3),
when it can be assumed that parameters u. and a are constant values over rather
short interval of observance. Such behavior is characteristic for example, for
chemical, electrochemical and metallurgical processes. There also is a meaning of
momentary sample — process parameters u.and a change insignificantly in com-
parison with a value over the time when the sample is drawn.
Over longer intervals of time parameters u. and a are subject to significant
but smooth changes.
In general, a real behavior of the technological process may represent a
combination of model 1 and model 2, however it is assumed that in any case it is
possible to indicate rather short time intervals for drawing momentary samples.
In many cases parameter a, corresponding to the momentary sample is
more stable and less subject to changes than parameter [i. Parameter <j is deter-
mined in general by accurate parameters of equipment, tool quality, a spread in
input material properties (raw material) and other conditions, their joint influence
may be approximately the same over a large period of observance.
1.4.2.Variance and standard deviation of process.
During short intervals of time corresponding to the meaning of momentary
sample a variability of output quality index is observed, however this variability is
relatively small and named own variability (inherent to the given process). The
reasons for this variability are named usual reasons (causes), for example, non-
absolute stiffness of a machine during mechanical treatment (machining), natural
inevitable fluctuations of the temperature and gas composition during thermal
treatment, etc.
Over a longer period of time special reasons are added to the influence of
the usual reasons. Special reasons become apparent at some moments between
the periods of statistical stability (model 1, Fig.2) or are in constant effect and
cause smooth, but significant changes of the quality index (model 2, Fig.3).
The total effect of usual and special reasons provides an overall variability
of the process.
30

An own variability of the process is characterized by inherent variance Dc
-cc
2
or an inherent standard deviation ac. An overall variability of the process is
characterized by overall variance Dn =an
2
or
overall standard deviation crn. An overall variance is always greater than an
inherent one, an overall standard deviation is always greater than an inherent one:
Dn >Dc (1)
<Jn > <*c (2)
In ideal practically unachievable case the inequalities (1) and (2) turn into
equalities.
An inherent and overall variances of the process have a different physical
meaning and a different value and are computed by different samples though
identical or similar formulas are used for their computations.
Inherent variance of the process Dc =CTC
2
is unknown in practice but it can
be estimated by one or several momentary samples.
An inherent variance is estimated by one momentary sample according to
the following formula:
A A
1 " / V
Dc=(Jc::=
Sn-i:=
~ : i ( x , — x) (3)
where: D c - estimate of unknown inherent variance of the process Dc;
ac — inherent standard deviation; ac - its estimate;
Sn.i - function for ac estimate;
n - momentary sample size;
x - sample average:
n
(4)
An inherent variance and inherent standard deviation of the process esti-
mated by one sample according to the formula (3) can have significant deviations
from true values of inherent variance Dc and inherent standard deviation ac due to
sample randomness. Accuracy of these estimates increases with a momentary sam-
ple size increase and can be considered acceptable provided two conditions are ful-
filled:
1) momentary sample size is at least 20;
2)from engineering considerations it is known that an inherent variance of
the process does not change, i.e. a spread in momentary samples does not depend
on the factors that can change during the process, that is why one momentary sam-
ple is enough for an estimation of CTC.
Inherent variance Dc can be estimated with a greater accuracy by several
momentary samples. However it can be done if Dc is constant. An estimation pro-
cedure for an inherent variance of the process based on several momentary samples
will be established in one of the national standards of Russia being currently de-
veloped. If the inherent variance of a process Dc changes in time the process is sta-
31

tistically unstable as to variance and it is impossible to estimate the inherent vari-
ance based on several momentary samples.
Total (overall) variance Dn of a process is also unknown in practice but it
can be estimated by a total sample drawn over a long period of observation. A
total sample may comprise separate observations or a few momentary samples
drawn under different states of a process (Fig.2 and 3). In any case over an ap-
propriate period of observation all natural changes of a process, typical for a par-
ticular process (adjustments, replacements of raw materials, operators, etc.) should
take place.
An estimate of the overall variance of a process can be obtained by the fol-
lowing formula:
(5)
T 7 r
A
where: Dn - estimate of unknown overall variance Dn of a process;
an - overall standard deviation; an - its estimate;
Sn_i - function for crn estimate;
N - total sample size;
x - total average:
~ '<• (6)
By a dispersion field of a process is understood a dispersion area of a quality
index which contains the main part of quality index values being close to 1. Fig.4
shows the dispersion field of a process with a quality index distributed under the
normal law.
w(x)
Fig.4. Process dispersion field.
Usually, an interval of total length 6a is assumed as a dispersion field, in
case of normal law the interval contains 99.73% of all quality index values.
32

The inherent dispersion field of a process has 6a and corresponds to the
quality index dispersion over a short interval of time that matches a momentary
sample. In practice 6ac estimate is used. The overall dispersion field of a process
has 6crn and corresponds to the total dispersion of the quality index over a long pe-
riod of observation. In practice 6an estimate is used. By ac and an relationship
(their estimates) or by the relationship of the inherent and overall dispersion fields
it is possible to judge an effect of special reasons of variability: the more an ex-
ceeds ac, the greater effect special reasons have, i.e. the less is a long-term stability
of a technological process and its potentials are less used to provide for a specified
tolerance.
1.4.3. Computations indexes and analysis of technological process state.
Index Cp of process capabilities is determined as tolerance (TB - T(I) to in-
herent dispersion field 6crc ratio. Cp index shows process potentials to fulfill the re-
quirements for tolerance over short periods of time corresponding to the mo-
mentary sample concept. The higher CPJ the greater accuracy of a process, the
bigger process potentials are for assurance of the given tolerance over short periods
of time. For statistically unstable processes as far as variance in concerned, index
Cp will change in time, thus it is neither used for a process, nor for estimation of
process potentials.
A process may be considered satisfactory for the assurance of the given tol-
erance if
G * 1,33 (7)
Index of centralization Cpk of a process (adjustment to the central line of
tolerance) is determined by the formula:
Cp,= (X
M v n
(8)
where: min. {A;B} - minimum from two values A and B;
[i- current value of the true average for a process;
o - inherent standard deviation of a process;
TB and TH - upper and lower tolerance boundaries.
For the processes being statistically unstable as to variance, index Cpkis
not applied.
Cpkindex shows a current quality level of a process regarding the fulfillment
of the tolerance requirements both from the point of average process value (^ ad-
justment to the tolerance center and from the point of the inherent dispersion field.
The higher Cpk the greater confidence a process provides in the fulfillment of the
tolerance requirements at a current time.
33

When process average x is located in the tolerance center index CPk is
maximum equal to Cp. If there are any deviations of (a. from the tolerance center,
Cpk decreases; it becomes equal to zero when i is located on either boundary of the
tolerance zone and becomes negative if xdeviates from the tolerance zone.
A process is considered to operate satisfactory at a current time if
The comparison of indexes Cp and CPkallows to draw a conclusion about
process adjustment u,to the tolerance center.
If CPk ^ 0.75 Cp a process is satisfactorily adjusted, if CPk ^ 0.75 Cpit means
that process potentials (potential accuracy) are not used satisfactorily due to poor
adjustment ju.of a process to the tolerance center.
Suitability index Pp of a process is determined as tolerance zone (TB - TH) to
overall dispersion field 6crn ratio. Index Pp is applied for any processes including
those being statistically unstable as to inherent variance ac
2
. Index Pp shows a
process suitability for the assurance of the tolerance requirements taking into ac-
count the overall process variance estimated over rather long time interval taking
into considerations all natural changes in the process (replacements of material
lots, operators, etc.). The higher Pp the greater confidence a process may assure to
meet the tolerance requirements. A process may be considered suitable for the as-
surance of a given tolerance if
PP>1 (10)
Serviceability index Ppkof a process is computed by a formula derived from
the formula (8) by replacing the inherent standard deviationCTCwith the overall
standard deviation an of a process as well as by means of substitution of the total
true process average for a current value of the true process average over the ob-
served period. Index Ppk is used for any processes including those being statisti-
cally unstable as to variance crc
2
. Index Ppk indicates a process quality achieved in
practice taking into account both the overall dispersion field and average adjust-
ment of the process during the observation time. The higher PPk the greater confi-
dence a process may assure to meet the tolerance requirements.
Conclusion.
Summary of the above said may confirm that the quality aspects of the prod-
ucts being solved with the help of QS have been given and are still given the high-
est priority. We realize that under market conditions it is a basis of our welfare
now and in the future.
34

References.
[I] - V.Okrepilov "Quality Management", M., "Economics", 1998.
[2] - N.Rubichev, V.Frumkin "Validation of Tolerance Quality
Control", M., "Standards Publishing House", 1990.
[3] - GOST 16263-70. Metrology. Terms and definitions.
[4] - Company standard - STP 11-272-98. "Quality system.
Control of inspection, measuring and test equipment.
COMPANY STANDARDS. Basic provisions".
METROLOGICAL EXPERTISE OF TECHNICAL
DOCUMENTATION. Organization and performance
procedure".
AUTOMATION DEVICES, NON-STANDARDIZED
CONTROL AND MEASURING EQUIPMENT. Design,
tests and implementation".
INFORMATION AND MEASURING SYSTEMS AND
MEASUREMENT CHANNELS. Metrological assurance".
MEASUREMENT PROCEDURES. PERFORMANCE OF
METROLOGICAL QUALIFICATION".
[9] - E.Schindowski, O.Schurz "Statistical Methods of Quality
Control", M., "Mir" Publishing House, 1976.
[10] - "Analysis of Measuring Systems", MSA guidelines.
Translation from English. N.Novgorod, AO "NITS KD",
SMTS "Prioritet", 1997.
II1] - D. Cowden "Statistical Methods in Quality Control",
translation from English, Publishing House of physical and
mathematical literature, M., 1961.
[12] - A.Afifi, S.Azen "Statistical Analysis. A Computer Oriented
Approach", M., "Mir", 1982.
[13] - HJ.Harrington "How America's Leading Corporations
Improve Quality". Translation from English. M.,
Economics", 1990.
35

cualificacion metrologos

Recommended

Recommended

More Related Content

What's hot

What's hot (11)

Similar to cualificacion metrologos

Similar to cualificacion metrologos (20)

Recently uploaded

Recently uploaded (20)

cualificacion metrologos