Sie sind auf Seite 1von 30

ASME B89.7.3.

(Technical Report)

Guidelines for
the Evaluation
of Dimensional


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale

ASME B89.7.3.2-2007
(Technical Report)

Guidelines for
the Evaluation
of Dimensional

Three Park Avenue • New York, NY 10016

Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
Date of Issuance: March 2, 2007

This Technical Report will be revised when the Society approves the issuance of a new edition. There
will be no addenda issued to this edition.

ASME is the registered trademark of The American Society of Mechanical Engineers.

ASME does not “approve,” “rate,” or “endorse” any item, construction, proprietary device, or activity.
ASME does not take any position with respect to the validity of any patent rights asserted in connection with any
items mentioned in this document, and does not undertake to insure anyone utilizing a standard against liability for
infringement of any applicable letters patent, nor assumes any such liability. Users of a code or standard are expressly
advised that determination of the validity of any such patent rights, and the risk of infringement of such rights, is
entirely their own responsibility.
Participation by federal agency representative(s) or person(s) affiliated with industry is not to be interpreted as
government or industry endorsement of this code or standard.

No part of this document may be reproduced in any form,

in an electronic retrieval system or otherwise,
without the prior written permission of the publisher.

The American Society of Mechanical Engineers

Three Park Avenue, New York, NY 10016-5990

Copyright © 2007 by
All rights reserved
Printed in U.S.A.


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale

Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
Committee Roster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Simplifications in the Evaluation of Measurement Uncertainty. . . . . . . . . . . . . . . . . . . . . . . . . . 1
3 Basic Concepts and Terminology of Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
4 Combining Uncertainty Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5 Basic Procedure for Uncertainty Evaluation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1 Measurement Uncertainty Quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1 Measurement and Validity Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Nonmandatory Appendices
A Type A Evaluation of Standard Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
B Type B Evaluation of Standard Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
C Influence Quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
D Thermal Effects in Dimensional Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
E Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale

The ISO Guide to the Expression of Uncertainty in Measurement (GUM) is now the internation-
ally accepted method of expressing measurement uncertainty [1]. The U.S. has adopted the GUM
as a national standard [2]. The evaluation of measurement uncertainty has been applied for some
time at national measurement institutes but more recently issues such as measurement traceability
and laboratory accreditation are resulting in its widespread use in calibration laboratories.
Given the potential impact to business practices, national and international standards commit-
tees are working to publish new standards and technical reports that will facilitate the integration
of the GUM approach and the consideration of measurement uncertainty. In support of this
effort, ASME B89 Committee for Dimensional Metrology has formed Division 7 — Measurement
Measurement uncertainty has important economic consequences for calibration and measure-

ment activities. In calibration reports, the magnitude of the uncertainty is often taken as an
indication of the quality of the laboratory, and smaller uncertainty values generally are of higher
value and of higher cost. ASME B89.7.3.1, Guidelines for Decision Rules in Determining Confor-
mance to Specifications [3], addresses the role of measurement uncertainty when accepting or
rejecting products based on a measurement result and a product specification. This document,
ASME B89.7.3.2, Guidelines for the Evaluation of Dimensional Measurement Uncertainty, provides
a simplified approach (relative to the GUM) to the evaluation of dimensional measurement
uncertainty. ASME B89.7.3.3, Guidelines for Assessing the Reliability of Dimensional Measurement
Uncertainty Statements [4], examines how to resolve disagreements over the magnitude of the
measurement uncertainty statement. Finally, ASME B89.7.4, Measurement Uncertainty and Con-
formance Testing: Risk Analysis [5], provides guidance on the risks involved in any product
acceptance/rejection decision.
With the increasing number of laboratories that are accredited, more and more metrologists
will need to develop skills in evaluating measurement uncertainty. This report provides guidance
for both the novice and experienced metrologist in this endeavor. Additionally, this report may
be used to understand the accuracy of measurements at a more comprehensive level than the
variation captured by “Gage Repeatability and Reproducibility” (GR&R) studies. This will provide
a higher level of confidence in the measurements and aid in determining if a measurement system
is capable of meeting the expected capability as a percentage of specified tolerance. Emphasis is
placed on simplified uncertainty evaluation appropriate for the reader who is experienced in
measurement procedures but is new to uncertainty evaluation.
Comments and suggestions for improvement of this Technical Report are welcome. They should
be addressed to The American Society of Mechanical Engineers, Secretary, B89 Main Committee,
Three Park Avenue, New York, NY 10016-5990.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
Dimensional Metrology
(The following is the roster of the Committee at the time of approval of this Technical Report.)


B. Parry, Chair
D. Beutel, Vice Chair
F. Constantino, Secretary


J. B. Bryan, Bryan Associates R. Hook, Metcon
T. Carpenter, U.S. Air Force Metrology Laboratories M. Liebers, Professional Instruments
T. Charlton, Jr., Charlton Associates S. D. Phillips, National Institute of Standards and Technology
D. Christy, Mahr Federal, Inc. J. Salsbury, Mitutoyo America
G. A. Hetland, International Institute of Geometric Dimensioning D. A. Swyt, National Institute of Standards and Technology
and Tolerancing B. R. Taylor, Renishaw PLC
R. J. Hocken, University of North Carolina at Charlotte


G. Hetland, Chair, International Institute of Geometric M. Krystek, Physikalisch–Technische Bundesanstalt
Dimensioning and Tolerancing M. Liebers, Professional Instruments
D. Swyt, Vice Chair, National Institute of Standards and Technology B. Parry, The Boeing Co.
D. Beutel, Caterpillar P. Pereira, Caterpillar
J. Buttress, Hutchinson Technology, Inc. S. Phillips, National Institute of Standards and Technology

T. Carpenter, U.S. Air Force Metrology Laboratories J. Salsbury, Mitutoyo America
T. Charlton, Charlton Associates C. Shakarji, National Institute of Standards and Technology
F. Constantino, The American Society of Mechanical Engineers J. Raja, University of North Carolina at Charlotte
W. T. Estler, National Institute of Standards and Technology


S. Phillips, Chair, National Institute of Standards and Technology M. Krystek, Physikalisch–Technische Bundesanstalt
J. Buttress, Hutchinson Technology B. Parry, The Boeing Co.
T. Carpenter, U.S. Air Force Metrology Laboratories P. Pereira, Caterpillar
T. Charlton, Charlton Associates J. Salsbury, Mitutoyo America
T. Doiron, National Institute of Standards and Technology R. Thompson, U.S. Air Force Metrology Laboratories
W. T. Estler, National Institute of Standards and Technology

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
--`,,```,,,,````-`-`,,`,,`,`,,`--- ASME B89.7.3.2-2007



ABSTRACT and at best represents a slight refinement of the uncer-

tainty statement. Indeed, even in the determination of
The primary purpose of this Technical Report is to
fundamental constants the practice of using degrees of
provide introductory guidelines for assessing dimen-
freedom has been abandoned [6].
sional measurement uncertainty in a manner that is less
Correlations can exist between uncertainty sources;
complex than presented in the Guide to the Expression
however, most uncertainty evaluations involve uncorre-
of Uncertainty in Measurement (GUM). These guide-
lated uncertainty sources. Consequently, correlation
lines are fully consistent with the GUM methodology
effects are omitted in this document, except for some
and philosophy. The technical simplifications include
guidelines to identify when they are present and hence
not assigning degrees of freedom to uncertainty sources,
more advanced methods (beyond the scope of this docu-
assuming uncorrelated uncertainty sources, and
ment) are needed.
avoiding partial differentiation by always working with
Accordingly, this guideline has the following two
input quantities having units of the measurand. A
detailed discussion is presented on measurement uncer-
(a) Uncertainty sources are not assigned any degrees
tainty concepts that should prove valuable to both the
of freedom (i.e., no attempt is made to evaluate the
novice and experienced metrologist (Nonmandatory
uncertainty of the uncertainty). Hence, it is assumed
Appendices A and B). Potential influence quantities that
that the expanded (k p 2) uncertainty interval has a
can affect a measurement result are listed in Nonmanda-
95% probability of containing the true value of the
tory Appendix C. Worked examples, with an emphasis
on thermal issues, are provided in Nonmandatory
(b) All uncertainty sources are assumed to be uncorre-
Appendix D. The bibliography is located in Nonmanda-
lated. Finally, for simplicity, all input quantities of the
tory Appendix E.
uncertainty budget are packaged in quantities that have
the unit of the measurand (i.e., length). This avoids the
1 SCOPE issue of sensitivity coefficients that typically involve par-
tial differentiation.
These guidelines address the evaluation of dimen-
sional measurement uncertainty. Emphasis is placed on
simplified methods appropriate for the industrial prac- 3 BASIC CONCEPTS AND TERMINOLOGY OF
titioner. The introductory methods presented are consist- UNCERTAINTY
ent with the Guide to the Expression of Uncertainty in
The formal definition of the term “uncertainty of mea-
Measurement (GUM), the nationally [2] and internation-
surement” in the current International Vocabulary of
ally [1] accepted method to quantify measurement
Basic and General Terms in Metrology (VIM) [7]
uncertainty. The use of these guidelines does not pre-
(VIM entry 3.9) is as follows:
clude the use of more advanced methods in the uncer-
tainty evaluation process. uncertainty (of measurement): parameter, associated with
the result of a measurement, that characterizes the dis-
persion of the values that could reasonably be attributed
MEASUREMENT UNCERTAINTY This can be interpreted as saying that measurement
To simplify and focus the uncertainty evaluation pro- uncertainty is a number that describes an interval cen-
cess in an industrial setting, issues associated with the tered about the measurement result where we have rea-
effective degrees of freedom of the uncertainty statement sonable confidence that it includes the “true value” of
and correlation between uncertainty sources are consid- the quantity we are measuring.
ered less important when compared to problems associ- expanded uncertainty (with a coverage factor of 2), U: a
ated with underestimating or omitting uncertainty number that defines an interval around the measure-
sources. The issue of effective degrees of freedom fre- ment result, y, given by y ± U, that has an approximate
quently confuses beginning uncertainty practitioners 95% level of confidence (i.e., probability) of including

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

Fig. 1 Measurement Uncertainty Quantities (d) The expanded uncertainty is a quantitative state-
ment about our ignorance of the true value of the meas-
Uncertainty interval urand.

Expanded Expanded influence quantity: any quantity, other than the quantity
uncertainty uncertainty being measured, that affects the measurement result.
Uk = 2 Uk = 2 Constructing the list of influence quantities is one of the
first steps of an uncertainty evaluation. This list includes
not only obvious sources of influence such as the uncer-
tainty in the value of a reference standard, or the value
of a force setting on an instrument, but also nuisance
quantities such as environmental parameters or gauge
contamination (dirt). (See Nonmandatory Appendix C.)
input quantity: a specific “line item” in the uncertainty
budget that represents one or more influence quantities
combined together into one quantity. That is, all signifi-
True value Error Measured value
cant influence quantities must be included (i.e., “pack-
aged”) in some input quantity. Different uncertainty
GENERAL NOTE: Figure 1 illustrates the uncertainty interval of width budgets developed by different metrologists might use
2U centered about the result of a measurement. There is a probability different input quantities, but all budgets include (in

of about 95% that the true value of the measured quantity lies in some input quantity) all the significant influence quanti-
this interval. The true value and hence the error are unknown; the
error shown in the figure is among an infinite number of possible
ties. The selection of the input quantities is usually based
values. The subscript k p 2 indicates that U has been calculated on the type of the data available about the influence
with a coverage factor of two. quantities. For example, if a long-term reproducibility
study using a check standard has been conducted
(e.g., measuring the same feature on a gauge once a
week, for several years), then the effects of many influ-
the true value of the quantity we are measuring. (In ence quantities such as temperature, different operators,
certain advanced applications of measurement uncer- recalibration of the instrument, and other factors, are all
tainty it may be necessary to have a different level of combined in the observed variation of the check stan-
confidence or even an asymmetric uncertainty interval; dard results. In this example, a very large number of
these topics involve modifying the coverage factor and influence quantities are combined into a single input
are beyond the scope of this document; refer to the quantity (i.e., the reproducibility of the check standard
GUM.) The expanded uncertainty is the end product results).1
of an uncertainty evaluation. In this document, unless
correlation: refers to a relationship between two input
otherwise stated, the term “measurement uncertainty”
quantities. Correlation between two input quantities
is considered to be the expanded uncertainty with a
means that these two quantities are not completely inde-
coverage factor of 2. (The issue of the coverage factor
pendent. One way in which input quantities can be cor-
will be discussed later.) Several aspects of measurement
related is that the same influence quantity can appear
uncertainty are described below.
in both input quantities. In this case the same influence
(a) Measurement results have uncertainty; measure-
quantity has the risk of being “double-counted.” In
ment instruments, gauges, and workpieces are sources
advanced uncertainty budgets this issue is addressed by
of uncertainty. For example, measuring the diameter of
calculating correlation coefficients and then the effect of
a steel ball using a caliper will generally have smaller the double counting is subtracted. In this document a
uncertainty than when measuring a foam rubber ball, more modest approach is suggested, namely that input
even though it involves the same instrument. quantities should be constructed such that an influence
(b) The expanded uncertainty, U, is always a positive quantity appears in only one input quantity.
number, and the uncertainty interval around a measure-
ment result is of width 2U. (See Fig. 1.) EXAMPLE: Suppose that gauge blocks are calibrated using a set
of master gauge blocks similar to the blocks under calibration.
(c) The expanded uncertainty (using the GUM proce- Suppose further that the laboratory’s temperature slowly varies
dures for evaluating uncertainty) is a statement of belief by ±1°C about 20°C and that no correction is made for the thermal
about the accuracy of a measurement result. When addi- expansion of either gauge block. A poor way to model the measure-
tional information becomes available the uncertainty is ment is to employ a separate input quantity for the temperature,
likely to be re-evaluated yielding a new value. Conse-
quently, there is no “true” or “correct” uncertainty value, 1
As will be described later, the variation captured by a reproduc-
only a statement of belief that is based on the informa- ibility study can be quantitatively evaluated by a “Type A” evalu-
tion available at the time the uncertainty is evaluated. ation.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

Tm, of the master block and for the temperature, Tc, of the custom- rotations that may be applied).3 Note the generic nature
er’s block under calibration. These two input quantities are strongly of this measurand, which avoids specifying any details
correlated. This is easily shown by asking the question, “If I knew
about potential experimental measurement setups.
for sure that Tm > 20°C, would such knowledge tell me anything
about Tc?” In this case the answer is affirmative (i.e., I would know
Unless careful consideration is given to the measurand,
that Tc > 20°C, because the blocks are similar and share the same different inspection techniques can lead to significantly
thermal environment). Indeed Tm ≈ Tc and the two input quantities different results. For example, when measuring a bore,
are fully correlated. This correlation can be completely removed a two-point diameter as measured with a micrometer,4 a
by the observation that both blocks will have the same temperature. least squares fit diameter as measured with a coordinate
Thus there is only a single temperature, T, associated with both measuring machine,5 and a maximum inscribed diame-
blocks, and all that is known is that T p 20°C ± 1°C. Hence, the
ter as found using a plug gauge, may each yield a differ-
correlation is removed by eliminating a redundant uncertainty
source. ent numerical value because each quantity realized by
a particular measurement method measures a different
measurand: the particular quantity subject to measure- measurand. No amount of improvement in the accuracy
ment. It is defined by a set of specifications (i.e., instruc- of these measurement methods will cause their results to
tions) that specifies what we intend to measure; it is not converge as they are fundamentally measuring different
a numerical value. It represents the quantity intended to quantities (two point, least squares, and maximum
be measured. It should specify, as generically as possible, inscribed diameters). The metrologist must recognize
exactly the quantity of interest, and avoid specifying that the two-point and least squares results are not the
details regarding experimental setups that might be measurand under consideration in this inspection, and
used to measure the measurand. For example, measur- differences must be accounted for by applying appro-
ands specified by ASME Y14.5 [8], such as the diameter priate corrections to the measurement result or, alterna-
of a feature of size or the concentricity of two bores, do tively, to account for their difference by increasing the
not attempt to describe the measurement procedure in uncertainty associated with the measurement result. In
detail.2 Ideally the measurand should be completely the GUM, reference is made to the quantity “realized”
independent of experimental measurement details so by the measurement system; again this points out that
that different measurement technologies can be used to many measurement systems do not yield a quantity fully
measure the same measurand and get the same result. consistent with the definition of the measurand and that

Indeed, the measurand is an idealized concept and it corrections (or an increase in the uncertainty) are needed
may be impossible to produce an actual gauge, artifact, to bring the results of the measurement into alignment
or instrument exactly to the specifications of the measur- with the definition of the measurand.
and. Consequently, a well-specified measurand provides A complete definition of the measurand will, in the
enough information, and is generic enough, to allow general case, allow corrections to be applied for different
different techniques to be used to perform the measure- measurement methods. For example, the calibration of
ment. The more completely defined the measurand, the a chrome-carbide gauge block using a gauge block com-
less uncertainty will (potentially) be associated with its parator and a steel master requires the correction for
realization. A completely specified definition of the the differential mechanical penetration of the probe tips
measurand has associated with it a unique value, and since the length of the block is defined as the unde-
an incompletely specified measurand may have many formed (i.e., unpenetrated) length.6 The use of appro-
values, each conforming to the (incompletely defined) priate corrections will allow convergence of the results
measurand. The ambiguity associated with an incom-
pletely defined measurand results in an uncertainty con- 3
tributor that must be assessed during the measurement In this example, it is assumed that no additional control for
orientation or location is specified for the bore and that the
uncertainty evaluation. ASME Y14.5 “Rule #1” is in effect.
As an example of the significance of the measurand, 4
A “two-point diameter” of a cross-section (defined as the “actual
consider a bore that has a size tolerance specified by local size” in ASME Y14.5-1982) is an ambiguous measurand since
ASME Y14.5. An inspection of the workpiece involves it is a one-dimensional length and different cross-sections will in
general yield different two-point diameters. For this reason, the
a measurand defined as the diameter of the maximum 1994 revision of ASME Y14.5 and the associated ASME Y14.5.1
inscribed cylinder that will just fit in the bore (i.e., this standard redefined this measurand to be a two-dimensional
is the largest diameter cylinder that is constrained by quantity.
the workpiece surface, regardless of any translations or A measurand defined as a least squares diameter fit has an
unambiguous value when computed from an infinite number of
perfectly known points distributed around the workpiece surface.
The nominal value that may be attributed to a measurand In practice the diameter will be measured with a finite number of
(e.g., the diameter of a feature of size) is not part of the measurand; imperfectly known points. The effects of finite sampling and errors
rather, it is the desired result of a measurement of the measurand. in the sampled points are uncertainty sources associated with that
Similarly, a tolerance associated with a feature is not part of the particular measurement.
measurand, but rather describes a region within which a measure- The calibration of gauge blocks made of the same material as
ment result is considered to demonstrate the feature to be in confor- the master generally do not require a penetration correction since
mance with the design intent. the deformation is the same on both blocks and hence cancels out.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

from different measurement methods and bring them that might have been made. Several aspects of an error
into accordance with the definition of the measurand.7 are described below.
Hence the methods divergence problem is actually a (a) An error is a quantitative statement about the dif-
problem with an incompletely specified measurand or ference between the measured value and the true value.
the failure to recognize that the measurement method (In contrast, an uncertainty statement expresses our
is not measuring the intended measurand. ignorance about the true value.)
The definition of the measurand must also be suffi- (b) Errors are typically measured only during the spe-
ciently complete to avoid improper use of the calibrated cial case of a calibration when calibrated artifacts (repre-
artifact or instrument. For example, consider a handheld senting the true value) are measured. [The measurement
micrometer that is calibrated for measuring workpieces of an uncalibrated object (e.g., a workpiece) is described
with flat and parallel surfaces by measuring several cali- using a measurement uncertainty statement.]
brated gauge blocks (with surfaces larger than the (c) Errors may be either positive or negative (in con-
micrometer anvil size). Hence the micrometer is cali- trast to uncertainty, which is always positive).
brated for this particular measurand; this means that For a measurement corrected for all known significant
a measurement result of this measurand has a known systematic effects, the best estimate of the error is zero.
expanded uncertainty provided the measurement is per- The set of reasonably probable errors is contained in an
formed within some set of conditions under which the interval centered on an error value of zero with its width
uncertainty statement is valid.8 This procedure does not equal to twice the expanded uncertainty.
calibrate the micrometer for measuring ball diameters Consider, for example, a measurement of the length,
because the flatness and parallelism of the anvils are L, of a block that yields the result L p 25 mm ± 10 ␮m
unknown and are significant influence quantities for the with a 95% level of confidence. It would be incorrect to
(ball diameter) measurand. say, for example, “The measurement error is equal to
Included in the definition of the measurand is a 10 ␮m,” but one could say, “The measurement error is
description of the set of conditions that specify the val- 0 ± 10 ␮m with a 95% level of confidence.” The best
ues of particular influence quantities relevant to the terminology, however, would be to state that the best
measurand. An obvious example is that the length of a estimate of the value of the measurand, with all correc-
gauge or workpiece is defined at 20°C (68°F); otherwise, tions applied, is 25 mm, with an expanded uncertainty
an object would have multiple “true lengths” depending of 10 ␮m.
on the temperature at the time of measurement. Typi- In the special case of an instrument calibration, well-
cally, the higher the accuracy requirements, the more calibrated artifacts are measured and the difference
extensive the list of specified influence quantities in between the measured and calibrated value is a good
order to have negligible uncertainty associated with the estimate of the error. For example, the calibration of a
definition of the measurand. Note that definition of the caliper using a series of gauge blocks uses the calibrated
measurand must address all significant conditions gauge block to represent a true value that is compared
(i.e., influence quantities, not just environmental with the caliper reading to reveal the instrument’s error.
conditions). The measured error will have an associated (usually
error: defined in a measurement result is the measured very small) uncertainty associated with the calibration of
value minus a true value of the measurand (VIM 3.10). the gauge block (i.e., this small uncertainty is associated
(See Fig. 1.) Strictly speaking, the error of a measurement with the “true value” of the block’s length).
result is never exactly known since the value of the systematic error: the (mathematical) expectation value of
measurand is never exactly known. Just as there is a the error. In an instrument calibration, it can be esti-
set of reasonably probable true values that could be mated by the mean (i.e., average) error in the reading of
attributed to the measurand after performing a measure- the measuring instrument when repeatedly measuring
ment, there is also a set of reasonably probable errors a calibrated standard. Similar to the case of error, the
systematic error is never exactly known because we
7 never know the “true value” and we cannot perform
In some cases a metrologist will deliberately choose (for econ-
omy or convenience) to measure a related quantity that differs an infinite number of measurements of a standard to
from the measurand (e.g., a least squares diameter instead of a produce the expectation value. The estimated systematic
maximum inscribed diameter). In this case an estimated systematic error may be determined from the mean of a series of
error may result; this must either be corrected or accounted for in
the uncertainty statement of the measurement.
repeated measurements or as a calculated value corres-
The set of conditions under which an uncertainty statement is ponding to a known systematic effect. If a systematic
valid is known as the validity conditions. This may restrict such error is known to be present in the measurement result,
parameters as the measurement temperature range, the amount then either a correction must be applied or it must be
of force applied, the quality of the workpiece surfaces, the type of accounted for in the uncertainty budget. (A correction
material measured (e.g., a dimensional measurement of a work-
piece made of foam rubber may have a higher uncertainty than has the same magnitude as the systematic error but the
one of steel). opposite sign.)


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

standard uncertainty: a quantitative value describing the In other cases a Type B evaluated standard uncertainty
magnitude of an uncertainty source. Each uncertainty involves deciding on the “probability density function”
source must be evaluated to yield its associated standard of the source; this is just a fancy way of stating how likely
uncertainty. The standard uncertainty can be thought of the uncertainty source might yield one value relative
as one standard deviation of potential variation associ- to another value. In this document several probability
ated with that uncertainty source. Each and every uncer- distributions will be used in the examples, including
tainty source, as represented as an input quantity to the the uniform distribution, the normal (or Gaussian) dis-
uncertainty budget, has an associated standard uncer- tribution, and the triangular distribution. The only infor-
tainty that is combined with others in order to produce mation needed to define a uniform distribution, for
the final uncertainty statement. example, are the upper and lower bounds of the possible
values of the uncertainty source. All other values that
combined standard uncertainty, uc: the result of combining occur between the two limits are considered to be
all of the standard uncertainties of the various uncer- equally likely. Nonmandatory Appendix B presents
tainty sources. The method of combining these sources some guidance on selecting the appropriate distribution
will be discussed later, but uc can be thought of as one as well as the details of calculating the standard uncer-
standard deviation’s worth of variation about the mea- tainty when using these distributions. While advanced
surement result due to all of the uncertainty sources uncertainty practitioners might employ many different
included in the uncertainty budget. In order to increase distributions, generally the benefits of this additional
the level of confidence that the uncertainty includes all fidelity are typically a small refinement to the numerical
reasonable values of the measurand, the combined stan- value of the expanded uncertainty.
dard uncertainty is multiplied by a “coverage factor,”
specified validity conditions: the conditions under which
which will always be equal to 2 in this document. Hence,
the results of the uncertainty statement are valid. For
the expanded uncertainty is just twice the combined
example, an industrial metrologist may wish to use a
standard uncertainty (i.e., U p 2uc). If the uncertainty
calibrated artifact at nonstandard thermal conditions
budget is well constructed such that it includes all rele-
(e.g., 23°C), and desires an expanded uncertainty that
vant uncertainty sources and these sources are well eval-
includes these nonstandard conditions (i.e., the metrolo-
uated, then generally there will be approximately a 95%
gist does not want to develop a new uncertainty state-
level of confidence (or probability) that the value of
ment for the nonstandard condition but rather wishes
the measurand is within the uncertainty interval of the
to refer directly to the calibration certificate to obtain
measurement result. (See Fig. 1.) the uncertainty relevant to the conditions of use). These
Type A and Type B uncertainty evaluations: designations conditions, which we will call the validity conditions,9
that refer to the method used to evaluate an uncertainty include the values (or range of values) of all significant
source. The GUM specifically avoids referring to uncer- influence quantities for which the uncertainty results
tainty sources as “random” or “systematic.” Such a clas- are valid. In the case of instruments, the validity condi-
sification is artificial since given more time and money tions also include the number of measurements used to
an apparently random source can be tracked down to compute a result, because if repeated measurements by
its systematic causes. Instead, the GUM focuses on the an instrument yield different results, then the mean
method of evaluation of the uncertainty source. (mathematical average) result will usually have a
Type A evaluations assign the standard uncertainty smaller uncertainty than a single result.
of an uncertainty source to be equal to the standard The validity conditions are either those specified in
deviation calculated based on repeated observations of the definition of the measurand (i.e., the “measurand
experimental data. Calculating the standard deviation defining conditions,” or are “extended conditions”). For
of experimental data is a simple and straightforward stating the uncertainty of a calibration, which typically
activity; see Nonmandatory Appendix A for details. involves such artifacts as master gauges and reference
Type B evaluations are based on anything other than standards, the validity conditions are often identical to
repeated observations (i.e., they involve more than just the measurand defining conditions. For example, the
numerically evaluating experimental data). Often this results of a NIST gauge block calibration are valid only
at exactly 20°C, which is the standard reference tempera-
information is from handbooks, manufacturer’s specifi-
ture for the length of a gauge block. Although the NIST
cation sheets, or just educated guesses. Whatever the
laboratory cannot actually realize the measurand defin-
source, the information must be converted into a stan-
ing conditions during the calibration (i.e., the prevailing
dard uncertainty. In some cases this conversion is trivial.
conditions during the calibration are never exactly 20°C),
For example, the standard uncertainty associated with
corrections for deviations from the measurand defining
the calibration value of a master artifact is easily
obtained from the expanded uncertainty (that would 9
The term “validity conditions” is used in order to avoid confu-
be typically stated on a calibration report) by simply sion with the conditions that happen to prevail at the time of the
dividing by two (the coverage factor). measurement.

5 --`,,```,,,,````-`-`,,`,,`,`,,`---

Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

conditions (e.g., not exactly at 20°C) are applied and their of the uncertainty statement. That is, it describes the set
associated uncertainties are included in the uncertainty of conditions under which the uncertainty statement is
budget of the calibration. Hence the calibration report valid; this is a critical piece of information for the user
states the best estimate of the block length and its uncer- of the uncertainty statement. Note that the validity con-
tainty at exactly 20°C. ditions of the uncertainty statement have no effect on
Subsequent use of these standards (e.g., in calibrating the definition of the measurand; they merely restrict or
other artifacts) will generally not be at the validity condi- extend the set of conditions under which the user may
tions (i.e., not exactly at the measurand defining condi- employ the results of the measurement with the stated
tions). Hence the metrologist is obligated to develop an uncertainty. Considering any particular measurement,
uncertainty budget that includes not only the uncer- the input quantities for an uncertainty evaluation with
tainty stated in the calibration report of the reference extended validity conditions will have larger standard
artifact, but also any failure to exactly realize the measur- uncertainties than those of a corresponding uncertainty
and defining conditions of the reference artifact during statement applicable only for the measurand defining
subsequent calibrations that use the reference artifact as conditions. Hence, the expanded uncertainty having
the “master.” Thus the uncertainty of each subsequent extended validity conditions will be larger than that of
calibration in a traceability chain will be greater than the corresponding case that is limited by the measurand
the uncertainty of the previous calibration since the defining conditions.
measurand defining conditions generally cannot be fully The rows of the table describe the conditions that
achieved. The benefit of having the validity conditions prevail during the measurement. The special test sce-
identical to those of the measurand defining conditions nario describes a specific measurement of a specific
is that it can yield the smallest uncertainty statement. object and the associated evaluation of the measurement
The disadvantage is that any subsequent use requires uncertainty. The multiple measurement scenario
the evaluation of a new uncertainty budget. (See para. 6.1 describes an ongoing measurement process where many
for an example of developing an uncertainty statement measurements will be performed now and in the future
for validity conditions constrained to be those specified and a single uncertainty statement is assigned to each
in the measurand defining conditions.) of these measurement results. For example, consider a
In contrast, industrial measurements often involve calibration lab that continuously calibrates gauge blocks
“extended validity conditions” that are appropriate for all year long; an uncertainty budget is developed once
their particular needs. These conditions may differ sig- and this uncertainty is attached to each of the future
nificantly from those specified in the measurand defin- measurement results. In both the special test and multi-
ing conditions. For example, a factory floor worker using ple measurement scenarios, sources of uncertainty must
an instrument may not want to develop an uncertainty be evaluated and corrections may need to be applied to
budget for every measurement performed. What may be bring the measurement results into alignment with the
desired is a calibration report that states an uncertainty definition of the measurand. The difference between the
under validity conditions that include the conditions of two cases involves the magnitude of the uncertainty
actual use. A common example is a voltmeter calibration sources and corrections. In the multiple measurement
that gives an uncertainty statement over a range of ambi- scenario the range of variation of each influence quantity
ent temperatures. (See para. 6.2 for an example of devel- over the entire time of the measurement process must
oping an uncertainty statement for extended validity be considered. For example, if the uncertainty statement
conditions.) of a gauge block calibration process is to be applicable
The calibration of an instrument or artifact under all year long, then variations that occur over that time
extended validity conditions must have its errors and period must be included. This will include, for example,
uncertainties assessed over this range of conditions. As seasonal temperature variations, multiple operators,
with the definition of the measurand, specifying the drift in the measurement system, etc. In general, the
extended validity conditions involves stating the permit- standard uncertainties evaluated for a multiple mea-
ted values of any influence quantity that affects the mea- surement scenario will be larger than the corresponding
surement. Uncertainty budgets that are associated with standard uncertainties for a special test scenario.
extended validity conditions typically have much larger
expanded uncertainties than those associated with
measurand defining conditions, since the effects of the
extended conditions must be included in the uncertainty Once all the influence quantities are known and
evaluation. included in some input quantity, the uncertainty compo-
Table 1 may further clarify the difference between the nents associated with each input quantity can be evalu-
validity conditions of the uncertainty statement and the ated (in the units of the measurand, i.e., length). These
conditions that happen to prevail at the time of measure- uncertainty components are then combined to yield the
ment. The columns of the table specify the applicability combined standard uncertainty, uc.


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

Table 1 Measurement and Validity Conditions

Prevailing Conditions at Validity Conditions of the Uncertainty Statement
the Time of Measurement Measurand Defining Conditions Extended Validity Conditions

Special test scenario: conditions limited to U1: the smallest uncertainty for a U3: U3 > U1 typically U3 > U2
those at the time of the measurement particular measurement result
Multiple measurement scenario: conditions U2: typically U2 > U1 U4: the largest uncertainty for a particular
varying over the time of all measurements measurement result

In the common case where all input quantities are Step 1: Define the quantity to be measured (the meas-
assumed to be uncorrelated, the method of combining urand).
uncertainty components is called a root sum of squares Step 2: State the desired validity conditions of the
(RSS) calculation, given by the following simple uncertainty statement.
expression: Step 3: Define the measurement method, equipment,
and environment.
uc p 冪u21 + u22 + . . . + u2N (1) Step 4: List the influence quantities.
Step 5: Determine the input quantities (select an
uncertainty evaluation plan).
uc p the combined standard uncertainty
Step 6: Evaluate and rank the input quantities (deter-
u1, u2,...,uN p the standard uncertainty components
mine the standard uncertainties).
associated with the N input quan-
Step 7: Combine the input quantities (combined stan-
dard uncertainty).
The expanded uncertainty, U, is stated as just twice Step 8: State the expanded uncertainty and the cover-
the combined standard uncertainty: U p 2uc. age factor used.
In actual practice, a metrologist may have determined
NOTE: The use of the uncertainty statement in a decision rule
that in order to have an acceptable level of risk11 during
regarding the acceptance or rejection of workpieces is discussed
conformance testing the expanded uncertainty must be in [3]. The analysis of the risk (accepting a nonconforming work-
no greater than a specific value. If the expanded uncer- piece or rejecting a conforming workpiece) associated with a deci-
tainty evaluated is greater than this value the metrologist sion rule is discussed in [5]. The evaluation of the reliability of
must consider which input quantities can economically uncertainty statements is discussed in [4].
have their standard uncertainties reduced in magnitude.
Due to the manner in which uncertainty sources com-
bine, given in eq. (1), the largest uncertainty source is 6 EXAMPLES
usually the target of this reduction since it will yield the
6.1 Calibration of Gauge Blocks by Mechanical
greatest reduction in the expanded uncertainty. Gener-
ally, once the expanded uncertainty has been reduced
to yield an acceptable amount of risk, it is economically A small calibration laboratory calibrates steel gauge
unprofitable to make further efforts to lower the evalu- blocks of lengths from 1 mm to 100 mm in an environ-
ated expanded uncertainty. ment of (21°C ± 1°C), and the master and customer
blocks are always within ±0.1°C of each other. An uncer-
tainty statement for the length of the gauge block with

validity conditions to be those specified by the measur-

and defining conditions is required. This is a multiple
Having discussed the general concepts of measure- measurement scenario since the uncertainty statement
ment uncertainty the basic procedure for producing an will be applied to many future gauge block measure-
uncertainty statement will now be discussed. For pur- ment results. (This example is intended to be a “crude”
poses of this document the evaluation will be considered calibration; for a high accuracy example of mechanical
in eight steps. comparison of gauge blocks, see [9].)

Due to the RSS manner by which the uncertainty components
6.1.1 Measurand. The measurand is the length of
are combined, the largest components dominate the combined the gauge block as defined by B89.1.9 [10]. This defini-
standard uncertainty. For example, consider two components: u1 p tion is based on an interferometric length measurement
1 ␮m and u2 p 5 ␮m; u2 will have 25 times more significance than at 20°C; fortunately, National Measurement Institutes
u1 in the combined standard uncertainty.
11 also provide gauge block calibrations based on mechani-
The issue of risk analysis (i.e., the probability of rejecting a
conforming workpiece or accepting a nonconforming workpiece cal comparison so the transfer of the interferometrically
during an inspection) is extensively discussed in [5]. based definition to a mechanical (point-to-point) length

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

is already included in the calibration report of the master between and within blocks (it is assumed that the ther-
blocks. mal properties of the master, customer, and check stan-
dard blocks are all similar since they are all steel), block
6.1.2 Uncertainty Validity Conditions. Since the cus- geometry effects, comparator’s transfer ability, and scale
tomer intends to use the gauge blocks as their master calibration because the check standard data includes
artifacts, the validity conditions of the uncertainty will several scale recalibrations. (In this example, we assume
be the same as the measurand defining conditions the comparator is routinely recalibrated.)
defined by B89.1.9; this yields the smallest possible
(b) Master Block Length
uncertainty statement given the conditions of the cali-
(c) CTEs of the Master and Customer Blocks
6.1.6 Evaluate Input Quantities
6.1.3 Measurement Method and Environment. The (a) Reproducibility. Using historical data for each
measurement method will be by mechanical comparison length of check standard block, the standard deviation
to a master gauge block in an environment of 21°C ± 1°C of the data is computed for each check standard block.
(over all measurements) and is homogeneous in the mea- The standard deviations from each check standard are
suring volume. The calibration includes “check stan- pooled as described in Nonmandatory Appendix A. The
dards” that are steel gauge blocks measured together standard uncertainty associated with reproducibility is
with the customer’s blocks as part of the quality assur- evaluated as the pooled standard deviation. Pooling the
ance system. The check standard plays the role of a standard deviations has the advantage that it includes
surrogate customer block for quality control purposes. several different block geometrical effects (e.g., flatness),
Data on the check standards from many years is avail- and these effects are quantified in the pooled standard
able and hence includes the long-term variation of many deviation.
influence quantities. Since the master, check standard,
EXAMPLE: For brevity, only 20 measurements per block and only
and customer’s blocks are all steel it is assumed that three check standard blocks will be considered. In a more complete
they have the similar temperatures, CTEs, and values example, more check standards might be used and the data could
of elastic modulus. be segregated based on length (e.g., all check standards between
1 mm and 30 mm in length would be pooled, allowing possible
6.1.4 Influence Quantities. Referring to Nonmanda- different standard uncertainties associated with block length).
tory Appendix C, the following influence quantities are Additionally, for quality control purposes, the check standards
relevant: could be calibrated (or at least the current measurement value
(a) master gauge block length uncertainty could be compared to the historical mean value) as a means of
detecting blunders. (See [9] for further details.)
(b) coefficient of thermal expansion (CTE) of the mas- (a) Check Standard 1 Data (Deviations From Mean Value in Microme-
ter block ters). 0.174467, 0.175691, 0.0852917, −0.0576673, −0.110542,
(c) CTE of the customer’s block 0.235682, 0.0954639, −0.283445, 0.0302104, −0.217087, 0.0156588,
(d) temperature of the master block 0.248976, 0.134756, 0.315471, 0.0509718, 0.0263296, −0.0868594,
0.203901, 0.00287366, −0.287881
(e) temperature of the customer’s block
Standard deviation p 0.17 ␮m
(f) thermal gradients between and within the blocks (b) Check Standard 2 Data (Deviations From Mean Value in Microme-
(g) operator effects ters). 0.103049, 0.246114, 0.0275914, 0.0778894, −0.0658711,
(h) calibration of the indicator on the gauge block 0.0374774, −0.0417286, 0.172826, 0.244662, 0.130969, 0.0290974,
comparator −0.172633, 0.0160234, 0.0659565, −0.200082, 0.173218, 0.111646,
0.172801, 0.0157569, 0.0378071
(i) comparator’s length transfer ability
Standard deviation p 0.12 ␮m
(j) master block geometry and modulus of elasticity (c) Check Standard 3 Data (Deviations From Mean Value in Microme-
(k) customer’s block geometry and modulus of elas- ters). 0.0128391, 0.0884897, 0.348164, 0.0382762, 0.0250351,
ticity −0.170214, −0.0939044, −0.246856, −0.134731, −0.113879,
0.00409127, −0.104183, 0.139396, −0.243031, 0.152802, −0.127018,
(l) check standard’s block geometry and modulus of
0.212681, 0.307211, 0.0618147, −0.0564749
elasticity Standard deviation p 0.17 ␮m
(m) cleanliness/contamination (d) Pooled standard deviation p
(n) resolution of the comparator

19 ⴛ (0.17)2 + 19 ⴛ (0.12)2 + 19 ⴛ (0.17)2
␮m p 0.16 ␮m
6.1.5 Input Quantities. Choosing the input quanti- 20 + 20 + 20 − 3
ties is guided by the availability of data and the measure- NOTE: Although repeated measurements are used in the uncer-
ment method. tainty evaluation, the resulting standard uncertainty (0.16 ␮m in
(a) Reproducibility. Creating an input quantity called this example) is applicable to a future measurement result based
on a single measured value.
reproducibility, based on the check standard data, is
advantageous as it include the effects of many (nuisance) (b) Master Gauge Calibration. From the certificate the
influence quantities including cleanliness/contamina- expanded (k p 2) uncertainty is 0.14 ␮m; hence, the
tion, multiple operator effects, thermal gradients standard uncertainty is 0.07 ␮m.


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

(c) Thermal Effects. From para. D-3 of Nonmandatory negligible, the difference between the measurand real-
Appendix D, the standard uncertainty component due ized by the caliper and that of interest (actual external
to thermal effects equals 0.17 ␮m.

size) is also negligible. The size of an object is defined

at 20°C, and with zero contact deformation (i.e., in its
6.1.7 Evaluate the Combined Standard Uncertainty.
free state).
The combined standard uncertainty is
6.2.2 Uncertainty Validity Conditions
uc p 冪0.162+ 0.072 + 0.172 ␮m p 0.24 ␮m
Caliper CTE: (11.5 ± 1) ⴛ 10−6/°C
6.1.8 State the Expanded Uncertainty. The Workpiece CTE range: 1 ⴛ 10−6/°C to 22 ⴛ 10−6/°C
expanded uncertainty and associated coverage factor Temperature range: 15°C to 25°C
(k p 2) is Temperature difference: −0.2°C to +0.2°C
Material hardness: All common metallic engi-
U p 2 ⴛ 0.24 ␮m p 0.48 ␮m neering materials
NOTE: A more detailed uncertainty evaluation may result in an Workpiece size: 0 mm to 100 mm continuous
uncertainty statement that depends on the length of the block Workpiece geometry: All geometries (planar, cylin-
under measurement. drical, etc.)
Workpiece form error: Negligible
6.2 Uncertainty of Shop Floor Measurements Using
a Caliper 6.2.3 Measurement Method and Environment.
Lengths are measured by hand using a calibrated caliper.
A metrologist is interested in evaluating the uncer- Data consists of a single caliper reading. The shop tem-
tainty of measurements made on the shop floor using perature varies between 15°C and 25°C.
a steel caliper (i.e., the uncertainty statement shall have
extended validity conditions including the conditions 6.2.4 Influence Quantities
on the shop floor).12 (a) calibration of caliper
(b) CTE of the caliper
NOTE: In this example, some of the validity conditions involve
workpiece material properties. (c) CTE of the workpieces
(d) temperature of caliper
The caliper has a calibration certificate stating that (e) temperature of the workpieces
the maximum permissible error (MPE) is less than 10 ␮m (f) thermal gradients
over its full range when measuring at 20°C. The shop (g) operator effects
floor measurand of interest is the actual external size at (h) geometry and modulus of elasticity of workpieces
a specified cross-section of the workpiece. The metal (i) cleanliness/contamination
workpieces have a variety of rectangular and cylindrical
(j) resolution of the caliper
shapes up to 100 mm in size, and consist of materials
(k) parallelism and flatness of the caliper anvils
with CTEs between 1 ⴛ 10−6/°C and 22 ⴛ 10−6/°C. The
shop temperature varies between 15°C and 25°C, and 6.2.5 Input Quantities. The effects of workpiece
the temperatures of the caliper and the workpiece are form error and modulus, operator variability, and con-
assumed to be within 0.2°C of each other during a mea- tamination are judged to be negligible compared to the
surement. effects of differential thermal expansion and caliper cali-
It is desired to have an uncertainty statement for a bration at 20°C.
single future measurement with extended validity con- The input quantities then consist of the following:
ditions covering the shop environment without applying (a) reading of the caliper with a resolution of 10 ␮m
corrections. The form error on the workpiece surface is (b) calibration of caliper
known to be small relative to the resolution of the caliper. (c) differential thermal expansion
6.2.1 Measurand. The caliper does not realize the (d) anvil parallelism and flatness
measurand of interest; rather it measures a point-to- 6.2.6 Evaluate Input Quantities
point distance (on cylindrical workpieces) or a line-to- (a) Caliper Resolution. The resolution (magnitude of
line distance (on rectangular workpieces). However, the last digit in the display) of the caliper is 10 ␮m.
since the form error of the workpiece is known to be Assigning a Type B uniform distribution of width 10 ␮m
(±5 ␮m) yields a standard uncertainty of 2.9 ␮m.
This example is a special test scenario; however, since the (b) Caliper Calibration. Since the calibration report of
extended validity conditions appropriate for the shop floor (includ- the caliper at 20°C states that the maximum permissible
ing different caliper operators, different workpiece materials, broad error (MPE) is 10 ␮m over the full range of travel and
and uncorrected thermal condition, etc.) result in significant uncer- there is no information about the actual calibration error
tainty contributions, a multiple measurement scenario (calibrating
many calipers for the shop floor) would likely have a similar uncer- in any particular measurement, the best estimate of this
tainty. error is zero, with an uncertainty evaluated as follows.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

Assigning a Type B uniform distribution of width 20 ␮m (see Nonmandatory Appendix A) equal to 4.5 ␮m. The
(±10 ␮m) yields a standard uncertainty of 5.8 ␮m. best estimate of the correction for these effects in any
NOTE: In this example, the calibration result was stated as a
particular workpiece measurement is zero due to lack
limiting value (i.e., an MPE). Had the calibration result been stated of specific information about the position and orienta-
as an expanded uncertainty with k p 2, then the standard uncer- tion of the anvils with respect to the measured feature.
tainty would be one-half the stated expanded uncertainty.
6.2.7 Evaluate the Combined Standard Uncertainty.
(c) Differential Thermal Expansion. For any particular The combined standard uncertainty is
measurement, the best estimate of the differential expan-
sion is zero since the average temperature on the shop uc p 冪2.92 + 5.82 + 2.32 + 0.12 + 4.52 ␮m p 8.2 ␮m
floor is 20°C and the average workpiece CTE is equal
to the CTE of the caliper. There is, however, a component
6.2.8 State the Expanded Uncertainty. The
of uncertainty evaluated as described in para. D-3 of
expanded uncertainty is U(k p 2) p 2 ⴛ 8.2 ␮m p
Nonmandatory Appendix D.
16.4 ␮m; valid for measurements of workpieces up to
Assuming a steel caliper with a CTE of 10.5 ⴛ 10−6/°C
100 mm in length, of any geometry, made of common
to 12.5 ⴛ 10 −6 /°C, the maximum difference, ⌬ ␣ max ,
metallic engineering materials with a CTE between 1 ⴛ
between the caliper and workpiece CTEs is 11.5 ⴛ
10−6/°C to 22 ⴛ 10−6/°C, measured within a temperature
10−6/°C. The maximum temperature deviation, ⌬Tmax,
range of 15°C to 25°C.
is 5°C, and the maximum measured length, L max , is
In this problem the temperature of the shop floor
100 mm. Then the possible values of the differential
happened to be symmetrically distributed about 20°C
thermal expansion lie in the interval [see eq. (D-15)]
(i.e., 15°C to 25°C). Had the temperature been biased
±L max ⌬ ␣ max ⌬T max p ±(0.1 ⴛ 11.5 ⴛ 10 −6 ⴛ 5)m p
away from 20°C (e.g., 18°C to 28°C) then, on average,
±5.8 ␮m. Assigning a Type B triangular distribution to
we would expect the measured length to be longer than
this possible error then yields a standard uncertainty of
the 20°C value, resulting in a systematic error. In the
5.8 ␮m/冪6 p 2.3 ␮m.
case of 18°C to 28°C, the magnitude of the expected
The maximum error due to a possible temperature
error is L ⴛ 11.5 ⴛ 10−6/°C ⴛ 3°C, where L is the length
difference between the workpiece and the caliper can be
of the measured workpiece. It is recommended that the
evaluated following eq. (D-17). The maximum measured
user apply this correction to the measurement results.
length, Lmax, is 100 mm. The maximum value of the
However, in this example it is stated that the user does
caliper CTE is 12.5 ⴛ 10−6/°C. The temperature differ-
not want to apply any corrections, so the largest length
ence, ␦T p Tm − T, between the caliper and the workpiece
(0.1 m) is assumed, yielding a systematic error of magni-
is assumed to lie in the interval ±0.2°C. Then error due
tude 3.5 ␮m. This value is then added in an arithmetic
to the temperature difference lies in the interval
manner to the previous analysis to yield 16.4 ␮m +
±L max ␣ max ␦ T max p ±(0.1 ⴛ 12.5 ⴛ 10 −6 ⴛ 0.2)m p
3.5 ␮m p 19.9 ␮m as the expanded uncertainty. This
±0.25 ␮m. Assigning a Type B triangular distribution
manner of including the systematic error in an uncer-
to this possible error gives a standard uncertainty of
tainty statement assures that the containment probabil-
0.25 ␮m/冪6 p 0.1 ␮m.
ity is at least as large as that associated with the coverage
NOTE: When an uncertainty source is evaluated and is less than factor (e.g., at least 95% at k p 2) [11]. However, the
10% of another uncertainty source typically it can be neglected. In resulting value is no longer, strictly speaking, an uncer-
this example we continue to include this small value for complete-
tainty (since it contains a known systematic error), and,
ness of the example.
while this procedure may be useful for workpiece con-
(d) Anvil Flatness and Parallelism. These effects are formance decisions, it is not appropriate for the state-
evaluated using a small gauge wire measured in multi- ment of calibration results where the uncertainty
ple positions and orientations. Variation of the results statement will be propagated into subsequent
leads to the assignment of a Type A standard uncertainty measurements.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007


A-1 MEASURES OF DISPERSION A-1.2 Pooled Standard Deviations

Repeated measurements will always have some varia- In some cases multiple sets of observations are avail-
tion between them. The extent of this variation is known able, but these sets of data do not have the same mean.
as the dispersion. There are several methods used to For example, the reproducibility of gauge block mea-
numerically characterize the dispersion. The simplest to surements could be evaluated by examining repeated
use and understand is the range, which is the arithmetic measurements on several different length gauge blocks.
difference between the largest and the smallest readings. Data sets with different means cannot be directly com-
A more commonly used estimate of the dispersion is bined to calculate the standard deviation but they can
the standard deviation. be pooled. To pool the data, the standard deviation of
each data set is first determined and then the standard
A-1.1 Standard Deviation deviations are then combined, weighted by the number
of observations in each set:
By definition, a Type A evaluation of standard uncer-

tainty involves the use of a statistical method on
repeated observations of the same measurand. From 兺 (ni − 1)s2i
sp (A-3)
these repeated observations the sample mean is N
defined by 兺
ni − N

x p ⴛ 兺 xi (A-1) N p the number of data sets
n 1 ni p the number of observations in the ith data set
s p the pooled standard deviation
si p the standard deviation of the ith data set
n p the number of observations
x1,x2,. . .,xn p the individual readings A-1.3 Standard Deviation of the Mean
Frequently, the best estimate of a value is obtained by
Given a set of n repeated observations of a quantity repeated observations and the calculation of the mean
x, the Type A standard uncertainty associated with x is value, x. [See eq. (A-1).] If the uncertainty statement
taken to be the sample standard deviation, defined by pertains to the mean value (as opposed to a single value
as described in para. A-1), then the standard uncertainty

associated with the mean is equal to the standard devia-
sx p 兺 ( x − xi)2
(A-2) tion of the mean, computed by
(n − 1)
For a Type A evaluation, the standard uncertainty u( x) p s x p (A-4)
associated with a single observation of x is equal to the 冪n
calculated standard deviation of eq. (A-2) (i.e., u p sx). where
This standard uncertainty is frequently used when an sx p the sample standard deviation given by
uncertainty statement pertains to future measurements eq. (A-1).
in a multiple measurement scenario. A measurement (in
NOTE: A Type A evaluation can be performed on repeated obser-
the future) is performed only once, and the standard vations regardless of whether the source of variation is from a
uncertainty (equal to sx) describes the dispersion of this systematic or random effect.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007


If repeated observations are available then the stan- Fig. B-1 Uniform Probability Distribution
dard uncertainty of a quantity can be evaluated by a
statistical Type A procedure, as described in Nonmanda-
tory Appendix A. Otherwise a Type B evaluation is
required. A Type B evaluation is based on available rele-
20.8 mm 21.2 mm
vant information; this could be a handbook value, expert 21.0 mm
opinion based on engineering judgment, prior history Best estimate
of similar measurement systems, calibration certificate GENERAL NOTE: Figure B-1 illustrates a measured length of
information, manufacturer’s specification, etc. 21.0 mm, with an uncertainty characterized by a uniform distribution
The essence of a Type B evaluation is assigning a with limits 20.8 mm and 21.2 mm.
probability distribution that describes the likelihood of
the possible values of the quantity. The metrologist must
estimate the likelihood of various values for a particular Fig. B-2 Normal Probability Distribution
input quantity that is assigned a Type B evaluation. That
is, are certain values of the input quantity more likely 95th percentile 95th percentile
to occur than others? How rapidly does the likelihood 2uc 2uc
of a value of an input quantity diminish as its value gets
farther away from the most likely value? What is the
maximum value that this input quantity may obtain?
These questions are answered by the metrologist by
assigning a probability distribution to the input quantity.
This typically involves specifying the shape of the distri-
bution and a measure of its width; once this is done the
standard uncertainty associated with the distribution is
readily evaluated.


The uniform distribution assigns equal probability for
a value anywhere between two limits. Figure B-1 shows
a uniform distribution for the possible values of the
length of a measured workpiece; the uniform distribu-
tion is centered about the best estimate of the value. (In 20.8 mm 21.2 mm
this document, the best estimate will always be at the
21.0 mm
center of the uniform distribution.)
To evaluate the standard uncertainty of a uniform GENERAL NOTE: Figure B-2 displays the uncertainty in a measured
distribution, take the half-width of the distribution and value of 21.0 mm having with a normal distribution and an expanded
(k p 2) uncertainty of 0.2 mm, yielding one standard uncertainty
divide by the square root of 3 (equivalent to multiplying
equal to 0.1 mm.
by 0.58). The standard uncertainty for the distribution
in Fig. B-1 is computed as u(L) p 0.58 ⴛ 0.2 mm p extends out to infinity, albeit with vanishingly small
0.12 mm. probability.
The normal distribution has the property that approxi-
B-2 THE NORMAL DISTRIBUTION mately 95% of its probability is contained within plus-
The normal distribution assigns a higher probability minus two standard uncertainties of its mean (best esti-
around the best estimate of the value than does the mate) value. (See Fig. B-2.) Consequently, if one believes
uniform distribution. The assigned probability decreases that there is a probability of about 95% that the value
as the difference between a possible value and the best of an unknown quantity lies between two limits (X1 and
estimate increases. Unlike the uniform distribution, the X2), and that it is most likely to lie midway between
normal distribution does not have limits; rather it the two limits, then a normal distribution is a reasonable
Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

Fig. B-3 Triangular and “U” Probability Distributions

20.8 mm 21.2 mm 20.8 mm 21.2 mm

21.0 mm 21.0 mm
Best estimate Best estimate

(a) (b)
GENERAL NOTE: Figure B-3 illustrates an example of a triangular distribution [illustration (a)] and a “U” distribution [illustration (b)]; for the
values shown the one standard uncertainty is 0.08 mm and 0.14 mm, respectively.

Table B-1 Various Probability Distributions and Their Standard Uncertainty

Standard Uncertainty Example: From
Distribution “Width” Specified By [Note (1)] Figs. B-1, B-2, and B-3

Triangular Upper minus lower bounding value 0.41 ⴛ 1⁄2 width 0.08 mm
Normal Upper minus lower 95th percentile value 0.50 ⴛ 1⁄2 width 0.10 mm
Uniform Upper minus lower bounding value 0.58 ⴛ 1⁄2 width 0.12 mm
“U” Upper minus lower bounding value 0.71 ⴛ 1⁄2 width 0.14 mm

(1) For rigor in this report, two significant figures are shown for calculating the standard uncertainty of the various distributions. In prac-
tice, a single significant digit is usually sufficient (e.g., using 0.4 instead of 0.41 is sufficient).

assignment with a standard uncertainty of are sufficient in most cases. Two other distributions are
0.5(X2 − X1)/2. shown in Fig. B-3; the triangular distribution has the
The normal distribution is also commonly assigned probability peaked at the center of the interval, while
to the distribution of reasonably probable values associ- the “U” distribution has the probability peaked at its
ated with a calibration report. The expanded uncertainty bounding values with low probability at the mid-value.
of a calibration result is usually (unless otherwise stated [The “U” distribution typically occurs when an input
on the calibration report) two standard uncertainties quantity has a sinusoidal time dependence (e.g., some
associated with a normal distribution. Hence to obtain temperature cycles).] The triangular distribution is often
the standard uncertainty from a calibration report, just used to describe the probability distribution that arises
divide the expanded uncertainty by the coverage factor. from the sum, product, or difference of two uniform
A simple example is shown in Fig. B-2. Similarly, if distributions. Table B-1 summarizes the standard uncer-
a metrologist believes an input quantity has a normal tainties of these probability distributions.
distribution, to evaluate its standard uncertainty the As seen in Table B-1, the normal and uniform distribu-
95th percentile limit of the distribution is estimated and tions yield a standard uncertainty that is roughly mid-
then this value is divided by two to obtain the standard way between the two more extreme distributions. Since
uncertainty. an uncertainty evaluation reduces all the information
about a distribution to a single number (the standard
uncertainty), the difference between using a normal or
uniform versus a more extreme distribution is a factor
For introductory uncertainty evaluations the uniform of roughly 20% and is unlikely to greatly change the
distribution, with its assignment of probability distrib- final combined standard uncertainty.
uted broadly over the range of possible values, and the NOTE: A Type B evaluation can be performed on any uncertainty
normal distribution, with its probability distribution source regardless if the source of uncertainty is from a systematic
peaked at the center of the range of possible values, or random effect.


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007


A large number of potential influence quantities can (i) geometrical imperfections

affect a measurement result. Some potential quantities (j) stiffness/rigidity
are listed in paras. C-1 through C-10. (k) sampling strategy
(l) probe/reading system
C-1 ENVIRONMENT (m) contact geometry
(n) stiffness of probe system
(a) temperature: absolute, time variance, spatial
(o) temperature, CTE, time constants
(p) temperature stability/sensitivity
(b) vibrations/noise
(q) Abbe, Cosine errors
(c) humidity
(r) time since last calibration
(d) contamination
(e) illumination (s) response characteristic
(f) ambient pressure (t) interpolation system
(g) air composition and flow (u) interpolation resolution
(h) EMI (electromagnetic interference) (v) digitization
(i) transients in power supply
(j) pressured air (e.g., air bearings)
(k) heat radiation
(l) gravity (a) Abbe, Cosine errors
(m) instrument thermal equilibrium (b) temperature sensitivity, warm up
(c) stiffness/rigidity/stability
EQUIPMENT (e) form deviation of tip
(f) stiffness of the probe system
(a) stability
(g) optical aperture
(b) scale mark quality
(h) interaction between workpiece and setup
(c) CTE, thermal time constant
(i) warming up
(d) physical principle: line scale, optical digital scale,
magnetic digital scale, spindle, rack and pinion, interfer-
(e) CCD-techniques
(f) uncertainty of the calibration (a) rounding/quantification
(g) resolution of the main scale (analog or digital) (b) algorithms and implementation
(h) time since last calibration (c) significant digits in computation
(i) wavelength error (d) sampling, filtering
(e) validity/certification of algorithm
(f) interpolation/extrapolation
C-3 MEASURING EQUIPMENT (g) outlier handling

(a) interpretation system

(b) magnification stability C-6 METROLOGIST
(c) wavelength error
(d) zero-point stability (a) heat source (breath, radiation)
(e) force stability/absolute force (b) physical ability
(f) hysteresis (c) experience, dedication
(g) guides/slideways (d) education, training, knowledge
(h) stylus/probe configuration (e) personal equation, honesty


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007


(a) surface roughness (a) conditioning
(b) form deviations (b) number of measurements
(c) elastic modulus (Young’s modulus) (c) order of measurements
(d) Poisson ratio (d) duration of measurements
(e) stiffness (e) choice of principle
(f) temperature, CTE (f) choice of reference
(g) thermal conductivity and diffusivity (g) choice of apparatus
(h) weight, shape, size (h) choice of metrologist
(i) magnetism (i) alignment
(j) hygroscopic characteristics (j) number of operators
(k) temperature (k) strategy
(l) internal stress, stability (l) clamping
(m) creep characteristics (m) fixturing
(n) workpiece distortion due to clamping (n) number of data points
(o) aging (o) probing principle
(p) cleanliness (p) probing strategy
(q) orientation (q) alignment of probing system
(r) drift check
(s) reversal measurements
C-8 DEFINITIONS OF THE CHARACTERISTICS (t) multiple redundancies, error separation

(a) datum, reference system

(c) toleranced feature FACTORS
(d) distance knowledge of the correct physical values of, for exam-
(e) angle ple, material properties (workpiece, measuring instru-
(f) reference conditions ment, ambient air, etc.)


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007



The thermally related issues of a dimensional mea- From eq. (D-1) we see that the measurand (the length
surement are often a major source of uncertainty. In this of the workpiece at 20°C) depends on three input quanti-
Appendix the basic methods of addressing this issue ties: L, ␣, and ⌬T. Uncertainties in each of these imper-
are described. The effects of temperature in dimensional fectly known quantities will contribute to the
metrology are treated in more detail in ISO/TR 16015 uncertainty in L20°C. In this Appendix we focus only on
[12] and ANSI/ASME B89.6.2 [13]. the thermal issues and hence consider only the uncer-
Consider a workpiece whose length, L, is measured tainty components due to CTE and temperature uncer-
at a nonstandard (other than 20°C) temperature. If the tainties. In a complete uncertainty evaluation,
result of the measurement is to be compared with a uncertainty components associated with the length mea-
tolerance requirement (specified on a drawing, for exam- surement, such as reproducibility and calibration uncer-
ple) then the measured length must be corrected for tainty, would also be included in the analysis.
thermal expansion. The length, L20°C, of the workpiece The GUM procedure for uncertainty evaluation
at standard temperature is involves techniques from differential calculus that may
be unfamiliar and seem overly complicated in an indus-
L20°C p L (1 − ␣⌬T) (D-1) trial environment. For simple thermal expansion prob-
lems an alternative procedure gives similar results. For
where the length measurement of eq. (D-1) the steps are as
L p the measured length at temperature, T follows:
Step 1: Assign uniform probability distributions to ␣
␣ p the coefficient of thermal expansion (CTE) of
and T. Thus ␣ is sure to lie between ␣min and
the workpiece material
␣max with best estimate ␣ p (␣min + ␣max)/2,
⌬T p T − 20°C
and T is sure to lie between Tmin and Tmax with
best estimate T p (Tmin + Tmax)/2.
The difference, ⌬L p L − L20°C, between the measured
Step 2: Evaluate the best estimate of the measurand
length and the length at standard temperature is a ther-
using the best estimates of the input quantities.
mally induced systematic error. From eq. (D-1) we see In this case L20°C p L(1 − ␣⌬T), according to
that eq. (D-1).
Step 3: Evaluate the change in length, L␣, when ␣ is
⌬L p L␣⌬T (D-2)
replaced by ␣max. Using eq. (D-1) the result is
In this Report it is assumed that corrections are always ⌬L␣ p L⌬T(␣max − ␣) (D-3)
applied to eliminate known significant systematic
errors.1 Hence the thermally induced length error, ⌬L,
This is the maximum length error that could be caused
in eq. (D-2) is eliminated by a correction (having the
by an error in the value of the CTE.
opposite sign) added to the measured length. After cor-
Step 4: Convert the error component, ⌬L␣, to an uncer-
recting the measured length for the known systematic
tainty component, u␣(L), by dividing by 冪3
error, the uncertainty in the measured length, L, must
(assigning a Type B uniform distribution; see
be evaluated. The thermal aspects of the uncertainty Nonmandatory Appendix B):
evaluation are described below.
L⌬T(␣max − ␣min) L⌬T(␣max − ␣)
u␣(L) p p (D-4)
2冪3 冪3
If a significant systematic effect is known to exist but there is Step 5: Evaluate the change in length, ⌬LT, when T is
no way to know the magnitude or sign of the associated error, replaced by Tmax. Using eq. (D-1) the result is
then the correction will be zero. The uncertainty budget, however,
should include a component (usually a Type B assignment) that
accounts for the possible magnitude of the unknown error. (See (Tmax − Tmin)
⌬LT p L␣ p L␣ (Tmax − T) (D-5)
para. 6.2.) 2

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

This is the maximum length error that could be caused The thermally related component of the combined
by an error in the temperature. standard uncertainty [see eq. (D-7)] is then
Step 6: Convert the error component, ⌬LT, to an uncer-
tainty component, uT(L), by dividing by 冪3 ucth (L) p 冪11.52 + 12.12 ␮m p 16.7 ␮m
(again assigning a uniform distribution):
The measurement result would then be stated L p
L␣ (Tmax − T) 2.000000 m, U(k p 2) p 33 ␮m
uT(L) p (D-6)
Step 7: Combine the uncertainty components in a
root-sum-of-squares procedure. The result is D-3 THERMAL SELF-COMPENSATION
the component of the combined standard In some length measurements no explicit temperature
uncertainty of the measured length due to measurement is taken for the purpose of calculating the
thermal effects: correction due to thermal expansion at a nonstandard
temperature. Typically this occurs when the measure-
c (L) p 冪u␣(L) + uT(L)
uth 2 2
(D-7) ment is “self compensating” in the sense that both the
instrument scale or master gauge and the artifact under
where the superscript th denotes the thermal component
measurement expand similarly. For example, when a
of uncertainty.
master gauge block and customer gauge block of the
EXAMPLE: Suppose a laser interferometer is used to measure same nominal length are both made of steel, and are in
the distance between two points on an aluminum rod, and the equilibrium in the same environment, it is reasonable
measurement result is 2.000220 m. The laser is compensated for
to assume that they expand thermally by the same
the wavelength of light in the measurement environment and hence
its measurement scale is adjusted to 20°C. The temperature mea-
amount. In calibrating the customer’s block by compari-
surement of the rod is 25°C using a calibrated thermometer, with son with the master block, the resulting nominal differ-
±0.5°C assumed to be the maximum plausible error in the tempera- ential expansion then “cancels out” and does not need
ture measurement. The length of a material object is defined at an explicit correction. The uncertainty in the calibrated
20°C, so a correction must be performed to account for the thermal length of the customer’s block will nevertheless have
expansion of the rod. thermally related components due to uncertainties in
The expected thermal expansion is corrected for the the CTEs of the master and customer blocks and uncer-
systematic error due to thermal expansion using tainties in the temperatures of the blocks.
eq. (D-2) is ⌬L p 2.0 m ⴛ 22 ⴛ 10−6/°C ⴛ 5.0°C p Although both the master and customer gauge blocks
220 ␮m. Hence the best estimate of the rod is 2.000000 m. are in the same thermal environment, so that their tem-
[Because the measured and corrected lengths of the rod peratures are expected to be the same, there might exist
are similar, a negligible error is committed by using only a small temperature difference between the blocks,
two significant digits for the rod length in eq. (D-2).] denoted by ␦T p TM − T, where TM and T are the
Assuming the rod is isothermal, the uncertainty in temperatures of the master and customer blocks, respec-
the length measurement is shown below. tively. The best estimate of ␦T is zero, but knowledge
(a) CTE of the Rod. Since the type of aluminum is of the thermal environment would suggest a uniform
unknown, a uniform distribution is assigned to its CTE, distribution of possible values between ␦Tmin and ␦Tmax.
with ␣min p 20 ⴛ 10−6 / °C, ␣max p 24 ⴛ 10−6 / °C, and Denote the lengths, at 20°C, of the master and cus-
␣ p (␣min + ␣max) / 2 p 22 ⴛ 10−6 / °C. The CTE-related tomer blocks by Lm and L, respectively, and let d be the
component of length uncertainty [see eq. (D-4)] is measured length difference. Then, using eq. (D-1),

L⌬T(␣max − ␣) d p L (1 + ␣⌬T) − LM (1 + ␣M⌬TM ) (D-10)

u␣(L) p (D-8)
where ␣ and ␣M are the associated CTEs. The desired
2 m ⴛ 5°C ⴛ 2 ⴛ 10−6/°C
p p 11.5 ␮m length, L, is then given by
d + LM (1 + ␣M⌬TM )
(b) Temperature of the Rod. Based on knowledge of the Lp (D-11)
1 + ␣⌬T
thermometer calibration, a uniform distribution is
assigned to the temperature, with Tmin p 24.5°C, Tmax p Since (1 + ␣⌬T)−1 ≈ 1 − ␣⌬T, and ⌬TM p ⌬T + ␦T,
25.5°C, and T p 25°C. The temperature-related compo- eq. (D-11) can be simplified to
nent of length uncertainty [see eq. (D-6)] is
L ≈ LM + d + C1 + C2 (D-12)
L␣(Tmax − T)
uT(L) p (D-9)
冪3 where
2 m ⴛ 22 ⴛ 10−6/°C ⴛ 0.5°C C1 p LM(␣m − ␣)⌬T
p 12.1 ␮m
C2 p LM␣M␦T


Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007

The desired length, L, is thus equal to the length, LM, 1 pLm (␣max − ␣min) ⌬Tmax pLm⌬␣ max⌬Tmax (D-15)
C max
of the master block plus the measured difference, d, with
additive corrections, C1 and C2, that account for the [In this expression, the small uncertainty, u(Lm), in the
effects of nominal differential expansion and the differ- length of the master block has been ignored.] Since it is
ence in temperatures between the blocks. unlikely that ⌬␣ and ⌬T will both be at their maximum
When the master and customer blocks are of the same limits, it is reasonable to assign a triangular distribution
material, then ␣m p ␣ and the best estimate of correction, to the possible values of C1. (See para. B-3 of Nonmanda-
C1, is equal to zero. Also, since the best estimate of the tory Appendix B.) Thus
difference, ␦T, is zero, the best estimate of correction,
C2, is also equal to zero. Then the resulting best estimate, Lm⌬␣ max⌬Tmax
u(C1) p (D-16)
Lest, of the length of the customer block is 冪6
D-3.1.2 Uncertainty Component Due to Correction,
Lest p Lm + d (D-13)
C2. Assume that knowledge of the temperature differ-
where ence, ␦T, is characterized by a uniform distribution over
d p best estimate of the measured length difference the interval ±␦Tmax. Then the reasonably probable values
Lm p best estimate of the length of the master block of C2 lie in the interval ±C2max, where

C 2max p Lm␣ max␦ Tmax (D-17)

The associated standard uncertainty is then
Since it is unlikely that the master block CTE and the
u (L) p 冪u2 (Lm ) + u2(d) + u2 (C1) + u2 (C2) (D-14) temperature difference will both be at their maximum
limits, a triangular distribution is assigned to the possi-
The uncertainty component, u(Lm), is evaluated based ble values of C2, so that
on the information supplied by the calibration certificate
Lm␣ max␦ Tmax
for the master block, and the component u(d) is evalu- u(C2) p (D-18)
ated based on what is known about the comparator 冪6
system used to measure the length difference. The ther- EXAMPLE: Suppose a 100-mm steel gauge block is being cali-
mally related components u(C1) and u(C2) may be evalu- brated by comparison to a 100-mm steel master gauge block. There
ated as follows. is no temperature measurement (and associated thermal correc-
tion) performed, but it is known that the blocks are in an environ-
D-3.1 Thermally Related Uncertainty Evaluation ment that varies between 20°C and 22°C. It is also known that
the blocks are always within ±0.1°C of each other. Determine the
Uncertainty components associated with the correc- thermally induced uncertainties.
tion terms C1 and C2 in eq. (D-12) are evaluated by
estimating their maximum values and assigning appro- For the steel blocks, assume ␣max p 12.5 ⴛ 10−6/°C
priate probability distributions. and ␣min p 10.5 ⴛ 10−6/°C. The maximum temperature
deviation is ⌬Tmax p 2°C, and the maximum tempera-
D-3.1.1 Uncertainty Component Due to Correction, ture difference is ␦ T max p 0.1°C. Then setting
C1. Knowledge of the block CTEs and the temperatures Lm ≈ 100 mm, the nominal length of the blocks, the
are modeled by uniform distributions. Thus ␣ and ␣m desired uncertainty components are evaluated using eqs.
are assumed to lie in the interval [␣min, ␣max],2 with best (D-16) and (D-18).
estimates ␣m p (␣min + ␣max)/2 and ␣ p (␣min + ␣max)/2.
Let ⌬Tmax be the maximum possible value of the tem- 100 mm ⴛ 2 ⴛ 10−6/°C ⴛ 2°C

u(C1) p ≈ 0.16 ␮m (D-19)

perature deviation ⌬T p T − 20°C, regardless of sign. 冪6
(For example, if the temperature is known to lie between
18°C and 21°C, ⌬Tmax p 2°C.) Then the reasonably prob- 100 mm ⴛ 12.5 ⴛ 10−6/°C ⴛ 0.1°C
u(C2) p ≈ 0.05 ␮m (D-20)
able values of the correction, C1, lie in the interval ±C1max, 冪6
The thermally related component of the combined
2 standard uncertainty is then
Although the two CTEs are assigned the same probability distri-
bution, knowledge of one of them would not change knowledge
of the other, so that they are uncorrelated. u(L)thermal p 冪0.162 + 0.052 ␮m ≈ 0.17 ␮m (D-21)

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007


The following sources were cited using numbered ref- [7] ISO International Vocabulary of Basic and General
erences throughout this Report: Terms in Metrology, International Organization for Stan-
[1] ISO Guide to the Expression of Uncertainty in dardization, 2nd Edition (1993).
Measurement, International Organization for Standard- [8] ASME Y14.5M-1994, Dimensioning and Toler-
ization (1995). ancing.
[9] T. Doiron and J. Stoup, Uncertainty and Dimen-
[2] ANSI/NCSL Z540-2, U.S. Guide to the Expression sional Calibrations, Journal of Research of NIST, 102, 647–
of Uncertainty in Measurement, National Conference of 676 (1997).
Standards Laboratories (1997). [10] ASME B89.1.9-2002, Gage Blocks.
[3] ASME B89.7.3.1-2001, Guidelines to Decision Rule: [11] S.D. Phillips, K.R. Eberhardt, and B. Parry, Guide-

Considering Measurement Uncertainty in Determining lines for Expressing the Uncertainty of Measurement
Conformance to Specifications. Results Containing Uncorrected Bias, Journal of Research
[4] ASME B89.7.3.3-2002, Guidelines for Assessing of NIST, 102, 577–585 (1997).
the Reliability of Dimensional Measurement Uncertainty [12] ISO/TR 16015:2003(E), Geometrical product
Statements. specifications (GPS) — Systematic errors and contribu-
tions to measurement uncertainty of length measure-
[5] ASME B89.7.4.1-2005, Measurement Uncertainty ment due to thermal influences, ISO, Geneva (2003).
and Conformance Testing: Risk Analysis. [13] ANSI/ASME B89.6.2-1973 (R1979, R1988), Tem-
[6] P. J. Mohr and B. N. Taylor, The fundamental phys- perature and Humidity Environment for Dimensional
ical constants, Physics Today, August 2000, 6–13. Measurement.

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale

Copyright ASME International
Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME Services

ASME is committed to developing and delivering technical information. At ASME’s Information Central, we make every effort to answer your
questions and expedite your orders. Our representatives are ready to assist you in the following areas:

ASME Press Member Services & Benefits Public Information

Codes & Standards Other ASME Programs Self-Study Courses
Credit Card Orders Payment Inquiries Shipping Information
IMechE Publications Professional Development Subscriptions/Journals/Magazines
Meetings & Conferences Short Courses Symposia Volumes
Member Dues Status Publications Technical Papers

How can you reach us? It’s easier than ever!

There are four options for making inquiries* or placing orders. Simply mail, phone, fax, or E-mail us and an Information Central representative
will handle your request.

Mail Call Toll Free Fax—24 hours E-Mail—24 hours

ASME US & Canada: 800-THE-ASME 973-882-1717
22 Law Drive, Box 2900 (800-843-2763) 973-882-5155
Fairfield, New Jersey Mexico: 95-800-THE-ASME
07007-2900 (95-800-843-2763)
Universal: 973-882-1167

* Information Central staff are not permitted to answer inquiries about the technical content of this code or standard. Information as to
whether or not technical inquiries are issued to this code or standard is shown on the copyright page. All technical inquiries must be
submitted in writing to the staff secretary. Additional procedures for inquiries may be listed within.

Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale
ASME B89.7.3.2-2007



Copyright ASME International

Provided by IHS under license with ASME
No reproduction or networking permitted without license from IHS Not for Resale