CO

PI
Ca

A
Torque Wrench: transducers and

N
rlo

O
calibration
Fe

N
rre

CO
ro

N
CN

FO
Ferrero Carlo
R-

RM
INRiM
IM

E
Training Course INRiM-ICE-NPL
G
C
 CO
PI
Ca

A
Torque wrenches (TW) are often considered and

N
rlo

O
used as “tools” (such as screwdrivers and similar),
Fe

N
rather than true measuring “instruments”.
rre

CO
They provide a “visible” measure of torque
ro

N
(dial indicating, flat beam TW) or are used to
CN

FO
control torque in assembly and fastening
operations (slipper, clicker-TW).
R-

RM
IM

E
G
C
 Torque Traceability in 2005
Calibration
National Standard = Torque Standard Testing

CO
Machine
(INRIM
INRIM)
+/-0.005 % ~

PI
Torque Transfer
Standards

Ca

A
N
rlo

O
Secondary Standard = Secondary Standard =
Torque Calibration Torque Calibration Machine

Fe

N
Machine for the reference torque
for the torque transducer wrench
+/-0.05 % ~
rre +/-0.05 % ~

CO
ro

N
Reference torque

CN Reference torque

FO
wrench ± 0.2 % ~
Reference torque drive ± 0.2 %
transducer ± 0.2 % ~ ~

R-

RM
Torque Driver
Torque Wrench Torque Wrench
Torque Driver Calibration Device
Calibration Device Calibration Device
Calibration Device without Loading
having Loading
without Loading
Mechanism ± 1 % ~
IMMechanism ±1 % ~
having Loading
Mechanism ± 1 % ~

E
Mechanism ± 1 % ~

G
To the user of the torque
transducer
Ex) Torsion Testing
±3%~ C
Machine, Engine Test Stand, Signal Type
Indicating Signal Type Indicating
etc Torque Wrench ± 3 %
Torque Driver ± 3 % ~ Torque Driver ± 3 % ~ Torque Wrench ± 3 % ~
~
 Basis knowledge

What`s Torque ?

CO
PI
"Torque" is defined as the product of a force and
Ca

A
the perpendicular distance of its line of application

N
rlo
from the point of rotation or fulcrum.

O
Fe

N
rre

CO
F
r
F ro α

N
β .
CN

FO
R-

RM
l
IM
M = F * r * sin α

E
M=F*l G
C
M = F * r * cos β
 Application fields
Torque applications

CO
Safety engineering

PI
Ca

A
Off-shore-industry

N
rlo
Material testing

O
Screw and assembly technology

Fe

N
Aviation industry

rre

CO
Marine drive technology

roAutomotive industry

N
CN

FO
Medical meas. technology

R-
Automation technology

RM
Viscometry
IM

E
1 mN·m 100 mN·m 10 N·m 1 kN·m
G 100 kN.m
10 mN·m 1 N·m 100 N·m
C
10 kN·m
Nominal torques at PMI
1 MN·m

0,1 N·m 80 kN·m

 CO
How to Make Torque Evaluation.

PI
Ca

A
N
Screw rlo

O
tighten
Fe

N
100 cm
rre
50 cm

CO
ro

N
CN

FO
R-

RM
IM

E
Force of 200 Force of 100
N G N
C
 Torque Evaluation

CO
PI
Ca

A
Hand A Torque = F(A) x L(A)

N
rlo

O
Fe

N
rre = 200 N x 50 cm

CO
ro = 1000 N*cm

N
CN

FO
Hand B R-

RM
Torque = F(B) x L(B)
IM

E
= 100 N x 100 cm
G
C
= 1000 N*cm
 Why is Torque Necessary?

CO
PI
Ca

A
In case of threaded objects:

N
rlo
Example: Bolts of car-wheel

O
Case 1 : Lightly tightened (lack of torque)
Fe

N
rre
Bolts will be loosened gradually and car accident might occur.

CO
ro

N
Case 2 : Strongly tightened (over torque)
CN

FO
Bolts are tightened just before breaking point, So bolts
R-

RM
might be suddenly broken, due to higher tension caused by
temperature or shocks IM

E
G
Properly tightened (suitable torque)
Ok C
 Why is Torque Necessary?

CO
PI
Ca

A
In case of unthreaded objects:

N
rlo

O
Fe
Example : Channel of TV-sets

N
rre

CO
Case 1 : Lightly tightened (lack of torque)
ro
Easy to turn/bad contact/ unstable picture.

N
CN

FO
Case 2 : Strongly tightened (over torque)
R-

RM
IM
Difficult to turn / large friction / short life,

E
difficult to use. G
C
 Bolt Tightening Method:

CO
PI
Ca

A
There are three ways to make sure that tightening

N
rlo
is correct.
correct

O
Fe

N
rre

CO
(A) Torque control method
ro

N
Using hand torque tool (wrench or driver), or
CN

FO
power operated torque tool (screwdriver or nut
R-

RM
runner) to tighten screws, nuts and bolts.
IM

E
G
C
 CO
PI
Ca

A
(B) Part – turn method:

N
rlo
Bolts are tightened until joint surfaces are in

O
Fe

N
close contact. At that point, just before solid
rre

CO
tightening starts, nut and protruding bolt threads
ro
are marked to show their relative positions.

N
CN

FO
Then, correct tightening is completed by turning
R-
the nut a half or 3/4 turn.

RM
IM

E
G
C
 CO
PI
Ca

A
N
rlo

O
(C) Bolt elongation measuring method
Fe

N
When bolts are tightened at the same time, the
rre

CO
elongation of the bolts are measured.
ro

N
CN

FO
R-

RM
IM

E
G
C
 What is Suitable Torque?

CO
PI
Ca

A
N
rlo
Case 1 : When the joint does not receive external

O
Fe
force or vibration

N
rre

CO
T = (0.5 to 0.7) P
ro
Case 2 : When the joint receives external force or

N
CN

FO
vibration
R-

RM
T = (0.45 to 0.55) P
IM
Where : T = Suitable Torque

E
G
P = Breaking Torque
C
 Dynamic Torque

CO
If the power (P) transmitted by a rotating shaft is to

PI
Ca
be measured, it is necessary to determine the speed of

A
N
rotation (ω) in addition to torque.
rlo

O
Fe

N
rre

CO
P=ωT ω= 2* π *N / 60 rev. / sec
ro

N
CN

FO
Then :P = 2 * π * N * T / 60
R-

RM
T = dynamic torque (Nm)IM

E
G
N = Number of rotation in minute (rpm)
C
P Nm/s = (watt)
Clamping force generated by applying

CO
torque

PI
Ca

A
N
rlo

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
G
C
 CO
PI
Calibration procedures for hand torque tools are defined
Ca

A
by the ISO 6789:1992 standard, which does not provide

N
rlo
any information or guideline on how to evaluate the

O
uncertainty of measurement.
Fe

N
rre

CO
This report analyses the situation in EU calibration
ro
laboratories, key issues and methods in calibration of hand

N
CN
torque tools in the calibration lab and the industry.

FO
R-
It also suggests a basic measurement uncertainty budget.

RM
These information can be relevant for quality managers
IM

E
with the need to set up a torque wrench calibration facility
in their own company. G
C
 CO
PI
1. Hand torque measuring “tools” or “instruments”?
Ca

A
Control of the applied torque is required especially when

N
rlo
particular materials or gaskets are being used with

O
bolts/nuts connections.
Fe

N
rre
Industries facing possible safety (pressure containers, nuclear,

CO
aerospace, etc.) or guarantee issues (i.e. automotive) do
ro
actually perform, also for this type of tools, calibration, time stability

N
CN

FO
and R&R (repeatability and reproducibility) studies.
Nevertheless, this activity has normally a lower priority,
R-

RM
compared to i.e. management and calibration of dimensional
measuring instruments. IM

E
G
C
 CO
PI
Ca

A
Measurement of torque in

N
rlo
production is generally performed

O
Fe
with hand torque measuring tools,

N
more generally
rredefined as

CO
"assembly tools for screws and
ro

N
nuts". They can be divided in two
CN

FO
classes: "dial indicating” or
R-

RM
"clicker" tools, defined in ISO
6789:1992 (see figure 1). We IM

E
describe here only the most
important ones: G
C
 2
Different types of application

E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 Dial indicating torque wrenches

CO
PI
These TW are designed for the
Ca precise application and

A
measurement of any torque

N
rlo value, without the need for

O
ancillary calibration or setting
Fe

N
equipment. They usually have bi-
rre

CO
directional tightening.
ro The instant torque value is

N
CN

FO
displayed continuously on an
analogic or digital display, the
R-

RM
user looks at the display while
Being used as an “instrument”,
these TW usually provide good IM
operating the tool and stops

E
tightening when the desired
results, and have generally a good
G
torque value is achieved.
repeatability
reproducibility.
and a medium
C
 Click torque wrenches (adjustable by the user)

CO
PI
Ca

A
In this case, the required nominal

N
rlo
value is pre-set on an indicator (a

O
scale) by the user, the tool is
Fe

N
then used as a normal spanner,
rre
often without a particular “care” in

CO
operation. When the required
ro
torque is reached , the tool

N
CN

FO
switches with a perceptible,
unmistakable and audible click
R-

RM
(see figure 10). At that point the
user should immediately stop
IM

E
tightening.
G
C
 CO
PI
Ca
These TW are used in many

A
different tightening applications, in

N
rlo
assembling departments, in

O
workshops, and they are easy to
Fe

N
be found in all car repair
rre

CO
workshops, where a “flexible”
ro
tool is required for various tasks.

N
CN

FO
This flexibility leads to a lower
repeatability and reproducibility of
R-

RM
the required torque value.
IM

E
G
C
 CO
PI
Figure shows the inside structure of a clicker-TW, with a
Ca

A
multi-lever action. The spring in the hilt applies a force

N
rlo
against one or more elements, which are connected,

O
through a multi-lever system, to the square drive. Setting
Fe

N
the TW means compressing the spring and applying a
rre

CO
higher force to the elements.
ro

N
CN

FO
R-

RM
IM

E
G
C
 CO
When the user operates the TW and the force balance

PI
is reached, there is a click point (peak value) and an
Ca

A
immediate decrease of the torque value (see figures

N
rlo
from 10 to 12 for the typical behavior of clicker-TW).

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM
The capacity of converting exactly the force F applied

E
from the spring (N) in the required torque (Nm) and the
G
C
“repeatability” of this “transformation” are the secrets of
a "good" clicker-TW design.
Slipper torque wrenches

CO
A variant of the previous one,

PI
this TW type cannot be adjusted
Ca

A
directly by the user, as it doesn’t

N
rlo have a setting scale or indicator.

O
Fe It is also known as “production

N
rre type TW”, is normally set against

CO
a torque wrench analyzer or
ro

N
calibration device TWCD and
CN

FO
then dedicated to a particular
R-

RM
application, where modifications
IM
of the value of the nominal

E
torque are not frequently
G
required. C
 CO
PI
Ca
From the metrological

A
point of view, this is the

N
rlo

O
“safest” tightening
Fe

N
system, as it eliminates
rre

CO
over-tightening and
ro
discourages unauthorized

N
alteration. It has a
CN

FO
smooth decreasing curve
R-

RM
after the peak value, with
the typical behavior IM

E
shown in figure 15. G
C
 Torque vs. time behavior

CO
PI
Ca

A
By analyzing different TW types and looking

N
rlo
closely what happens in proximity of the click

O
point,
point we can pinpoint some issues in TW
Fe

N
calibration.
rre

CO
Figures 10 to 12 represent the ideal situation, a
ro

N
well defined peak value, an almost immediate
CN

FO
and consistent decrease, which can be clearly
R-

RM
identified by the user.
IM
It is not possible to avoid completely the

E
G
increase in torque after the click, with click
torque wrenches. C
 CO
PI
Ca

A
N
rlo

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
Figure 10. A good torque
wrench shows a well remarkable IM
Figure 11. In this case the click is

E
peak, corresponding to the pre- even more clear and easier to be
G
identified
set torque value
C
 CO
PI
Ca

A
N
rlo

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
Figure 12. A variation like this can also be admitted,
G
C
provided that the click point can be easily identified.
 CO
PI
Let’s have a look at some “undesirable” behaviors. The
Ca

A
irregular curve shown in figure 13, with many local peaks,

N
rlo
relatively close to each other,

O
Fe

N
can confuse the user and reduce tightening accuracy
rre

CO
ro

N
CN

FO
R-

RM
IM

E
G
C
 CO
PI
Ca

A
N
Typical behavior of a
rlo

O
slipper TW
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
The advantages of using a slipper TW are even
IM

E
greater if the torque decreases, as shown here,
G
C
after having achieved the peak value
 The situation in Europe

CO
PI
The lack of ”metrological awareness” for these kind of instruments
Ca

A
leads, in general, to a poor overall quality level of “controlled

N
fastening”. rlo

O
Fe
Traceability of measured quantities to corresponding national

N
references (metrological chain) is not so well established for “torque”,
rre

CO
compared to other quantities
ro

N
Torque calibration is usually related to departments, in National
CN

FO
Metrological Institutes (NMI), dealing with Force and Mass, but in
many European countries NMIs do not perform calibration of torque
R-

RM
wrenches.
IM

E
Calibration of TW is therefore normally performed by accredited
calibration laboratories (ACL). G
C
 Calibration of torque wrenches

CO
PI
Ca

A
Calibration according to ISO 6789:1992 requires five

N
rlo
consecutive clicks (pre-load) without recording the

O
measure, at the maximum nominal torque (MNT),
Fe

N
followed by the measurement of five consecutive
rre

CO
positions at 20%, 60% and 100% of the MNT.
ro

N
If none of the 15 measured values is outside the
CN

FO
tolerance values (defined as 4% of the nominal value)
R-

RM
the TW is considered conforming according to ISO
6789:1992. IM

E
G
C
 CO
PI
Ca

A
N
4. Calibration of torque wrenches
rlo

O
Fe
Calibration can be performed with automatic or

N
rre

CO
manuals devices.
devices
ro

N
Automatic devices are normally used by calibration
CN

FO
laboratories, while manual devices, with a higher
R-

RM
uncertainty of measurement, are used in company
IM

E
calibration labs.
G
C
 Manual calibration devices (testers) for
torque wrenches

CO
Devices like the one shown in figure can be used by

PI
companies as “master calibration device” by companies, to
Ca

A
perform internal calibration or setting of torque wrenches.

N
rlo

O
They are also
Fe

N
used by some
rre

CO
calibration
ro laboratories, but

N
they show a
CN

FO
relatively higher
uncertainty,
R-

RM
compared to
IM automatic

E
G systems.
C
To improve the reproducibility of the click points (influence
of different users), the TWCD is located on a light and rigid

CO
aluminum structure. The crank on the sliding carriage

PI
allows a more continuous and linear application of the
Ca

A
force.

N
rlo

O
Fe

N
rre

CO
ro

N
TWCD
CN

FO
R-

RM
IM

E
G
C
 Automatic calibration devices for torque wrenches

CO
PI
Ca

A
An automatic calibration

N
rlo
device, for laboratory

O
Fe
use, is shown in figure

N
The torque is applied by
rre

CO
a motor, connected to
ro

N
the reference torque
CN

FO
transducer and, through
a two-speed gear box, R-

RM
to the TW in calibration. IM

E
G
C
 CO
PI
Ca
When torque is applied by the

A
motor, the transducer acquires

N
rlo
up to 2000 measuring values

O
Fe
every second and draws

N
torque vs. time graphs. The
rre

CO
software identifies the
ro

N
maximum value, compares it
CN

FO
with the nominal value and
print a calibration report
R-

RM
according to ISO 6789:1992
and/or torque vs. time IM

E
characteristic curves G
C
Automatic
system
CAGI-

E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 CO
PI
Ca

A
5. Software for calibration of torque wrenches

N
rlo
Next Figures show an example of a software being used

O
for calibration of TW, allowing the acquisition of the
Fe

N
measured values, management of different instruments
rre

CO
(calibration due-date, history, etc.) and print of calibration
ro

N
/ test certificates according to ISO 6789:1992.
CN

FO
R-

RM
Initially developed for length metrology, it is now being
IM
completed with a standard module for the evaluation of

E
uncertainty of measurement (CAGI) G
C
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 Key issues on calibration and use of torque
wrenches

CO
Initial shock

PI
Ca
The following graph shows a typical measuring sequence on

A
a common clicker-TW, set at 20 Nm (14.74 lbf.ft).

N
rlo

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
G
C
 Key issues on calibration and use of torque
wrenches

CO
The first click (usually only this one, sometimes the

PI
Ca
first two points) falls very often outside the upper

A
4% limit. Measurement values decrease gradually and

N
rlo

O
stabilize after a few clicks.
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
G
C
 CO
PI
Ca

A
This is the main reason why ISO 6789:2000

N
rlo
requires five initial clicks before taking any

O
measurement. Experience shows that the effects of
Fe

N
this “initial shock” do not prevent the TW from
rre

CO
having the same behavior during calibration, when
ro
measurements are taken.

N
CN

FO
The initial click, therefore, often falls outside the
R-

RM
upper 4% tolerance value even for relatively
“good” torque wrenches. IM

E
G
C
 CO
PI
This behavior also points out a strong correlation of
Ca

A
the measured values and makes the evaluation of

N
rlo
the uncertainty of measurement more difficult.

O
Fe

N
We cannot assume a Gaussian distribution for
rre

CO
this process.
ro

N
We should also consider that this phenomenon is not
CN

FO
identical for all the types of torque wrench designs,
R-

RM
and depends from many sources.
IM

E
For these reasons we suggest a conservative
top-down approach.
approach G
C
 Short-term time effect
ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

CO
2 sec 19,65 19,54 19,30 19,18 19,08 18,90 18,98 18,90 18,90 18,79 18,78 18,71 18,71 18,69 18,64 18,69 18,63 18,65 18,64 18,60
20 sec 19,30 19,17 19,04 18,98 18,92 18,92 18,93 18,90 18,95 18,94 18,85 18,84 18,83 18,82 18,87 18,75 18,77 18,74 18,78 18,78
00 sec 19,95 19,42 19,25 19,18 19,13 19,11 19,07 19,04 19,02 19,07 19,02 18,99 18,99 18,97 18,97 18,97 18,95 18,94 18,89 18,89
Multi-Line - individuals
20,0
300 s

PI
19,8

Ca

A
Figure shows a sample of 19,6
2s

twenty consecutive

N
rlo 19,4

O
measurements on the same
Fe
19,2
120 s

N
TW, set at 20 Nm (14.74 19,0

rre

CO
lbf.ft) and operated at time 18,8

ro
intervals of 2, 120 and 300

N
18,6

2 sec
seconds on an automatic
CN

FO
120 sec
300 sec

calibration device. The first

R-

RM
click point has always been The TW has been turned down (spring
considered in the IM
unloaded) for at least 6 hours, between a

E
sequence of 20 measures and the
calculations. G
following sequence, in order to reduce
interactions between sequences
C
 CO
The graphs are similar and show a correlation between

PI
torque and tightening frequency. Users should be made
Ca

A
aware of this behavior

N
rlo

O
ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
2 sec 19,65 19,54 19,30 19,18 19,08 18,90 18,98 18,90 18,90 18,79 18,78 18,71 18,71 18,69 18,64 18,69 18,63 18,65 18,64 18,60

An interval of 2 seconds
Fe
20 sec 19,30 19,17 19,04 18,98 18,92 18,92 18,93 18,90 18,95 18,94 18,85 18,84 18,83 18,82 18,87 18,75 18,77 18,74 18,78 18,78

N
00 sec 19,95 19,42 19,25 19,18 19,13 19,11 19,07 19,04 19,02 19,07 19,02 18,99 18,99 18,97 18,97 18,97 18,95 18,94 18,89 18,89
Multi-Line - individuals

(blue) between
20,0

rre
300 s

CO
consecutive releases is 19,8

not enough for the ro

N
19,6
2s

spring to go back to its CN

FO
19,4

original length, the click

R-

RM
19,2
120 s
is always at a lower
IM
19,0

torque (even after the

E
first 10 clicks
18,8

G
18,6

2 sec
120 sec
300 sec
C
 CO
PI
Ca

A
N
rlo
The green (t=120s) and brown (t=300s) curves

O
are very similar.
Fe

N
rre
Experience also shows that a longer the interval

CO
between clicks leads to overall click point values, the
ro

N
difference here is around 0.7 Nm (0.51 lbf.ft).
CN

FO
The graph shows that the spread of
R-

RM
measurement values reduces after about 10
clicks or more IM

E
G
C
 The importance of the first click points

CO
PI
The importance of the first click points is clear when
Ca

A
examining the histograms simulating “populations”

N
rlo
of consecutive clicks measured in a determined

O
position.. Fe

N
rre

CO
How many click points should we remove, in
ro
order to have a normal (Gaussian) distribution?.

N
CN

FO
We use the Chi-square test to define when the
distribution can be considered as a normal one
R-

RM
IM

E
G
C
 This histogram shows 60 measures, taken in 3

CO
consecutive cycles, each one with 20 consecutive
click points, including also the first click.The

PI
distribution is clearly skewed.
Ca

A
N
rlo

O
How many click
Fe

N
points should we
remove, in order to rre

CO
have a normal
ro

N
(Gaussian) distribution?
CN

FO
We use the Chi-square
R-

RM
test to define when the
distribution can be IM

E
considered as a normal
G
one. C
By excluding the first click point, and repeating
each time the chi-square test, we find out that

CO
the distribution is gaussian when we remove the

PI
first four points for each sequence
Ca

A
N
rlo

O
Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
G
C
 Long-term time effect

CO
PI
Ca

A
From the above pattern we saw a short-term interaction

N
rlo
between the click point number and the time.

O
Unfortunately, there is also a long term effect, which is
Fe

N
often not considered by TW users.
rre

CO
Clicker-type TW should always be turned down and
ro

N
stored at the lowest possible setting,
setting but this is usually
CN

FO
not the case.
R-

RM
IM

E
G
C
 Long-term time effect

CO
PI
Ca

A
N
rlo
The continuous “stress” on the internal spring can

O
Fe
be responsible for a long-term decrease in

N
performance (accuracy).
rre

CO
ro
It is necessary, before performing calibration,

N
CN

FO
to keep the torque at the minimum setting for a
R-

RM
certain amount of time (12 hours), in order to
IM
exclude or reduce this effect.
effect

E
G
C
 Evaluation of the uncertainty of measurement

CO
PI
Ca

A
The following example (top-down approach with an over-estimation)

N
rlo
refers to a calibration of a TW with an TWCD (figures 6 and 7) and

O
Fe
should be taken as a preliminary example, not as an official and

N
general way of determining uncertainty in torque calibration.
rre

CO
ro
It might be used, in principle, at the factory level, since it

N
CN
considers globally only three elements (reference standard, TW

FO
design and user) in determining the expanded uncertainty of
R-

RM
measurement.
IM

E
G
C
 Example

CO
PI
Ca

A
Calibration position: 100 Nm

N
rlo

O
Nominal value of ATWCD: 100 Nm
Fe

N
TW resolution: 5 Nm
rre

CO
TW measurement values:
ro

N
CN

FO
101.00, 100.00, 99.00, 98.50, 98.00
R-

RM
mean value of the 5 measurements = 99.300 Nm
IM

E
standard deviation = 1.204 Nm G
C
 Following contributions can generally be taken into
account:

CO
PI
Ca

A
• Contribution of the second line reference

N
rlo

O
(uCSL), which is basically the torque transducer of the
Fe

N
TWCD. This value should be taken from the
rre

CO
manufacturer’s calibration certificate, for the
ro
corresponding measuring range or defined position,

N
divided by the coverage factor (usually k=2). It’s not
CN

FO
unusual to take the maximum value of the expanded
R-

RM
uncertainty (i.e. 1%), convert it to Nm and then consider
0.5%. IM

E
G
C
 Following contributions can generally be taken into
account:

CO
PI
• Contribution of the second line reference (uCSL),
Ca

A
If the TWCD shows a drift over time, another source of

N
rlo
uncertainty should be added, based on historical data of

O
Fe
calibration certificates or experience.

N
rre

CO
In the numerical example we prefer to take the actual value
ro
0.270 Nm, corresponding to the 0,27% in the 100 Nm

N
CN

FO
position (interpolation can be necessary if that particular
calibration position is not included in the certificate), and
R-

RM
obtain 0.135 Nm.
IM

E
G
C
 CO
• Contribution of the resolution (ur).

PI
With this contribution we simply refer to the user’s error
Ca

A
when setting the required torque value on the TW scale.

N
rlo
This can roughly be related to the scale resolution r,

O
Fe
divided by a standard value. TW with a lower resolution

N
rre
(i.e. 1 Nm) can slightly reduce this contribution. We

CO
optimistically suggest to have a “normal distributed” user,
ro

N
divide the resolution by a factor 5 (20%) and then take ½
CN

FO
of this value. Some calibration laboratories consider a
R-

RM
rectangular distribution and divide this value by . In this
IM
particular case our value would then be 0.577 Nm instead

E
of 0.500 Nm.
G
C
 CO
PI
• Contribution of the TW inherent repeatability
Ca

A
(ub). We can try to define here the key issues discussed in

N
rlo
the previous paragraphs and evaluate it, to be on the safe

O
side, with the sample standard deviation (s) of the 5
Fe

N
measurements in each position according to ISO
rre

CO
6789:1992. To be on the safe side, we include the first
click point. ro

N
CN

FO
We consider this contribution as a type A uncertainty and
R-

RM
we do not divide this value by the coverage factor.
IM

E
Note: use (n-1) at the denominator when
evaluating s. G
C
 CO
PI
• Contribution of reproducibility,
reproducibility imperfect
Ca

A
connection of TW and TWCD fittings, local deformation,

N
rlo
friction and other factors. Their contribution, especially

O
reproducibility, has already been considered in ur ,
Fe

N
with the r/5 value (presumably high enough).
rre

CO
• ro
Contribution of environmental factors such as

N
temperature, humidity, pressure, etc. They are not
CN

FO
considered as relevant, we consider T = 22.0 ± 3 °C (71.6
R-

RM
± 5.4 °F), humidity lower than 70% and a temperature
IM
variation less than 1 °C (1.8 °F) during calibration (which

E
takes about 20 minutes). G
C
 CO
PI
To simplify, we assume no correlation and then have,
Ca

A
according to EA-4/02:

N
rlo

O
Fe U = k ⋅ u(X)

N
rre

CO
with ro

N
CN

FO
u ( X ) = (u CSL
2
+ u 2b + u 2r )
R-

RM
IM

E
G
C
 CO
PI
Ca

A
Source of Value Divisor u
Symbol Distribution
uncertainty Nm Nm

N
uCSL
rlo
Uncertainty of CD normal 2 ½ of the value of the

O
TWCD manufacturer’s expanded uncertainty
calibration on the calibration

Fe

N
certificate certificate, if given as
k=2

rre

CO
ub Contribution of the
inherent TW sample standard sample standard
repeatability
ro deviation normal 1 deviation

N
CN

FO
Contribution of r/5 2 r/10
ur normal
the user
Combined u
R-

RM
u normal
uncertainty
Expanded U
U
uncertainty
IM
normal, k=2

E
G
C
 Example

CO
PI
Ca

A
Calibration position: 100 Nm

N
rlo

O
Nominal value of ATWCD: 100 Nm
Fe

N
TW resolution: 5 Nm
rre

CO
TW measurement values:
ro

N
CN

FO
101.00, 100.00, 99.00, 98.50, 98.00
R-

RM
mean value of the 5 measurements = 99.300 Nm
IM

E
standard deviation = 1.204 Nm G
C
 CO
PI
Ca

A
Value Divisor u

N
Symbol
rlo
Source of uncertainty
Nm
Distribution
Nm

O
uCSL Uncertainty of TWCD
Fe

N
0.270 normal 2 0.135

rre

CO
ub’ Contribution of the inherent
TW repeatability
ro normal

N
1.204 1 1.204

CN

FO
ur Contribution of the user 1 normal 2 0,500
R-

RM
Combined uncertainty 1,310
u
Expanded uncertainty IM 2.621

E
U
G
C
 CO
PI
Ca

A
N
rlo
The uncertainty of measurement of the torque

O
wrench at the measuring position of 100 Nm would
Fe

N
then be given as
rre

CO
ro
99.300 ± 2.621 Nm, with a coverage factor of

N
k=2.
CN

FO
R-

RM
IM

E
G
C
 References

CO
PI
Ca

A
1. ISO 6789:1992 Assembly tools for screws and nuts -

N
rlo
Requirements and test methods. www.iso.ch.

O
Fe

N
2. EA-4/02 – Expression of the uncertainty of
rre

CO
measurement in calibration.-
ro

N
3. EA-10/14 – EA Guidelines on the calibration of static
CN

FO
torque measuring devices.
R-

RM
4. Prore, Milano, Italy – www.prore.it
IM

E
G
C
 CO
PI
Ca

A
Discussion and decision about: New proposals for

N
rlo
EA WG onTorque

O
Fe

N
All the attendants agree about the need of new EA-
rre

CO
Guides on
ro

N
STATIC CALIBRATION OF TORQUE WRENCH
CN

FO
CALIBRATION DEVICES and on CALIBRATION OF
R-

RM
TORQUE WRENCHES (TW),
IM
that fix some common rules in this measurement field

E
of high industrial interest.G
C
 CO
The following proposals have a chronological order.
order

PI
Ca
1. New guideline for “STATIC CALIBRATION OF TORQUE

A
WRENCH CALIBRATION DEVICES” DEVICES (based on DKD-R 3-8,

N
rlo
2003), add. with a test of defined defective square drive and

O
different side forces correlated to the nominal torque.
Fe

N
rre
2. EA ILC for STATIC CALIBRATION OF TORQUE WRENCH

CO
CALIBRATION DEVICES.
DEVICES
ro

N
3. New guideline for “CALIBRATION OF TORQUE
CN

FO
WRENCHES”
WRENCHES (see e.g. DKD-R 3-7, 2003 and ISO 6789, type I
and II).
R-

RM
IM
4. EA ILC for CALIBRATION OF TORQUE

E
WRENCHES
G
C
 Reference Torque Wrench RTW

CO
PI
Istituto di
Ca

A
Metrologia
“Gustavo

N
Colonnetti”
rlo

O
Torino

Fe

N
rre

CO
ro HBM TTS 100

N
CN

FO
R-

RM
IM

E
CONSIGLIO HBM TTS 500 G
NAZIONALE
DELLE
RICERCHE
C
 Calibration system for RTW

CO
PI
Istituto di
Ca

A
Metrologia
“Gustavo

N
Colonnetti”
rlo

O
Torino

Fe

N
rre

CO
ro

N
CN

FO
R-

RM
IM

E
CONSIGLIO
G
NAZIONALE
DELLE
RICERCHE
C
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
PI
A rre
Fe
CO rlo
Ca
 E
RM C
FO G
N IM
CO R-
N CN
O
N ro
THANKS
rre

ATTENTION
A
PI Fe

for YOUR
CO rlo

MANY
Ca