Chapter 4

Force, Torque, Stress, and Pressure

Measurement

Our standards of force are ultimately based on weight, magnified by levers

as necessary. Sometimes direct application of weights is convenient, or we
use spring balances which are easily checked by attaching weights. A
minor problem with weights is that gravity varies perceptibly, mainly due
to the earth’s non-spherical shape. Gravity has been measured by free-
falling body methods to be about 9.8127 Newtons per kg (or m / s 2 ) in
London but would be about 9.7725 N/kg in Quito, near the equator and
2800 m above sea level [data from Kaye and Laby (3)]. Local gravity
variations would not matter at all when weighing by comparing masses and
only rarely in engineering work. It would affect the time-keeping of
pendulum clocks, or could cause arguments about the price paid for a given
mass of gold if weighed using a strain gauge device or spring balance.

This chapter is mainly devoted to measurements involving transducers.

However, the simple mechanical methods still have a useful role. For
example, spring balances may be spurned by the modern engineer or
technician as being imprecise and incapable of being connected to a
computer. However, they are cheap, portable, and do not drift
unpredictably or go out of calibration. (Obviously, the spring may stretch
or creep if permanently loaded, but this is readily checked by unloading.)
They are also valuable for in situ approximate calibration of load cells, to
check that the instrumentation as a whole is set up correctly.

4.1 Load cells

These devices are widely used for many purposes including overload
warnings, tank content indication, measuring material quantities passing on
a belt conveyor; if individually calibrated they can take the place of
expensive mechanical weigh-bridges, which shows how much confidence
they have earned. However, they require electrical instrumentation and
36 Engineering Measurements

may show a drift with time so that for critical purposes re-calibration may
be required at intervals.

Most load cells use resistance strain gauges but there is no need to
understand their detailed construction in order to use them. The main
forms of load cell are the in-line design and the cantilever type known as a
shear beam. In-line cells may be made for tension or for compression;
many can be used for either, fitted with a domed button for compression,
or hooks or other fixings for tensile applications. The loading need not be
exactly central since the strain gauges can be arranged to compensate for
some eccentricity (see below). In the absence of individual calibration the
output is usually guaranteed only to 2% accuracy.

Commercially, cells are available for loads of many tons, others down to
1 g. For some applications the regular cells are inconvenient; devices such
as those shown in Fig. 4.1 can be made and calibrated by dead-weights.
Two precautions should be taken: firstly the electrical instrumentation
should be as identical as possible for calibration and for the actual work,
secondly the position of the load line may give non-linear performance,
thus the calibration should cover the whole expected range, not depending
on extrapolation from small loads or vice versa.

PLAN OF
2 STRAIN TIP
GAUGES

SOLDERED
JOINT

HANGER

Fig. 4.1 Strain gauge set-up

 Force, Torque, Stress, and Pressure Mensuremertt 37

Traditional platform scales have a lever system ensuring parallel motion,

so that the potential energy change for a given deflection, and hence the
reading, is the same at all load positions. When testing scales, the loads are
placed at a variety of positions to check on this aspect. Scales using several
load cells but no parallel motion mechanism should have matched cells
with equal load-deflection properties and should have a highly linear
output-load relation, otherwise the outputs will not add up correctly for all
loads and load positions.

4.2 Metal foil strain gauges

4.2.1 Principles
The resistance of a strain gauge increases when put under tensile stress.
The main reason is an intrinsic change in the material while under load
(not a permanent change) but part of the effect is the increase in length and
the reduction of cross-section as noted in Chapter 6 (the Poisson
contraction). The effects are lumped together under the title of gauge
factor. The gauges are bonded to the relevant surface; the meaning of the
gauge factor is that if the gauge is extended by 1 part per million (one
micro-strain or pe) and its gauge factor is 2, then its electrical resistance
increases by 2 parts per million. If a gauge is attached to a surface with
non-uniform strain, the resistance change relates to the average strain
under the thin portion of the grid (the gauge length).

Thermal expansion of most structurally common metals exceeds 11 parts

per million for each degree Celsius, whilst the gauge material could have a
very different coefficient of expansion. Since working strains may often be
below 100 parts per million, it is clear that small changes of temperature
could give serious errors. The change of resistance with stress has been
known for over a century but the temperature compensation problem was
only overcome much later.

It is normally assumed that the gauge foil itself and the plastic carrier are
so thin compared with the component to which they are bonded that it has
the same elastic extension and the same actual thermal expansion (or
contraction). Therefore unless its natural thermal expansion is identical
with that of the component, the gauge goes into tension or compression as
the temperature changes, even with no external load. The intrinsic
resistance of the gauge material also tends to change with temperature. It
has been found possible to select and treat metals used for strain gauges in
38 Engineering Measurements

such a way that the thermal expansion compensates for the resistivity
change in the gauge material. The effect is called self-compensation; it can
be made to suit typical steel values, aluminium or many other substances,
for example concrete. In this connection one should note that, for example,
stainless steels have a higher coefficient than mild steel, whilst high-nickel
alloys range widely, including Invar with a low coefficient near 0 "C
(though it has a much higher coefficient at high temperatures). Some
aluminium alloys used for pistons have a low coefficient, others a
coefficient higher than pure aluminium. A long-established supplier [(4),
formerly under a different name] offers a great many levels of
compensation. Nevertheless where the object is likely to vary considerably
in temperature, the dummy gauge principle is adopted (see below). This
also tends to equalize the heating effect which may come from the
resistance-measuring instruments.

Modem strain gauges can stand extensions well beyond those commonly
used in metal components, even the larger strains which may arise in
plastics. It may be noted that in compression the resistance decreases, the
gauge factor remaining the same over quite a large range. The permissible
strains on some gauges are so large that they can be used to monitor crack
growth and plastic extension.

The most common errors associated with strain gauges are not the gauges
themselves being overstrained or damaged, but that they become unbonded
from the surface to which they are attached. The readings will then drift or
be unrepeatable.

Another problem with strain gauges is positioning relative to

discontinuities. Often it is desired to measure the strains at a position with
a high stress concentration such as a fillet radius. The exact positioning on
the fillet is difficult to achieve consistently, and it is also difficult to
maintain good bonding on a sharply curved surface. Therefore under these
circumstances it is advisable to fit a second strain gauge nearby on a less
curved surface where the conditions are better. The output from this
second gauge then acts as a check on the condition of the first.

The second 'check' gauge also has the merit that it will probably be
unaffected if a crack forms in the component in the vicinity of the first
gauge. This effect was seen on a fatigue test rig, where the measured strain
at a critical fillet radius decreased by about 30% as the test proceeded,
while a second gauge about 10 mm away maintained a steady reading.
 Force, Torque, Stress, and Pressure Measurement 39

Investigation showed that the reason was the formation of a sub-surface

crack on the fillet radius, close to the main strain gauge. This crack caused
a re-distribution of strain, effectively casting a ‘shadow’ which reduced the
strain reading from the gauge. On a test rig with defined loading, this effect
was interesting but not critical. However, if using strain gauges to monitor
the effects of random loads in service, for example on an aircraft wing, the
effect could be dangerously misleading, indicating decreasing strains and
loads when the component was actually cracking and close to failure.

Semi-conductor strain gauges are occasionally used instead of the metal

foil gauges. These give a much higher output, e.g. 5-10 times that of the
metal foil gauges, but are reported to be very temperature sensitive, and
suffer from drift and hysteresis.

4.2.2 Construction of metal foil strain gauges

Figure 4.2 shows a typical foil strain gauge, though many have a greater
length-to-width ratio. The delicate foil is attached to a plastic carrier
(matrix). The shape of the grid is produced photochemically. There are two
reasons for the large end-turns. One is that they provide secure attachment
for the long thin grid bars, the other is that since the electrical resistance of
the wide portions is very small, any change in resistance due to transverse
strain can nearly always be ignored. BS 6888 1988 requires

I U D V

COPPER COATED FOIL STRAIN ELEMENT

TABS

Fig. 4.2 A typical foil strain gauge

40 Engineering Measurements

the transverse sensitivity to be below 0.15% of the linear sensitivity unless

stated otherwise by the supplier. Some gauges are supplied with short leads
attached, others provide anchor points prepared for easy soldering-on of
leads. The top usually comes covered by a protective plastic film but can
be obtained bare, which helps with heat dissipation but should at least be
protected from moisture by a suitable varnish.

Strain gauges are normally available with an active length (gauge length)
ranging from 2 to 13 mm. The shortest gauges are intended for non-
uniform stresses. The finite length of the gauge itself can lead to some
averaging, particularly when trying to pick up peak stresses around stress
raisers. The overall length is greater than the active length, to ensure good
bonding. There are also multiple patterns available to measure two strains
at right angles, or three strains at 60" to each other. The multiple gauges
are known as rosettes, the name being taken from one form which has three
linear gauges superimposed. Many multi-directional arrangements are side-
by-side rather than stacked, yet are still called rosettes.

The most usual resistance values provided are 120 ohms or 350 ohms,
whilst the most usual gauge factors are 2.0 or 2.1.

Normally a gauge with its carrier is bonded to the surface of a component,

which need not be flat though it should not be sharply curved. Double
curvature calls for the use of very narrow versions and special attention
during bonding, such as using pre-shaped resilient pressure pads to ensure
that the bonding film is thin and uniform. For some applications it is
convenient to purchase gauges ready bonded to a thin steel sub-base which
can be attached to a component by spot-welding. This would apply mainly
to flat or slightly convex surfaces, under tensile stress only.

4.2.3 Resistance measurement

The measurement may be by a balanced Wheatstone bridge (see Chapter
7), or by a bridge which is balanced at zero load and which measures the
imbalance current in the galvanometer. Since the resistance changes are
small it is not necessary to have a full-range variable accurately known
resistor in the arm adjacent to the gauge, but to fit a fixed backing-off
resistor equal to the gauge resistance, Fig. 4.3a. The arm may also contain
a small, accurate variable resistor which is set, manually or automatically,
to balance the bridge. This resistor gives the increase or decrease of the
gauge resistance; it can be calibrated directly in micro-strains by setting up
the gauge parameters. Alternatively, the backing-off arm is set up for a
 Force, Torque, Stress, and Pressure Measurement 41

zero reading at no load and then kept constant so that the bridge becomes
unbalanced under load, the galvanometer current being used to find the
strain.

Instead of a backing-off resistor in the bridge box, it is often better to place

a dummy gauge near the working gauge, mounted on similar material and
located where the temperature is the same but not subject to load, Fig. 4.3b.

If a strain gauge amplifier is used the output may be given a zero offset to
prevent it bottoming out if there is any drift. If the strain gauge is to be
used bi-directionally, it is necessary to remember to remove this offset, or
the output will be asymmetrical and incorrect.

4-

WORKING
GAUGE

(b)

Fig. 4.3 Strain gauge resistance measurement

42 Engineering Measurements

4.2.4 Leads resistance

Although the leads resistance may only be a fraction of an ohm,
nevertheless it seems to be good practice to use the three-wire method for
single gauges as well as for the dummy gauge method. This compensates
out changes in the leads resistance in a similar way to the compensation of
resistance thermometer leads, see Chapter 5. One of the two power leads is
in series with the gauge, the other with the backing-off resistor or the
dummy gauge in the adjacent arm. The third lead only carries the
imbalance current. Alternatively a four-wire system can be used,
corresponding to Fig. 5 . 6 ~ .

The dummy gauge method is used when there are significant temperature
changes at the gauge and/or there is uncertainty about the coefficient of
expansion of the object, or worry about heating due to the instrument current.

It may be noted that these schemes still give an error, though it is usually
insignificant: for example, if the gauge is of 120 ohms resistance and the
leads each of 0.5 ohms (which could represent 50 m of copper cable,
depending on size), then the fundamental output for a given strain is that of
the gauges alone, but the reading obtained is only 120/121 of that value.

4.2.5 Bending effects

In cases of bending it is often possible to use one gauge on the tensile face,
and another on the compressive face. Then there is no need for a dummy
gauge since each resistance offsets the other in the unloaded state. Under
bending, one resistance increases while the other diminishes, therefore the
outputs are additive. If the component is thin it is worth remembering that
the gauges are further from the neutral axis than the object’s surface and
therefore show a higher strain; it may also be that the stiffening effect of
the gauges slightly reduces the deflection. These two effects work in
opposing senses, reducing the error.

The converse of this principle is used to compensate for unwanted bending

due to non-central loading. If we have a member in tension or compression,
we can fit matched pairs of gauges on opposing faces and connect them in
series. The pure tension or compression outputs are simply additive but any
bending increases the resistance of some gauges while decreasing the
resistance of their opposites by the same amount. The gauge factor of a
gauge which satisfies BS 6888 has to be uniform to 0.5 % in tension and
compression up to 1000 pe (microstrain) unless stated otherwise by the
 Force, Torque, Stress, and Pressure Measurement 43

supplier. It is also possible to put gauges into all four arms of the bridge to
increase the output.

4.2.6 Excitation
The power for the bridge is known as the excitation, as opposed to any
power to the indicating or recording equipment. It is always specified by
the supplier, usually 10 V. It may be ax., d.c., or pulsed. Alternating
current causes heating in the fine grid bars and may also cause eddy
current heating in the component or in the dummy gauge base; any
temperature difference between gauge and base, component or dummy
gauge defeats the temperature compensation. As heating of the gauge foil
is likely to predominate, the makers recommend using gauges as large as
possible. A bare-topped gauge shows a further advantage in this way but it
is only slight, since moisture protection has to be provided.

With d.c. the grid heating is still present, though the eddy current problem
is absent. On the other hand there is another problem, though we have not
seen any reference to its severity. At the junction of the leads to the gauge
metal the current produces a Peltier effect heating at one joint, cooling at
the other, by the excitation current. This gives a thermocouple voltage
which, it seems, must influence the output voltage at the meter. The pulsed
form of excitation seems to be a way of reducing the Peltier heating
problem. Presumably the instruments using this method provide a hold
facility for the last reading and either a manual control or automatic ways
of preventing needlessly frequent pulsing.

4.3 Torque measurement

For measurement of torque on static components, the most effective and

simple method is to use a torque arm and a force transducer. Such an
arrangement can be readily calibrated by means of weights. This is the
method traditionally used for engine dynamometers, where the torque
absorbed by the dynamometer is measured at the casing of the unit. If there
are rapid torque fluctuations, or significant transient accelerations of the
rotating components, this method can be inaccurate, and it is then
preferable to measure the torque on the rotating shaft.

Torque transducers are available for mounting on rotating shafts. They are
generally strain gauge or piezo-electric based, and transmit their output to
44 Engineering Measurements

the instrumentation either by slip rings or by radio signals. Use of such

transducers is appropriate if they can be incorporated at the design stage,
on a test rig for example. However, if torque measurements are required on
an existing machine it is often physically impractical to incorporate a
torque transducer, and even if space were available, the transducer may
affect the dynamic characteristics of the machine. As the reason for
carrying out the measurements is often to establish the dynamic torques
and torsional vibration characteristics, use of torque transducers could give
false indications.

The approach commonly adopted is to apply strain gauges directly to a

portion of the rotating shaft, and to transmit the signals by a radio (or infra-
red) link to the instrumentation. The equipment attached to the rotating
shaft can be smaller than a matchbox, and has self-contained batteries
suitable for some hours of operation. This method has been used
successfully on a wide range of machinery in some adverse environments.
If continuous operation is required, it may be necessary to provide a power
input via slip rings or a transformer. Slip rings are sometimes used for
transmission of the signal, but they are prone to noise, and therefore radio
transmission is preferred where possible.

One potential source of error is the presence of undetected static torques. If

the strain gauges are applied to the machine in service, then there may be a
significant static torque applied to the shaft; possible reasons include
friction in gears or bearings, or the use of automatic brakes which engage
when the machine is stopped. The gauge will read the locked-in torque as
zero, which can lead to quite large errors. A simple cross-check may be to
measure the motor current under steady state running, and calculate the
torque from data available from the motor manufacturer. Alternatively, it
may be possible by manually rotating the machine slightly, to establish a
position where the shaft is free of torque, and set the zero accurately.

Axial loads and bending moments can also significantly affect the reading.
Ideally the torque should be measured directly at the point of interest, but
this is not always possible. If there are couplings, gearboxes, or just
substantial inertias or flexibilities between the transducer and the point
where the torque is desired to be known, these can substantially modify the
dynamic shape of the signal. The case study below illustrates this point.
 Force, Torque, Stress, arid Pressure Measurement 45

Case Studv
Problems with an in-line torque transducer
On a valve train wear rig used for testing of lubricants, an in-line torque
transducer was used to measure the torque supplied to the cam-shaft via the drive
belt. An unexpected increase of peak torque with speed was obtained, and when
the results were studied in more detail it was found that the roughly sinusoidal
signal correspondingto the valve lift had a higher frequency signal superimposed
upon it. This affected the peak torques, giving unrepeatable results.

The cause was found to be frictional torque fluctuations in an Oldham type

coupling downstream of the torque transducer. Modifications to the
coupling were tried, but the ultimate solution was to use a bellows type
coupling with no frictional effects.

4.4 Pressure gauges and pressure transducers

The familiar pressure gauges with dial and pointer (needle) mostly employ
the Bourdon tube principle. A tube of oval cross-section bent into a C-
shape reduces its curvature, i.e. straightens out slightly, when fluid
pressure is applied. This happens because the outer and inner parts of the
curve move further apart while remaining essentially unchanged in length.
The movement of the free end is quite small and is magnified by some
convenient means such as a tie-bar and lever or magnified further by a gear
sector and a pinion on the pointer shaft, Fig. 4.4a. The method works under
suction too, the tube radius becoming smaller. The zero is not normally
adjustable, except by removing the pointer from its shaft and refitting in a
different position. This lack of adjustment means that if the zero is found
to have shifted, the gauge is likely to have been damaged or overloaded.

As the curved tube is almost impossible to clean out, the Bourdon type of
gauge is not favoured in food, pharmaceutical, and many other processes.
A diaphragm gauge is preferable, Fig. 4.4b. The fluid side of the
diaphragm-type gauge is made cleanable by removing the gauge together
with its diaphragm from the plant, leaving the mechanism undisturbed.

For remote indication and also for rapidly changing conditions pressure
transducers are used; these have a diaphragm exposed to the fluid. The
diaphragm may carry strain gauges, or press upon a piezo-electric crystal
as described in Section 4.5. Alternatively, the deflection may be inferred
from the capacity change between the diaphragm and a fixed electrode.
46 Engineering Measurernents

(a)
Schematic
of GASKET

Bourdon
gauge DIAPHRAGM SECURED
TO GAUGE BODY

(b)
Diaphragm gauge

Fig. 4.4 Pressure gauge types

The strain gauge type uses a Wheatstone bridge instrument as described

earlier; under steady conditions it could perhaps be used with a manual
Wheatstone bridge to measure the resistance change under pressure,
though a fixed bridge is more convenient; this becomes unbalanced by the
resistance change, the unbalance being used for the pressure indication.
The other types require electronic circuits to create a signal suitable for
transmission to an indicating or recording instrument. To have confidence
in the readings it is best to have a static instrument (gauge or manometer)
either permanently connected to the same service, or ready to connect by
opening a valve, so that a static comparison may be made of the transducer
plus its instrumentation. The dynamic problems are discussed below.

Most transducers are made insensitive to temperature changes by choosing

materials with appropriate coefficients of thermal expansion. In this way
the resistance, the capacity, or the force on the crystal can be made
independent of slow temperature changes. However, there is a serious
 Force, Torque, Stress, and Pressure Measurement 47

difficulty with rapid temperature changes. Supposing that the fluid

temperature increases quickly, the exposed face of the diaphragm expands
first so that the diaphragm bows outwards. This has the same effect on the
signal as a reduction of pressure. When measuring fluctuating gas
pressures there is normally a temperature change accompanying each
pressure change, hence false readings are quite common. An extreme case
is the measurement of pressures in the combustion chamber of an IC
engine, where there are very large and rapid temperature fluctuations. In
this application, some pressure transducers can give dramatically incorrect
readings. The error may be reduced if some non-rigid thermal shielding can
be placed over the diaphragm, which hinders heat flow but allows the
pressure change to reach the diaphragm. Coatings of silicone rubber have
been used to good effect. It may be thought that Invar alloy could be used
for the diaphragm in some cases. Invar has a very low coefficient of
thermal expansion over a certain range of temperature.

If a pressure transducer is to respond to rapid changes in a gas, an obvious

precaution is to avoid creating a resonance chamber. Unfortunately many
pressure transducers are constructed like a Helmholtz resonator, with a
reduced passage leading to a chamber in front of the diaphragm. If a
volume V is connected to the pressure region by a duct of cross-sectional
area A, length L, then the natural frequency is readily derivable from first
principles, being given by
1
(Hz)
24-
where yis the adiabatic index of the gas, P is the absolute gas pressure, p is
the gas density (note that (y.P/p)'~' is the speed of sound in the gas
concerned, at the prevailing temperature). LE is the equivalent length of the
passage allowing for end effects. An approximate figure for LEfor a round
hole is the actual length plus one diameter. A similar expression in another
form for slots is given by Turner and Pretlove (5). Their expression should
also apply to liquids where the cavity frequency is much higher than in
gases provided the cavity is really full of liquid.

In liquids the installation should be such that no gas bubbles can

accumulate. As it seems to be customary to place a transducer either
vertically above a pipe or vessel, or horizontally, the chamber in front of
the diaphragm is almost certain to hold air bubbles For fast-changing
conditions the transducer diaphragm should be as nearly flush with the
48 Engineering Measurements

surface as feasible and installed to allow gas bubbles to rise away, and
preferably not on top.

4.5 Piezo-electric transducers for force, pressure,

and acceleration measurement
These transducers are relatively expensive, and their charge amplifiers
even more so, but they have advantages in a number of situations. They are
active transducers, utilizing the piezo-electric effect by which if the crystal
lattice of, say, a quartz crystal is deformed under load it generates an
electric charge. If a number of the quartz plates are then loaded in series
and connected in parallel this increases the magnitude of the charge, which
can then be fed through a charge amplifier and measured without serious
leakage. Piezo transducers are always used with a charge amplifier to
oppose the leakage, but nevertheless they are intended for dynamic
conditions and not for static loads or pressures.

Between the transducer and the charge amplifier a special high-insulation,

low-noise cable is used. This is manufactured under highly controlled
conditions and in a vacuum to avoid oxidation at the joints. It is not a good
idea to remove the end connector and then shorten the cable and re-solder
the joint, as this can lead to severe loss of signal at this point! Downstream
of the charge amplifier any cable can be used.

Piezo transducers are commercially available for dynamic force, pressure,

and acceleration measurement. They were first used in IC engine
development, measuring the cylinder head pressure, with the advantages of
compactness, high natural frequency and a low temperature sensitivity.
They are not particularly susceptible to acceleration, but for measuring low
pressures under severely vibrating conditions it may help to use a low pass
filter to remove the vibration signals, or resort to an acceleration-
compensated version. Normally extensive temperature compensation is
used, so that the temperature error is very small, for static or slowly
varying temperatures. As well as the quartz crystals, both silicon and
ceramic piezo-electric sensors are now available, with the ceramic sensors
having a wider temperature range and much higher sensitivity than quartz.

In force measurement the piezo-electric sensors have the advantage of high

spring stiffness so that there is almost zero displacement and an excellent
frequency response. If high stiffness is critical it may make the difference
between choosing a piezo transducer or a load beam. They can be
 Force, Torque, Stress, and Pressure Measurement 49

constructed to measure in one to three planes; the 'cross-talk' between the

different planes caused by manufacturing limitations being normally
insignificant. However, in one case where low friction coefficients were to
be measured a 1% cross-talk from the normal load would have swamped
the friction reading. Their high stiffness and the possibility of cross-talk
mean it is particularly important to de-couple any loads you do not wish to
measure upstream of the transducer.

For example, in a test machine simulating the loading on artificial hip and
knee joints, the transducer was to be used to measure the normal loading
and the hydrodynamic friction in the joint during movement. Ultimately
two transducers were used; a load washer in line with the normal load, and
a transducer at 90" to the load, de-coupled with flexures, to measure the
friction. The load washer was pre-loaded, avoiding any tensile stresses due
to bending moments, with an elastic-shank type bolt to reduce the loss of
sensitivity due to loads being transferred through the bolt.

Attention has to be paid to the mounting of load washers. For example, the
supporting and mounting surfaces should be level and rigid (see Fig. 4.5a)
to give a uniform stress distribution across the washer. Figure 4.5b shows
deflection of the pre-load screw loading the inner edges of the washer to
give an incorrect output value. Insulating washers can also be used if
earthing presents a problem.

(a) CORRECT (b) INCORRECT

GIVES UNIFORM STRESS DISTRIBUTION

Fig. 4.5 Method of mounting load washers

50 Engineering Measurements

F F

(a) INCORRECT (b) CORRECT

MEASURES ONLY REDUCTION IN PRESSING MEASURES LOADING FORCE EXCEPT
FORCE BETWEEN FLANGE AND FOUNDATION THAT DIVERTED THROUGH ELASnc
SHANK FIXING BOLT

Fig. 4.6 Position of mounting load washers

The position of mounting the washer is also important; Fig. 4.6a shows a
load washer being incorrectly used to measure a bearing force, as it
measures only the reduction in pressing force between the flange and the
foundation, not an increase in the screw force. Figure 4.6b shows the
correct mounting arrangement, so that the load washer sees all of the force
except that diverted through the fixing bolt. It is calibrated in situ to
determine this. Further details of the correct use of the transducers are
given in reference (6).

4.6 Optical and fibre-optic based sensors

In recent years a number of sensors for force, pressure, etc. have been
developed based on optical devices (light or infra-red), communicating their
signals via optical fibres. These have several merits, the principal one being
complete immunity to electromagnetic interference. Other advantages include
inherently high-frequency response, and the ability to pass signals across small
air gaps or through windows. These are not necessarily exotic or expensive
devices, and are appearing in various applications including automotive. These
devices are effectively an extension of the principle of the opto-isolator, a well-
known and economic device for transmitting electrical signals while
maintaining electrical isolation.
 Force, Torque, Stress, and Pressure Measurement 51

4.7 Manometers

For low pressure differences manometers are widely used, using the levels
of a liquid to give the pressure difference. The chief problems lie in the
lines which connect the manometer to the source of pressure; since a major
use is with flow measurement devices, this aspect has been covered at
some length in Chapter 6; however, it applies to manometer use in general.

At the manometer itself, the first point of attention is the scale, which may
be a direct measurement or may be scaled so that only one reading is
needed, with compensation for level change in the other branch (usually an
enlarged reservoir). The second point is the liquid used. In many instances
the scale is intended for water but some manometers use a mixed fluid
which has more consistent properties of wall-wetting, to minimize surface
tension problems. For some applications with large pressure differences,
mercury is used; if so one must watch out for an oxidized skin which may
affect readings. The laboratory may have an apparatus for filtering
mercury. Mercury vapour is toxic, hence only properly enclosed equipment
should be used.

A sensitive manometer useful for calibration is the Casella instrument.

This uses mercury, with a fairly large surface, perhaps 20 mm in diameter,
to cut down meniscus curvature errors. It is fitted with a microscope and
the observation is made accurate by using a conical point on a micrometer
screw; the reflection of this point in the mercury pool allows a sensitive
detection of the level. It is important to stand the instrument on a very firm
table and check its position by the inbuilt spirit level.

Manometer readings may be magnified by using a sloping tube, in which

case the above remarks about a firm table and levelling are particularly
important. Another magnification method is to use two non-mixing liquids
in a differential mode, so that the pressure difference is the level change
times the difference between the two densities. This is a little cumbersome
since both densities need to be known accurately. Industrial flow recorders
are available which include a manometer for low differentials, in a
chamber strong enough for high pressure. The manometer uses mercury,
the height change being transmitted by a float operating a lever system or
an electrical displacement transducer (thus avoiding a moving seal
problem). Another method does not use a float but detects the mercury
level by carbon rods dipping into the mercury so that their resistance
decreases as the mercury rises.
52 Engineering Measurenients

4.8 Floating bell instruments

A brief mention is justified of floating bell instruments, used chiefly for

chimney draught or ventilating systems. They can conveniently carry
clockwork chart recorders. The action is by opposing the pressure above or
below ambient against the buoyancy of a hollow metal can and appropriate
counterweights, partly immersed in water. The bell rises or falls with the
measured pressure.

4.9 Method for measuring small near-horizontal forces

The advantages of this method are absence of electrical apparatus, absence

of pulley friction and low cost. Figure 4.7 shows a stand carrying a thread,
0.5 to 1 metre long. This carries a simple weight carrier easily produced
from wire and plastic or aluminium, a ruler and another thread leading to
the point of interest. The application in this case was measuring the drag of
objects in slow-moving water. The force is found from the weight carrier
plus added weights. For improved accuracy, allowance can be made for
part-weight of the threads, and corrections for the slope of the near-
horizontal thread (downward component of drag force added to the applied
weight and cosine correction for the drag). This very simple apparatus has
been used to measure forces of a few milli-Newtons in a virtually self-
calibrating manner.

Fig. 4.7 Method for measuring small near-horizontal forces