engg measurements
TRANSCRIPT
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Mechanical Measurements
Module 2:
1. Thermometry
Formally we start the study of “Mechanical Measurements” now!
Module 2 will consider the measurement of field quantities like temperature,
pressure and fluid velocity. First topic to be covered is “Thermometry” or the
Science and Art of “Temperature Measurement”
Sub Module 2.1
1. Thermometry or the science and art of temperature measurement
Preliminaries
Temperature along with pressure is an important parameter that
governs many physical phenomena. Hence the measurement of temperature
is a very important activity in the laboratory as well as in industry. The lowest
temperature that is encountered is very close to 0 K and the highest
temperature that may be measured is about 100000 K. This represents a
very large range and cannot be covered by a single measuring instrument.
Hence temperature sensors are based on many different principles and the
study of these is the material of this sub module.
We take recourse to thermodynamics to provide a definition for
temperature of a system. Thermodynamics is the studies of systems in
equilibrium and temperature is an important intensive property of such
systems. Temperature is defined via the so called zeroth law of
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
thermodynamics. A system is said to be in equilibrium if its properties remain
invariant. Consider a certain volume of an ideal gas at a specified pressure.
When the sate of this volume of gas is disturbed it will eventually equilibrate in
a new state that is described by two new values of volume and pressure.
Even though we may not be able to describe the system as it is undergoing a
change we may certainly describe the two end states as equilibrium states.
Imagine two such systems that may interact through a wall that allows
changes to take place in each of them. The change will manifest as changes
in pressure and/or volume. If, however, there are no observable changes in
pressure and volume of each one of them when they are allowed to interact
as mentioned above, the two systems are said to be in equilibrium with each
other and are assigned the same temperature. The numerical value that is
assigned will have to follow some rule or convention as we shall see later.
The zeroth law of thermodynamics states that is a system C is in
equilibrium separately with two thermodynamic systems A and B then A and
B are also in equilibrium with each other. At once we may conclude that
systems A and B are at the same temperature! Thermometry thus consists
of using a thermometer (system C) to determine whether or not two systems
(A and B in the above) are at the same temperature.
Principle of a thermometer
Principle of any thermometer may be explained using the facts
indicated in Fig. 1. Consider a system whose state is fixed by two properties
– coordinates – X and Y. It is observed that several pairs of values of X, Y
will be in equilibrium with a second system of fixed temperature (or a fixed
state). These multiplicity of sates must all be characterized by the same
temperature and hence represent an isotherm.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 1 Principle of thermometry explained
Assume that one of the coordinates of the system (Y) is fixed at a value
equal to Y0. Then there is only one sate that will correspond to any given
isotherm. If the system is allowed to equilibrate with a system characterized
by different isotherms, the property X will change as indicated by the points of
intersection X1, X2 and so on. These will then correspond to the respective
temperatures T1, T2 and so on. We refer to X as the thermometric property
and the system as a thermometer.
Y = Y0
X2 X3 X1 X4 X
Isotherms Y
T2 T3 T1 T4
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Table 1 Thermometers and thermometric properties
Table 1 shows several thermometers that are actually used in
practice. The thermometric property as well as the symbol that is used to
indicate it is also shown in the table.
Constant volume gas Thermometer
Figure 2 Schematic of a constant volume gas thermometer
Thermometer Thermometric property
Symbol
Gas at constant volume Pressure P Electric resistance under constant tension
Electrical resistance R
Thermocouple Thermal electromotive force
E
Saturated vapor of a pure substance
Pressure P
Blackbody radiation Spectral emissive power
,bE λ
Acoustic thermometer Speed of sound a
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
We look at the constant volume gas thermometer in some detail now.
Schematic of such a thermometer is shown in Fig. 2. It consists of a certain
volume of a gas contained in a rigid vessel. The vessel is connected to a U
tube manometer with a side tube and height adjustable reservoir as shown.
The volume is kept constant by making the meniscus in the left limb of the U
tube always stay at the mark mad3e on the left limb of the U tube. The right
limb has a graduated scale attached to it as shown. The gas containing
vessel is immersed in a constant temperature environment. The graduated
scale helps in determining the pressure of the confined gas in terms of the
manometer head.
The following experiment may be performed. Choose the pressure of
the gas to have a definite value when the constant temperature environment
corresponds to standard fixed state such as the triple point of water (or the ice
point at one atmosphere pressure). Now move the thermometer into an
environment at the steam point (boiling point of water at one atmosphere).
The pressure of the gas has to be adjusted to a higher value than it was
earlier by adjusting the height of the reservoir suitably so as to make the
meniscus in the left limb of the U tube stay at the mark. The above
experiment may be repeated by taking less and less gas to start with by
making the pressure at the triple point of water to have a smaller and smaller
value (the vessel volume is the same in all the cases). The experiment
may also be repeated with different gases in the vessel. The result of such
an experiment gives a plot as shown in Fig. 3.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 3 Gas thermometer characteristics
The ratio of the pressure of the gas corresponding to the steam point to
that at the triple point of water tends to a unique number as tpp 0→ (the
intercept on the pressure ratio axis) independent of the particular gas that has
been used. This ratio has been determined very accurately and is given
by sttp
tp
p 1.366049 as p 0p
→ → . The gas thermometer temperature scale is
defined based on this unique ratio and by assigning a numerical value of
0tpT 273.16 K or 0.01 C= . The defining relation is
st sttp
tp tp
T p as p 0T p
= → (1)
This last value is referred to as the single fixed point for
thermometry or the primary fixed point. At this temperature ice, liquid
water and water vapor all coexist, if in addition, the pressure is maintained at
4.58 mm Mercury column or 610.65 Pa. The ice point is at 273.15 K or 0°C
and was used in early times as the primary fixed point in thermometry. In
00.20.40.60.8
11.21.41.61.8
2
0 100 200 300 400 500
ptp
pst/
ptp
Gas A Gas B Gas C
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
ordinary laboratory practice the ice point is easier to achieve and hence is
commonly used.
Equation 1 may be generalized to define the constant volume gas
thermometer temperature scale as
tptp tp
T p as p 0T p
= → (2)
Thus the temperature ratio and pressure ratios are the same in the case of a
constant volume gas thermometer. The latter is measured while the former is
inferred. The message thus is clear! A measurable property that varies
systematically with temperature is used to infer the temperature! The
measured property is termed the thermometric property.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example1
In determining the melting point of a certain alloy with a gas
thermometer, an investigator finds the following values of the pressure
p when the pressure ptp at the triple point of water has the indicated
value.
ptp 100 200 300 400 p 233.4 471.6 714.7 962.9
If the triple point of water is taken as 273.16 K, what is the melting point
of the alloy?
o In order to determine the melting point we need the limiting value of the
ratio tp
pp as tpp 0→ . This value is obtained by extrapolation. The
ratios are calculated and are given by the following table.
p 100 200 300 400
tp
pp
2.334 2.358 2.382 2.407
Difference 0.024 0.024 0.025
o Since the common differences are constant we may extrapolate
linearly to get
31.2024.0334.2pp
limtp0ptp
=−=→
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Practical thermometry
We have mentioned earlier that the range of temperatures encountered
in practice is very wide. It has not been possible to device a single
thermometer capable of measuring over the entire range. Since all
thermometers must be pegged with respect to the single fixed point viz. the
temperature at the triple point of water it is necessary to assign temperature
values to as many reproducible states as possible using the constant
volume gas thermometer. Subsequently these may be used to calibrate other
thermometers that may be used to cover the range of temperatures
encountered in practical thermometry. These ideas are central to the
introduction of International Temperature Scale 1990 (or ITS90, for short).
The following is a brief description of ITS90.
Specification of ranges and corresponding thermometers according to
ITS 90
• Between 0.65 K and 5.0 K T90 is defined in terms of the vapor-pressure
temperature relations 3He and 4He.
• Between 3.0 K and the triple point of neon (24.5561 K) T90 is defined
by means of a helium gas thermometer calibrated at three
experimentally realizable temperatures having assigned numerical
values (defining fixed points) and using specified interpolation
procedures.
• Between the triple point of equilibrium hydrogen (13.8033 K) and the
freezing point of silver (961.78 ºC) T90 is defined by means of platinum
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
resistance thermometers calibrated at specified sets of defining fixed
points and using specified interpolation procedures
• Above the freezing point of silver (961.78ºC) T90 is defined in terms of a
defining fixed point and the Planck radiation law.
It is noted that the above uses several “secondary fixed points” to define the
temperature scale. These are shown in Table 2.
Table 2 Secondary fixed points used in ITS90
Equilibrium state
T90 K T90oC Equilibrium state T90 K T90
oC
Triple point of H2
13.8033
-259.3467 Triple point of Hg 234.3156
-38.8344
Boiling point of H2 at 250 mm Hg
17 -256.15 Triple point of H2O
273.16 0.01
Boiling point of H2 at 1 atmosphere
20.3 -252.85 Melting point of Ga
302.9146
29.7646
Triple Point of Ne
24.5561
-248.5939 Freezing point of In
429.7483
156.5985
Triple point of O2
54.3584
-218.7916 Freezing point of Sn
505.078
961.928
Triple Point of Ar
83.8058
-189.3442 Freezing point of Al
933.473
660.323
Freezing point of Ag
1234.93
961.78
Even though the ITS90 specifies only a small number of thermometers,
in practice we make use of many types of thermometers. These are
discussed in detail below.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
How do we make a thermometer?
Properties that vary systematically with temperature may be used as the
basis of a thermometer. Several are listed here.
Thermoelectric thermometer
• Based on thermoelectricity - Thermocouple thermometers using two
wires of different materials
Electric resistance
• Resistance thermometer using metallic materials like Platinum,
Copper, Nickel etc.
• Thermistors consisting of semiconductor materials like Manganese-
Nickel-cobalt oxide mixed with proper binders
Thermal expansion
• Bimetallic thermometers
• Liquid in glass thermometer using mercury or other liquids
• Pressure thermometer
Pyrometry and spectroscopic methods
• Radiation thermometry using a pyrometer
• Special methods like spectroscopic methods, laser based methods,
interferometry etc.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Sub Module 2.2
Thermoelectric thermometry
Thermoelectric thermometry is based on thermoelectric effects or
thermoelectricity discovered in the 19th century. They are:
• Seebeck effect discovered by Thomas Johann Seebeck in 1821
• Peltier effect discovered by Jean Charles Peltier in 1824
• Thomson effect discovered by William Thomson (later Lord Kelvin) in
1847
The effects referred to above were all observed experimentally by the
respective scientists. All these effects are reversible unlike heat diffusion
(conduction of heat) and Joule heating (due to electrical resistance of the
material) which are irreversible. In discussing the three effects we shall
ignore the above mentioned irreversible processes. It is now recognized that
these three effects are related to each other through the Kelvin relations.
Thermoelectric effects:
Figure 4 Sketch to explain Peltier effect
Consider two wires of dissimilar materials connected to form a circuit
with two junctions as shown in Fig. 4. Let the two junctions be maintained at
Current I
T1 T2
A
B
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
different temperatures as shown by the application of heat at the two
junctions. An electric current will flow in the circuit as indicated with heat
absorption at one of the junctions and heat rejection at the other. This is
referred to as the Peltier effect. The power absorbed or released at the
junctions is given by ABPP Q I= = ±π where ABπ is the Peltier voltage (this
expression defines the Peltier emf), PQ is rate at which the heat absorbed or
rejected. The direction of the current will decide whether heat is absorbed or
rejected at the junction. For example, if the electrons move from a region of
lower energy to a region of higher energy as they cross the junction, heat will
be absorbed at the junction. This again depends on the nature of the two
materials that form the junction. The subscript AB draws attention to this fact!
The above relation may be written for the two junctions together as
( )1 21 2 T T
T,Tπ = π − π (3)
Figure 5 Sketch to explain Thomson effect
Note that the negative sign for the second term on the right hand side
is a consequence of the fact that the electrons move from material A to
material B at junction 1 and from material B to material A at junction 2.
Consider now a single conductor of homogeneous material (wire A
alone of Fig. 4) in which a temperature gradient exists. The current I is
maintained by heat absorption or heat rejection along the length of the wire.
Current I
T1 T2
A
B
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Note that if the direction of the current is as shown the electrons move in the
opposite direction. If 2 1T T> , the electrons move from a region of higher
temperature to that at a lower temperature. In this case heat will be rejected
from the wire. The expression for heat rejected is2
1
T
T AT
Q I dTσ= ∫ where TQ the
Thomson heat is and Aσ is the Thomson coefficient for the material. A
similar expression may be written for the Thomson heat in conductor B.
Figure 6 Sketch to explain the Seebeck effect
If we cut conductor B (or A) as indicated the Seebeck emf appears across
the cut. This emf is due to the combined effects of the Peltier and Thomson
effects. We may write the emf appearing across the cut as
( ) ( )
( ) ( ) ( )
2 1
2 1 1 2
2
2 1 1
T TS P T A B A B A BT T T T
TA B A B A BT T T
V V V σ dT σ dT
σ -σ dT
= + = π − π + +
= π − π +
∫ ∫
∫ (4)
We define the Seebeck coefficient ABα through the relation sAB
dVdT
= α . In
differential form, equation 4 may then be rewritten as (assume 2 1T T dT− = )
A
B
T1 T2
VS
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
( )
( )
S ABAB A B
S AB A B
dV ddT dT
dV d dT
π= α = + σ − σ
= π + σ − σ
(5)
Kelvin relations:
Since the thermoelectric effects (Peltier and Thomson effects) are
reversible in nature there is no net entropy change in the arrangement
shown in Fig. 4.The entropy changes are due to heat addition or rejection at
the junctions due to Peltier effect and all along the two conductors due to
Thomson effect. The entropy change due to Peltier effect may be obtained as
follows:
At junction 1, the entropy change is P11
ABP
1 1
Qs I
T Tπ
= = . Similarly at
junction 2 the entropy change is P 2
1
BA ABP
2 2 2
Qs I I
T T Tπ π
= = = − . Again if we assume
that the temperature difference is 2 1T T dT− = the net change in entropy is
Pds IdTπ⎛ ⎞= ⎜ ⎟
⎝ ⎠. The net change in entropy due to Thomson heat in the two
conductors may be written as A B
T
ds I dTT
σ − σ= . Combining these two we get
A BP T
AB A BAB 2
ds=ds ds I d dT 0T T
d dTI dT 0T TT
σ − σ⎡ π ⎤⎛ ⎞+ = + =⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦π σ − σ⎡ ⎤= − π + =⎢ ⎥⎣ ⎦
(6)
The total entropy change is equated to zero since both the
thermoelectric processes are reversible. The current I can have arbitrary
value and hence the bracketed term must be zero.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
From equation 5 ( )A B s ABσ σ dT dV d− = − π . Introducing this in equation 6 we
get
sAB ABAB 2
dVd ddT 0T T TT
π π− π + − =
or
sAB AB S
dVT T TdT
π = = α = α (7)
We have renamed the Seebeck coefficient as Sα according to normal practice.
Differentiating equation 7 we get AB s sd Td dTπ = α + α . Introduce this in
equation 5 to get ( )s s s A B sdV Td dT dT dT= α + α + σ − σ = α or
( ) sA B
dTdTα
σ − σ = − (8)
Equations 7 and 8 constitute the Kelvin relations.
How do we interpret the Kelvin relations?
The Seebeck, Peltier and Thomson coefficients are normally obtained
by experiments. For this purpose we use the arrangement shown in Fig. 6
with the junction labeled 2 maintained at a suitable reference temperature,
normally the ice point (0°C). The junction labeled 1 will then be called the
measuring junction. Data is gathered by maintaining the measuring junction
at different temperatures and noting down the Seebeck voltage. If the
measuring junction is also at the ice point the Seebeck voltage is identically
equal to zero. The data is usually represented by a polynomial of suitable
degree.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
For example, with Chromel (material A) and Alumel (material B) as the
two wire materials, the expression is a quartic of
form 2 3 4S 1 1 1 1V a t a t a t a t= + + + where The Seebeck voltage is in µV and the
temperature is in °C. An inverse relation is also used in practice in the form
2 3 41 S 2 S 3 S 4 St A V A V A V A V= + + + . Two examples follow. We shall see later that
the coefficients in the polynomial are related to the three thermoelectric
effects.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 2
For the Chromel-Alumel pair the Seebeck voltage varies with temperature
according to the fourth degree polynomial
3 2S
6 3 10 4
V 39.44386 t 5.8953822 10 t
4.2015132 10 t 1.3917059 10 t
−
− −
= + ×
− × + ×
The Seebeck voltage is in µV while the temperature is in °C. Discuss the
behavior of this thermocouple near the ice point.
o It is clear that, near the ice point, the Seebeck coefficient (it is also
called the thermoelectric power) is
. s
sdV Vα 39.4 Cdt
µ= ≈ °
o Using the Kelvin relations, we also have the following:
1. ( ) ( )2
s SA B Chromel Alumel 2
dα d Vσ σ σ σ T -T dT dT
− = − = − =
2.
A B Chromel Alumel
S
T 39.44386T
− −π = π
= α ≈
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
o Note that T in the Kelvin relation is in Kelvin and t in the polynomial is
in 0°C. Also note that d ddT dt
≡ . Hence the second derivative of the
Seebeck voltage is given by
2 22S S
2 3 42 2d V d V 2a 6a t 12a tdT dt
= = + +
o Near the ice point we may take t = 0 and write
( )Chromel Alumelσ σ ( 2)(0.0058953822)(273.15) 3.2206 µV.
− ≈ −
= −
and
Chromel Alumel ST (273.15)(39.44386) 10774.1 V 0.0108 V−π = α = = µ ≈
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 3
The thermocouple response shown below (Copper Constantan
thermocouple with the cold junction at the ice point) follows the law VS
= a t + b t2. Obtain the parameters a and b by least squares. Here t is
in °C and VS is in mV.
Temperature, °C 37.8 93.3 148.9 204.4 260
VS, mV 1.518 3.967 6.647 9.523 12.572
o Since the fit follows the form specified above, it is equivalent to a linear
relation between E= VS/t and t. Since VS/t is a small number we shall
work with 100 VS/t and denote it as y. We shall denote the temperature
as x. The following table helps in evolving the desired linear fit.
x=t y=100VS/t x2 y2 xy yfit
37.8 4.015873 1428.84 16.12724 151.8 4.036073
93.3 4.251876 8704.89 18.07845 396.7 4.24048
148.9 4.46407 22171.21 19.92792 664.7 4.445255
204.4 4.659002 41779.36 21.7063 952.3 4.649661
260 4.835385 67600 23.38094 1257.2 4.854436
Sum: 744.4 22.22621 141684.3 99.22085 3422.7 6.638481
Mean = 148.88 4.445241 28336.86 19.84417 684.54 4.445181
(Note: I have used EXCEL to solve the problem)
o The statistical parameters are calculated and presented in the form of a
table:
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Variance of x= 6171.606
Variance of y= 0.084001
Covariance = 22.73252
Slope of fit line = 0.003683
Intercept of fit line = 3.896856
o Using the fit parameters the calculated values of yfit are shown in the
last column of the first table. Reverting back to t and VS, we get the
following relation:
( ) 25S t10683.3t03897.0100/3.896856t0.003683tV −×+≈+=
o We compare the data with the fit in the table below:
t VS VS(fit)
37.8 1.518 1.526
93.3 3.967 3.956
148.9 6.647 6.619
204.4 9.523 9.504
260 12.572 12.622
o Standard error of the fit with respect to data may be calculated as
mV035.03
003774.03
))fit(VV( 2SS
Err ==−
=σ
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
o The 3 in the denominator is the degrees of freedom. A plot is made to
compare the data and the fit. It is clear that the thermocouple behavior
is mildly nonlinear. The standard error of fir translates to approximately
±1°C!
o The inverse relation may similarly be obtained by fitting a linear relation
between S
tV
and SV . This is left as an exercise to the student. The
result obtained is 2S St 25.173V 0.3769V= − with a standard error
of t 2.209 Cσ = ± ° .
0
2
4
6
8
10
12
14
0 50 100 150 200 250 300Temperature oC
Seeb
eck
volta
ge m
V
VS VS(fit)
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 2 has shown that the thermoelectric data may be expressed
in terms of global polynomial to facilitate interpolation of data. A simple
quadratic fit has been used to bring home this idea. In practice the
appropriate interpolating polynomial may involve higher powers such that the
standard error is much smaller than what was obtained in Example 2. As an
example, the interpolating polynomial recommended for the K type
thermocouple with the reference junction at the ice point is given as:
3 2 6 3s
10 4
V 39.44386 t 5.8953822 10 t 4.2015132 10 t
1.3917059 10 t
− −
−
= + × − ×
+ ×
This is a fourth degree polynomial and passes through the origin. The
Seebeck voltage is given in µV and the temperature is in °C. Using Kelvin
relations, the appropriate parameters are calculated near the ice point as:
OSS
AB S
SA B
dV 39.444 V / CdtT 273.15 39.444
10774.1 VdT 273.15 2 0.005895dt
3.22064 V
α = = µ
π = α = ×
= µα
σ − σ = − = × ×
= − µ
The variation of the Seebeck coefficient over the range of this thermocouple is
given in Fig. 7 below
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 7 Variation of Seebeck coefficient with temperature
for K type thermocouple
A small excerpt from a table of Seebeck voltages is taken and the
corresponding fit values as calculated using the above fourth degree
polynomial is given in Table 1. Note that the voltages in this table are in mV.
Table 3 comparison of actual data with fit
t VS VS(fit) 37.8 1.52 1.499 93.3 3.819 3.728 148.9 6.092 5.990 204.4 8.314 8.273 260 10.56 10.581
371.1 15.178 15.237537.8 22.251 22.276815.6 33.913 33.874
1093.3 44.856 44.8791371.1 54.845 54.827
The maximum deviation is some 0.102 mV. The standard error is
approximately ±0.053 mV! This translates to roughly an error of ±1.2°C.
0
5
10
15
20
25
30
35
40
45
0 200 400 600 800 1000 1200 1400
Temperature oC
Seeb
eck
coef
ficie
nt µ
V/o C
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The above shows that the three effects are related to the various terms
in the polynomial. The Seebeck coefficient and the Peltier coefficients are
related to the first derivative of the temperature. The contributions to the first
derivative from the higher degree terms are not too large and hence the
Seebeck coefficient is a very mild function of temperature. In the case of K
type thermocouple this variation is less than some 2% over the entire range
of temperatures. The Thomson effect is related to the second derivative of
the polynomial with respect to temperature. The value of this is again small
and varies from -3.22 V at 0°C to a maximum value of +31.16 V at
1350°C.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Sub Module 2.3
Resistance Thermometry
Resistance thermometry depends on the unique relation that exists
between resistance of an element and the temperature. The resistance
thermometer is usually in the form of a wire and its resistance is a function of
its temperature. Material of the wire is usually high purity Platinum. Other
materials also may be used. The resistance variation of different materials is
indicated by Table 1.
Table 1 Resistance variation of different wire materials
Material Temperature Range °C
Element Resistance in Ohms at 0°C
Element Resistance in Ohms at 100°C
Nickel -60 to 180 100 152 Copper -30 to 220 100 139 Platinum -200 to 850 100 136
Platinum resistance thermometer is also referred to as (Platinum
Resistance Thermometer) PRT (or PT) or Resistance Temperature Detector
(RTD). Usually the resistance of the detector at the ice point is clubbed with it
and the thermometer is referred to as, for example, PT100, if it has a
resistance of 100 Ω at the ice point. The resistance of standard high purity
Platinum varies systematically with temperature and it is given by the
International standard calibration curve for wire wound Platinum elements:
( )( )
2 3 O Ot 0 1 2 3
2 O Ot 0 1 2
R R 1 K t K t K t 100 t , 200 C t 0 C
R R 1 K t K t ,0 C t 250 C
= + + + − − < <
= + + < < (9)
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Where
3 O 7 O 2 12 O 41 2 2K 3.90802 10 / C; K 5.802 10 / C ;K 1.2735 10 / C− − −= × = − × = − ×
The ratio 100 0
0
R R100R
− is denoted by and is given as 0.00385/°C. It is
seen that the resistance temperature relationship is non linear. The response
of a Platinum resistance thermometer is usually plotted in the form of ratio of
resistance at temperature t to that at the ice point as a function of temperature
as shown in Fig. 1. The sensor in the Fig.2 is shown with a three wire is
shown with three wire arrangement. The resistance sensor is also available
with four wire arrangement. These two aspects will be discussed later.
0
0.5
1
1.5
2
2.5
3
3.5
-200 -100 0 100 200 300 400 500 600
Temperature, oC
Res
ista
nce
ratio
, R(t)
/R0
Figure 1 Characteristics of a Platinum resistance thermometer
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 2 Typical PRT sensor schematic with three wire
arrangement
Platinum resistance thermometer and the Callendar equation
As seen from equation (1) the Platinum resistance thermometer has
essentially a non-linear response with respect to temperature. We define a
temperature scale defined as the Platinum resistance temperature that is
basically given by a linear scale defined through the relation]
t 0Pt
100 0
R Rt 100R R
−= ×
− (10)
The quantities appearing in the above are:
RPt = Platinum resistance temperature, Rt = Resistance of sensor at
temperature t, R0 = Resistance of the sensor at the ice point and R100 =
Resistance of the sensor at the ice point. Obviously the non-linearity will have
to be taken into account to get the correct temperature from the linear value
Ceramic Powder
Protective SheathPlatinum Element
1 2 3
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
obtained by Equation 10. This is done by applying a correction to the
Platinum resistance temperature as suggested by Callendar.
From Equation 9 we have
( )2100 0 1 2R R 1 100K 100 K= + + .
Hence
2100 0 1 2R R 100K 100 K− = + .
We then have
2t 0 1 2
2100 0 1 2
2 2 2 21 2 1 2 2 2
1 2 1 22 2
22 2 2 2 2
1 2 1 1
R R K t K t100 100R R 100K 100 K
K t K t K t 100K t 100K t K tK 100K K 100K
100K t K t 100K t K t K t tt t t 100 1K 100K K K 100 100
− +× = ×
− +
+ + − += =
+ +
⎛ ⎞− − ⎡ ⎤= − ≈ − = + × −⎜ ⎟ ⎢ ⎥+ ⎣ ⎦⎝ ⎠
With the K’s given earlier, we
have7
2 423
1
K 5.802 10100 10 1.485K 3.90802 10
−
−− ×
× = × = − = −δ×
. Thus we have
t 0
100 0
R R t t100 t 1R R 100 100
− ⎛ ⎞⎛ ⎞× = − δ −⎜ ⎟⎜ ⎟− ⎝ ⎠⎝ ⎠
This may be rephrased as (using the definition given in Equation 10)
t 0 Pt PtPt
100 0Correction c
R R t tt tt 100 1 t 1R R 100 100 100 100
− ⎛ ⎞⎛ ⎞⎛ ⎞⎛ ⎞≈ × + δ − = + δ −⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟− ⎝ ⎠⎝ ⎠ ⎝ ⎠⎝ ⎠ (11)
This is referred to as the Callendar equation and the second term is the
Callendar correction, represented as c. The Callendar correction is evidently
zero at both the ice and steam points. The correction is non-zero at all other
temperatures. Figure 10 shows the Callendar correction as a function of the
Platinum resistance temperature over a useful range of the sensor.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 10 Callendar correction as a function of tPt
-101234567
0 50 100 150 200 250 300
Platinum resistance temperature, tPt
Cor
rect
ion,
c
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 4
The resistance of a Platinum resistance sensor of 0R 100= Ω was
measured to be119.4Ω . This sensor has α value of 0.00385. What is
the corresponding temperature without and with correction?
o We have R0 = 100Ω, α = 0.00385 and δ = 1.485.
o Hence ( ) ( )100 0R R 1 100 100 1 100 0.00385 13.5= + α = × + × = Ω
o The measured sensor resistance is given as tR 119.4= Ω .
o By definition the Platinum resistance temperature is
Ot 0Pt
100 0
R R 119.4 100t 100 100 50.39 CR R 138.5 100
− −= × = × =
− −
o This is also the uncorrected value of the temperature.
o The Callendar correction is calculated as
OPt Ptt t 50.39 50.39c 1 1.485 1 0.37 C100 100 100 100
⎛ ⎞ ⎛ ⎞= δ − = × − × = −⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠
o The corrected temperature is thus given by
OPtt t c 50.39 0.37 50.02 C= + = − =
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
RTD measurement circuits
The resistance of the resistance sensor is determined by the use of a
DC bridge circuit. As mentioned earlier there are two variants, viz. the three
wire and the four wire systems. These are essentially used to eliminate the
effect of the lead wire resistances that may adversely affect the
measurement. There are two effects due to the lead wires: 1) they add to the
resistance of the Platinum element 2) the resistance of the lead wires may
also change with temperature. These two effects are mitigated or eliminated
by either the three or four wire arrangements.
The lead wires are usually of higher diameter than the diameter of the
sensor wire to reduce the lead wire resistance. In both the three and four wire
arrangements, the wires run close to each other and pass through regions
experiencing similar temperature fields (refer Figure 9). Hence the change in
the resistance due to temperature affects all the lead wires by similar
amounts. The resistances of the lead wires are compensated by a procedure
that is described below.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Bridge circuit for resistance thermometry:
Three wire arrangement for lead wire compensation
Figure 11 Bridge circuit with lead wire compensation
(Three wire arrangement)
Figure 11 shows the bridge circuit that is used with three lead wires.
The resistances R1 and R3 are chosen to be equal and the same as R0 of the
RTD. Two lead wires (labeled 2 and 3) are connected as indicated adding
equal resistances to the two arms of the bridge. The third lead wire (labeled
1) is used to connect to the battery. Thus the bridge will indicate null (milli-
ammeter will indicate zero) when R2 = R0 when the RTD is maintained at the
ice point. During use, when the RTD is at temperature t, the resistance R2 is
adjusted to restore balance. If the lead wires have resistances equal to Rs2
and Rs3, we have
RTD
1
3
Battery
2
Milli- Ammeter
Compensating Leads
R3
R1
R2
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
( )t s3 2 s2 t 2 s2 s3R R R R or R R R R+ = + = + − (12)
If the two lead wires are of the same size the bracketed terms should
essentially be zero and hence the lead wire resistances have been
compensated.
Four wire arrangement for lead wire compensation
The four wire arrangement is a superior arrangement, with reference to
lead wire compensation, as will be shown below. Figure 12 is the bridge
arrangements that are used for this purpose.
Figure 12 Bridge circuit with lead wire compensation
(four wire arrangement)
(a) (b)
R1
R2 R2 R3
3 42 1
Milli- Ammeter
Battery
Compensating Leads
RTD
1 243
Milli- Ammeter
Battery
Compensating Leads
RTD
R1
R3
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The choice of the resistances is made as given for the three wire
arrangement. If the lead wires have resistances equal to Rs1 - Rs4, we have
the following.
Condition for bridge balance in arrangement shown in Figure 12(a):
t s4 2(a) s2
For balancarrangement a
R R R R+ = + (12)
Condition for bridge balance in arrangement shown in Figure 12(b):
t s2 2(b) s4
For balancearrangement b
R R R R+ = + (13)
We see that by addition of Equations 5 and 6, we get
2(a) 2(b)t
R RR
2+
= (14)
The lead wire resistances thus drop off and the correct resistance is nothing
but the mean of the two measurements. Since the lead wire resistances
actually drop off, the four wire scheme is superior to the three wire scheme.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 5
An RTD has C/004.020 °=α . If 20R 106= Ω (resistance at 20°C),
determine the resistance at 25°C. The above RTD is used in a bridge
circuit with 1 2 3R R R 100= = = Ω and the supply voltage is 10 V.
Calculate the voltage the detector must be able to resolve in order to
measure a 1°C change in temperature around 20°C.
o Note the definition of α viz. tt
tt
s1 dRR dt R
α = = at any temperature.
Symbol s stands for the slope of the resistance versus temperature
curve for the sensor. (The earlier definition assumes that α is constant
and is evaluated using the resistance values at the ice and steam
points.)
o The circuit used for measurement is shown in the following figure.
R1 Voltmeter
Battery
RTD Rt
R3 R2
10 V
VA
VB
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
o With the given data of 20 0.004 / Cα = ° the slope may be determined as
20 20 20s R 0.004 106 0.424 / C= α = × = Ω °
o Assuming the response of the sensor to be linear over small changes
in temperature, the resistance of the sensor at 25°C may be
determined as
( ) Ω=×+=×−+= 12.108424.05106s2025RR 202025
o Infer all voltages with reference to the negative terminal of the battery
taken as zero (ground). The voltmeter reads the potential difference
between A and B. If there is change of temperature of 1°C the
temperature of the RTD may either be 21°C or 19°C.
o Case (a): t = 21°C. The potentials are given by the following:
A 21 2
10 10V 10 R 10 100 5 VR R 100 100
= − = − × =+ +
o If there is a change of 1.0oC in temperature the resistance changes by
0.424 Ω as given by the slope. The resistance of the RTD will be
106.424 Ω in this case.
o The potential VB is then given by
B 2121 3
10 10V 10 R 10 106.424 4.844 VR R 106.424 100
= − = − × =+ +
o The voltmeter should read
A BV V 5 4.844 0.156 V or 156 mV− = − =
o Case (b): t = 19°C. The potentials are given by the following:
V5100100100
1010RRR
1010V 221
A =×+
−=+
−=
o If there is a change of -1.0oC in temperature the resistance changes by
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
-0.424 Ω as given by the slope. The resistance of the RTD will be
105.576 Ω in this case.
o The potential VB is then given by
B 2119 3
10 10V 10 R 10 105.576 4.864 VR R 105.576 100
= − = − × =+ +
o The voltmeter should read
A BV V 5 4.864 0.136 V or 136 mV− = − =
o The smaller of these or 0.136 V or 136 mV is the resolution of the
voltmeter required for 1oC resolution. Practically speaking we may
choose a voltmeter with 100 mV resolution for this purpose.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Effect of self heating
The bridge arrangement for measuring the sensor resistance involves
the passage of a current through the sensor. Heat is generated by this
current passing through the RTD. The heat has to be dissipated by an
increase in the sensor temperature compared to the medium surrounding the
sensor. Thus the self heating leads to a systematic error. Assume that the
conductance (dissipation constant) for heat transfer from the RTD to the
surrounding medium is PD W/K. The temperature excess of the RTD is given
by CPRItD
t2
°⎥⎦
⎤⎢⎣
⎡=∆ . Example 6 demonstrates this.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 6
An RTD has 20 0.005 / Cα = ° , 20R 500= Ω and a dissipation constant of
DP 30 mW / C= ° at 20oC. The RTD is used in a bridge circuit with
1 3R R 500= = Ω and 2R is a variable resistor used to null the bridge. If
the supply voltage is 10 V and the RTD is placed in a bath at C0° , find
the value of R3 to null the bridge. Take the effect of self heating into
account.
o Note: Figure in Example 2 is appropriate for this case also.
o Given data: 320 20 D s0.005 / C, R 500 , P 30 10 W / C, V 10 V−α = ° = Ω = × ° =
o Since 1 3R R 500= = Ω , at null RTD 2R R= . Thus the current through the
RTD is ( )
s
2
VR 500+
and hence the dissipation in the RTD is
( )
2s
22
V RR 500
⎡ ⎤⎢ ⎥+⎣ ⎦
. The self heating leads to a temperature change
of( )
2s 2
2 D
V RR 500 P
⎡ ⎤⎢ ⎥+⎣ ⎦
C° . The temperature of the RTD is thus
( )
2s 2
2 D
V RR 500 P
⎡ ⎤⎢ ⎥+⎣ ⎦
C° instead of 0oC as it should have been.
o The resistance of the RTD is thus given by (assuming linear variation
of resistance with temperature)
( )
2s 2
2 20 202 D
V RR R [1 (20 ]R 500 P
⎡ ⎤= − α − ⎢ ⎥+⎣ ⎦
o This has to be solved for 2R to get the variable resistance which will
null the bridge.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
o The solution may be obtained by iteration. The iteration starts with the
trial value
02 20 20R R (1 20 ) 500 (1 0.005 20) 450= − α = × − × = Ω
o Substitute this in the right hand side of the previous expression to get
( )
20
1 s 22 20 20 0
D2
2
3
V RR R [1 (20 ]PR 500
10 450500 1 0.005 20 454.16450 500 30 10−
⎡ ⎤⎢ ⎥= − α −⎢ ⎥+⎣ ⎦
⎡ ⎤⎧ ⎫⎪ ⎪⎡ ⎤⎢ ⎥= × − × − × = Ω⎨ ⎬⎢ ⎥+⎣ ⎦ ×⎢ ⎥⎪ ⎪⎩ ⎭⎣ ⎦
o It so happens that we may stop after just one iteration! Thus the
required resistance to null the bridge is 454.16 Ω.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 7
Use the values of RTD resistance versus temperature shown in the
table to find the equation for the linear approximation of resistance
between 100 and 130oC. Assume 0T 115 C= ° .
t°C 100 105 110 115 120 125 130 R Ω 573.40 578.77 584.13 589.48 594.84 600.18 605.52
o For the linear fit we calculate the α value by using the mean slope near
the middle of the table. We use the values shown in blue to get
0594.84 584.13 0.00182
120 110−
α = =−
o The linear fit is:
( )ft 115 0R R 1 t 115= + α −⎡ ⎤⎣ ⎦
o We make a table to compare the linear fit with the data.
t Rft R(data) Difference 100 573.39 573.40 0.01 105 578.75 578.77 0.02 110 584.12 584.13 0.01 115 589.48 589.48 0.00 120 594.84 594.84 0.00 125 600.21 600.18 -0.03 130 605.57 605.52 -0.05
Linear fit appears to be very good
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Sub Module 2.4
Thermistors
Resistance thermometry may be performed using thermistors. Thermistors
are many times more sensitive than RTD’s and hence are useful over limited
ranges of temperature. They are small pieces of ceramic material made by
sintering mixtures of metallic oxides of Manganese, Nickel, Cobalt, Copper,
Iron etc. They come in different shapes as shown in Figure 13. Resistance of
a thermistor decreases non-linearly with temperature. Thermistors are
extremely sensitive but over a narrow range of temperatures. The resistance
temperature relation is known to follow the law
0
1 1T T
T 0R R e⎛ ⎞
β −⎜ ⎟⎝ ⎠= (15)
Here is a constant and all the temperatures are in Kelvin, To is the ice point
temperature and R0 the corresponding resistance of the thermistor.
Figure 13 Typical thermistor types
Disc type Thermistor
Rod type Thermistor
Bead type Thermistor
Glass Envelope
Bead
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
A typical thermistor has the following specifications:
025
70
RR 2000 , 18.64R
= Ω = (16)
This thermistor has a value of 3917 K and a resistance of 6656 at
0oC. We may compare the corresponding numbers for a standard Platinum
resistance element.
0 70 0 0R 100 , R R (1 0.00385 70) 1.2695R= Ω = + × =
Hence, we have
0
70
R 1 0.788R 1.2695
= = (17)
The resistance change is thus very mild in the case of a PRT as
compared to a thermistor. The variation of the resistance of the thermistor
described above is shown plotted in Figure 14.
Figure 14 Variation of thermistor resistance with temperature
0
1000
2000
3000
4000
5000
6000
7000
0 20 40 60 80 100Temperature, oC
Res
ista
nce,
Ω
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Typical thermistor circuit for temperature measurement
Figure 15 Typical thermistor circuit
Since a thermistor has a highly non-linear temperature response it is
necessary to use some arrangement by which the resistance variation with
temperature is nearly linear. One way of achieving this is to connect a
suitable parallel resistance as shown in Figure 15. As an example we
consider the thermistor whose characteristics are given by (2). A parallel
resistance of 500 is chosen for the simulation. The equivalent resistance of
the thermistor in parallel with Rp varies nearly linearly as shown in Figure 16.
The circuit is basically a voltage divider circuit. The potential difference
across the series resistance or the thermistor provides an output that is
related to the temperature of the thermistor.
Battery Thermistor
Voltmeter
Rs
Rp
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
0
50
100
150
200
250
300
350
400
450
500
0 10 20 30 40 50 60 70 80 90 100
Temperature, oC
Equi
vale
nt R
esis
tanc
e, Ω
Data Linear (Data)
Figure 16 Equivalent resistance variations for the circuit shown in
Figure 3
If the series resistance is chosen as sR 500= Ω and the battery voltage
is 9 V, the voltmeter reading varies as shown in Figure 17. The output
decreases monotonically with temperature and the non-linearity is mild.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0 10 20 30 40 50 60 70 80 90 100
Temperature, oC
Out
put a
cros
s th
erm
isto
r, V
Figure 17 Variation of voltage across thermistor as a function of its
temperature
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 8
A thermistor has a resistance temperature relationship given by
00
1 1R R expT T
⎡ ⎤⎛ ⎞= β −⎢ ⎥⎜ ⎟
⎝ ⎠⎣ ⎦. For a certain thermistor R0 =80000 Ω where
T0 = 273.16 K. The resistance of the thermistor has been measured
accurately at three other temperatures as given below:
t,°C 50 100 150
T, K 323.16 373.16 423.16
R, Ω 10980 2575 858
Using the above data estimate β. Use this β to estimate the resistance
of the thermistor at 10 and 110 oC. Compare the data with the values
calculated using the β determined above.
o Given Data: T0 = 273.16 K, R0 = 80000 Ω
o Taking logarithms and rearranging, we have
( )[ ]
0
0
ln R R1 T 1/ T
β =−
o The tabulated data in the above expression will thus give three
values for β. The average of these three values should give the
best estimate for β. The values of β are 3506, 3503 and 3495 K.
The mean value is thus equal to 3501 K (whole number retained).
o Now we calculate the resistances at the three temperatures using
the value of β obtained above. The data is conveniently tabulated.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
t, °C 50 100 150
R (measured), Ω 10980 2575 858
R (calculated), Ω 11070 2602 861
o The match is very good!
Sub Module 2.5
Pyrometery
Pyrometry is the art and science of measurement of high temperatures.
According to the International Practical Temperature Scale 1968 (IPTS68)
Pyrometry was specified as the method of temperature measurement above
the gold point. Pyrometry makes use of radiation emitted by a surface
(usually in the visible part of the spectrum) to determine its temperature. The
measurement is thus a non-contact method of temperature measurement.
We shall introduce basic concepts from radiation theory before discussing
Pyrometry in detail.
Radiation fundamentals
Black body radiation exists inside an evacuated enclosure whose walls
are maintained at a uniform temperature. The walls of the enclosure are
assumed to be impervious to heat transfer. The black body radiation is a
function of the wavelength λ and the temperature T of the walls of the
enclosure. The amount of radiation heat flux leaving the surface, in a narrow
band dλ aroundλ, of the enclosure is called the spectral emissive power and
is given by the Planck distribution function
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
2
1b 5 C
T
C 1E (T)
e 1λ
λ
=λ
−
(18)
C1=First radiation constant = 3.742 x 108 W µm4 / m2 and C2 = Second
radiation constant = 14390 µm K. It is seen that the wavelength is specified in
µm (= 10-6 m). It is to be noted that there is no net heat transfer from the
surface of an enclosure and hence it receives the same flux as it emits. It
may also be seen that the -1 in the denominator is much smaller than the
exponential term as long as λT<<C2. This is indeed true in Pyrometry
applications where the wavelength chosen is 0.65 µm and the measured
temperature may not be more than 5000 K. It is then acceptable to
approximate the Planck distribution function by the Wein’s approximation
given by
2C1 T
b 5CE (T) e
−λ
λ =λ
(19)
Figure 18 Error in using approximate Wein’s approximation
instead of the Planck function
It is clear from Figure 18 that the error in using the Wein’s
approximation in lieu of the Planck function is around 1% even at a
temperature as high as 5000 K.
0.00E+00
2.00E-01
4.00E-01
6.00E-01
8.00E-01
1.00E+00
1.20E+00
1.40E+00
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000Temperature, K
% D
iffer
ence
Wavelength = 0.65 M icron
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The spectral Black body emissive power has strong temperature
dependence. In fact the emissive power peaks at a wavelength temperature
product of
maxT 2898 m Kλ = µ − (20)
This is referred to as Wein’s displacement law. Figure 19 shows this
graphically.
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
1.E+08
0.1 1 10 100
Wavelength , mm
Mon
ochr
omat
ic e
mis
sive
pow
er E
b, W
/m2
m
T = 300 K
T = 5800 K
T = 1500 K
T = 2898 m K
Figure 19 Black body characteristics and the Wein’s displacement law
(Red line corresponds to a wavelength of 0.65 µm normally used in
Pyrometry)
If you imagine keeping the wavelength fixed at say 0.65 µm, we see
that the ordinate is a strong function of temperature! The surface will appear
brighter to the eye higher the temperature. This is basically the idea central to
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Pyrometry. Actual surfaces, however, are not black bodies and hence they
emit less radiation than a black surface at the same temperature. We define
the spectral emissivity λε as the ratio of the emissive power of the actual
surface to that from a black surface at the same temperature and wavelength.
a
b
E (T)E (T)
λλ
λε = (21)
Brightness temperature
It is defined such that the spectral emissive power of the actual surface
is the same as that of a hypothetical black body at the brightness temperature
TB. Thus:
a b BE (T) E (T )λ λ= (22)
If the emissivity of the surface (referred to as the target) is λε , we use
Equation 22 to write b b BE (T) E (T )λ λ λε = . Using Wein’s approximation this may
be re written as22B
CCT1 1T
5 5C Ce e
−− λλλε =
λ λ. We may cancel the common factor on
the two sides, take natural logarithms, and rearrange to get
( )B 2
1 1 lnT T C λ
λ− = ε (23)
Equation 23 is referred to as the ideal pyrometer equation. This
relation relates the actual temperature of the target to the brightness
temperature of the target. The brightness temperature itself is measured
using a vanishing filament pyrometer. Equation 23 indicates that BT T≤ since
1λε ≤ for any surface. In actual practice the intervening optics may introduce
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
attenuation due to reflection of the radiation gathered from the target. It is
also possible that one introduces attenuation intentionally as we shall see
later. We account for the attenuation by multiplying the emissivity of the
surface by a transmission factor 1λτ ≤ to get
( )B 2
1 1 lnT T C λ λ
λ− = ε τ (24)
We refer to Equation 28 as the pyrometer equation.
We infer from the above that the brightness temperature of a surface depends
primarily on the surface emissivity. If a surface is as bright as a black body at
the gold point, the actual temperature should vary with spectral emissivity of
the surface as indicated in Figure 20. Since no surface has zero emissivity
we allow it to vary from 0.02 to 1 in this figure. As the spectral emissivity
decreases the actual temperature increases as shown.
1300
1400
1500
1600
1700
1800
0 0.2 0.4 0.6 0.8 1
Spectral emissivity, ελ
Act
ual t
empe
ratu
r T, K
Figure 20 Temperature a surface whose brightness temperature equals
gold point temperature of 1337.6 K. λ = 0.66 µm.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 20 also may be interpreted in a different way. Consider an
actual surface whose spectral emissivity is known. If we introduce a
transmission element with variable attenuation before making a comparison
with a reference black body at the gold point, the abscissa in Figure 20 may
be interpreted as the spectral emissivity transmission factor product. In that
case the comparison is with respect to a single reference temperature, the
gold point temperature. If this fixed point is determined with great precision
on the ideal gas scale we have achieved measurement of an arbitrary
temperature higher than the gold point temperature with reference to this
single fixed point. This has the advantage that the pyrometer reference
source (usually a standard tungsten filament lamp) runs at a constant
temperature resulting in long life for the lamp.
Black body reference – Cavity radiator
Pyrometry requires a reference black body source that may be
maintained at a desired temperature. This is achieved by making use of a
black body cavity shown schematically in Figure 21.
Figure 21 Schematic of a black body furnace
Electrically Heated Refractory Sphere 0.3 m diameter
Temperature Controller
Insulation
Temperature Sensor
Opening – 50 mm diameter, 150 mm long
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The black body reference consists of an electrically heated refractory
sphere with a small opening as shown. The surface area of the sphere is
much larger than the area of the opening through which radiation will escape
to the outside. The radiation leaving through the opening is very close to
being black body radiation at a temperature corresponding to the temperature
of the inside of the sphere. Figure 22 shows that the effective emissivity of
the opening is close to unity. The emissivity of the surface of the sphere is
already high, 0.96, for the case shown in the figure). The area ratio for the
case shown in Figure 21 is
2opening
sphere
A 1 0.05 0.007A 4 0.3
⎛ ⎞= ≈⎜ ⎟⎝ ⎠
(25)
The effective emissivity for this area ratio is about 0.995!
0.9
0.92
0.94
0.96
0.98
1
0 0.2 0.4 0.6 0.8 1
Ratio of Opening Area to Cavity Surface Area
Eff
ect
ive
E
mis
siv
ity
Figure 22 Variation of effective emissivity of a cavity
radiator with area ratio
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Many a time the cavity radiator has the sphere surrounded on the
outside by a material undergoing phase change (solid to liquid). The melting
point of the material will decide the exact temperature of the black body
radiation leaving through the opening, as long as the material is a mixture of
solid and liquid.
Vanishing filament pyrometer
This is fairly standard equipment that is used routinely in industrial
practice. The principle of operation of the pyrometer is explained by referring
to the schematic of the instrument shown in Figure 23.
Figure 23 Schematic of a vanishing filament pyrometer
The pyrometer consists of collection optics (basically a telescope) to
gather radiation coming from the target whose temperature is to be estimated.
The radiation then passed through an aperture (to reduce the effect of stray
radiation), a neutral density or grey filter (to adjust the range of temperature)
and is brought to focus in a plane that also contains a source (tungsten
filament standard) whose temperature may be varied by varying the current
through it. The radiation from the target and the reference then passes
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
through a red filter and is viewed by an observer as indicated in Figure 23.
The observer will adjust the current through the reference lamp such that the
filament brightness and the target brightness are the same. The state of
affairs in the image seen by the eye of the observer is shown in Figure 24.
Figure 24 Pyrometer adjustments
If the adjustment is such that the filament is brighter than the target the
setting is referred to as “high”. The filament appears as a bright object in a
dull background. If the adjustment is such that the filament is duller than the
target the setting is referred to as “low”. The filament appears as a dull object
in a bright background. If the adjustment is “correct” the filament and the
background are indistinguishable. Thus the adjustment is a null adjustment.
The filament vanishes from the view! In this setting we have a match between
the brightness of the target and the filament. The temperature of the filament
is in deed the brightness temperature of the object. The wavelength
corresponds to roughly a value of 0.66 µm. This is achieved by a combination
of the response of the red filter and the eye of the observer (Figure 25). The
average eye peaks around 0.6 µm and responds very poorly beyond 0.67 µm.
The red filter transmits radiation beyond about 0.62 µm. On the whole the red
High Correct Low
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
filter – eye combination uses the radiation inside the red triangle shown in
Figure 25. This region corresponds to a mean 0f 0.66 µm with a spectral
width of approximately 0.03 µm.
00.10.20.30.40.50.60.70.80.9
1
0.45 0.5 0.55 0.6 0.65 0.7 0.75Wavelength, Micron
Tra
nsm
ittan
ce
Red FilterEye Response
Figure 25 Effective wavelength for the pyrometer
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 9
A pyrometer gives the brightness temperature of an object to be
800oC. The optical transmittance for the radiation collected by the
pyrometer is known to be 0.965 and the target emissivity is 0.260.
Estimate the temperature of the object. Take λ = 0.655 µm as the
effective wavelength for the pyrometer.
Given data:
2 b0.655 m, C 14390 m K, 0.965, T 800 C 1073 Kλ = µ = µ − τ = = ° =
Target emissivity is
0.260ε =
The emissivity transmittivity product is
0.260 0.965 0.251ετ = × =
Using the pyrometer equation, we get:
( ) ( )
a
b 2
1 1T 1151 K1 1 0.655ln ln 0.251T C 1073 14390
= = =λ
+ ετ + ×
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 10
The brightness temperature of a metal block is given as 900°C. A
thermocouple embedded in the block reads 1015°C. What is the
emissivity of the surface? The pyrometer used in the above
measurement is a vanishing filament type with an effective λ of 0.65
µm. Assuming that the thermocouple reading is susceptible to an error
of ±10°C while the brightness temperature is error free determine an
error bar on the emissivity determined above.
o The first and second radiation constants, in SI units are:
16 21 2C 3.743 10 W m , C 14387 m K−= × = µ −
o The data specifies the actual (Ta) as well as the brightness (Tb)
temperatures and it is desired to determine the emissivity of the metal
block at the stated wavelength.
a b0.65 m, T 1015 C 1288K,T 900 C 1173 Kλ = µ = ° = = ° =
o We make use of the pyrometer equation to estimate ε.
a b
2
1 1 1 1T T 1288 1173exp exp 0.1850.65
C 14387
⎡ ⎤ ⎡ ⎤− −⎢ ⎥ ⎢ ⎥⎢ ⎥ε = = =⎢ ⎥λ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎣ ⎦⎣ ⎦
o This is the nominal value of the emissivity.
o The error in ε is due to error in Ta. This is calculated using the
standard method discussed in the class. We need to calculate the
derivative of ε with respect to Ta.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
a b2 22
a a a b a2
2
1 1T TC Cd d 1 1 1exp ( )
dT dT T T TC
14387 1 0.185 0.00247 / K0.65 1288
⎡ ⎤−⎢ ⎥⎡ ⎤⎧ ⎫ε ⎢ ⎥= − × = −⎢ ⎨ ⎬⎥ λλ λ⎢ ⎥⎩ ⎭⎣ ⎦⎢ ⎥⎣ ⎦
= − × × = −
o Since K10Ta ±=∆ , the error in emissivity is given by
aa
d T 0.00247 10 0.025dT
ε∆ε = ∆ = − × ± ≈ ∓
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Emissivity values
Temperature measurement using a pyrometer requires that the surface
emissivity of the target be known. Useful emissivity data is given in Table 5.
These are representative values since the nature of the surface is application
specific and much variation is possible. One way out of this is to perform an
experiment like the one presented in Example 2.
Table 5 Approximate emissivity values
Temperature °C
Surface 600 1200 1600 1800
Iron, Un-oxidized 0.2 0.37
Iron, Oxidized 0.85 0.89
Molten Cast Iron 0.29
Molten Steel 0.28 0.28
Nickel, Oxidized 0.75 0.75
Fireclay 0.52 0.45
Silica Bricks 0.54 0.46
Alumina Bricks 0.23 0.19
In fact one may vary the surface temperature over a range of values
that is expected to occur in a particular application and the measure the
emissivity values over this range. An emissivity table may then be made for
later use. Uncertainty in emissivity affects the measurement of temperature
using a pyrometer. Figure 26 shows a plot of temperature error as a function
of emissivity error, both in percent. For small errors the relationship is almost
linear with aT 0.65∆ = ∆ε . The case considered is a vanishing filament
pyrometer operating at 0.65 µm.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
0
2
4
6
8
10
0 2 4 6 8 10Error in Emissivity, %
Err
or
in
Tem
pera
ture
, %
Figure 26 Effect of uncertainty in emissivity on pyrometer
measurement of temperature
Ratio Pyrometry and the two color pyrometer
Uncertainty in target emissivity in the case of vanishing filament
pyrometer is a major problem. This makes one look for an alternate way of
performing pyrometric measurement, based on the concept of color or ratio
temperature. Consider two wavelengths 1 2andλ λ close to each other. Let
the corresponding emissivity values of the target be 21 andεε . The color
temperature Tc of the target is defined by the equality of ratios given by
1 1
2 2
a a b c
a a b c
E (T ) E (T )E (T ) E (T )
λ λ
λ λ= (26)
Thus it is the temperature of a black body for which the ratio of spectral
emissive powers at two chosen wavelengths λ1 and λ2 is the same as the
corresponding ratio for the actual body.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
We may use the Wein’s approximation and write the above as
2 21
1 a 1 c 1 2 2
2 a 1 2 b 1 22 22
2 a 2 c
C Cexp expT T C C1 1 1 1or exp exp
T TC Cexp expT T
⎡ ⎤ ⎡ ⎤ε − −⎢ ⎥ ⎢ ⎥ ⎡ ⎤ ⎡ ⎤λ λ ⎧ ⎫ ⎧ ⎫ε⎣ ⎦ ⎣ ⎦= − − = − −⎨ ⎬ ⎨ ⎬⎢ ⎥ ⎢ ⎥ε λ λ λ λ⎡ ⎤ ⎡ ⎤ ⎩ ⎭ ⎩ ⎭⎣ ⎦ ⎣ ⎦ε − −⎢ ⎥ ⎢ ⎥λ λ⎣ ⎦ ⎣ ⎦
Taking natural logarithms, we have
1 2 2
2 a 1 2 c 1 2
C C1 1 1 1lnT T
⎡ ⎤ ⎡ ⎤ε− − = − −⎢ ⎥ ⎢ ⎥ε λ λ λ λ⎣ ⎦ ⎣ ⎦
This may be rearranged as
1
2
a c 2
1 2
ln1 1 1
1 1T T C
εε
− =−
λ λ
(27)
This equation that links the color and actual temperatures of the target
is the counterpart of Equation 23 that linked the brightness and actual
temperatures of the target. In case the emissivity does not vary with the
wavelength we see that the color and actual temperatures are equal to each
other. We also see that the actual temperature may be either less than or
greater than the color temperature. This depends solely on the ratio of
emissivities at the two chosen wavelengths. Equation 27 by itself is not
directly useful. We measure the ratio of emissive powers of the target, at its
actual temperature, at the two chosen wavelengths. Thus what is measured
is the ratio occurring on the left hand side of Equation 26. Thus we have
1
2
25
a 1 a1 25
a 2 21
2 a
CexpE TE Cexp
T
λ
λ
⎡ ⎤⎢ ⎥λε λ ⎣ ⎦=
ε ⎡ ⎤λ⎢ ⎥λ⎣ ⎦
Again we may take natural logarithms to get
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
1
2
22 1
a 5a 1 2
5a 2 1
1 1CT
Eln
Eλ
λ
⎛ ⎞−⎜ ⎟λ λ⎝ ⎠=
⎡ ⎤λ ε⎢ ⎥
λ ε⎢ ⎥⎣ ⎦
(28)
Similarly we may show that
1
2
22 1
c 5a 1
5a 2
1 1CT
Eln
Eλ
λ
⎛ ⎞−⎜ ⎟λ λ⎝ ⎠=
⎡ ⎤λ⎢ ⎥
λ⎢ ⎥⎣ ⎦
(29)
Equation 28 is useful as a means of estimating the actual temperature
for a target, whose emissivities at the two chosen wavelengths are known,
and the measured values of emissive powers at the two wavelengths are
available. The instrument that may be used for this purpose is the two color
pyrometer whose schematic is shown in Figure 27.
Figure 27 Schematic of a two color pyrometer
SM
Amplifier
Amplifier
SC Display
HT O
M2
M1 RF
Sh BF
Ph2
Ph1
W
HT – hot target; O – objective lens; W – wedge; M1 and M2 – mirrors; Sh – shutter; SM – synchronous motor; RF – red filter;
BF – blue filter; Ph1, Ph2 – photocells; SC – scaler
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The radiation from the target is split up in to two beams by the use of a
wedge, as indicated. Two mirrors redirect the two beams as shown. These
pass through a rotating wheel with openings, as indicated. The beams then
pass through two filters that allow only a narrow band around a well defined
wavelength (indicated as red and blue). The rotating wheel chops the two
beams and creates ac signal to be detected by the photocells. The ratio of
the two signals is equal to the ratio of emissive powers in Equation 28. The
scaler introduces the other ratios that appear in the same equation. The
display then indicates the actual temperature.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 11
A certain target has a brightness temperature of 1000 K when viewed
by a vanishing filament pyrometer. The target emissivity at 0.66 m is
known to be 0.8. What is the true temperature of the target? What is
the colour temperature of the same target
if21 20.66 m, 0.5 m, 0.50λλ = µ λ = µ ε = ?
Given Data: 1B 1T 1000 K, 0.66 m, 0.8λ= λ = µ ε =
The ideal pyrometer equation may be recast to read as
( )1
a1
B 2
1T 1 lnT C λ
=λ
+ ε
Thus the true target temperature is
( )a
1T 1010.3 K1 0.66 ln 0.81000 14390
= =+
For determining the colour temperature the required data is:
1
2
1
2
0.66 m, 0.8
0.5 m, 0.5λ
λ
λ = µ ε =
λ = µ ε =
We determine the color temperature by using the relation
2
1
c
a2
2 1
1T
ln1T 1 1C
1 969.7K0.76ln
1 0.861 11010.3 14390
0.5 0.66
λ
λ
=⎡ ⎤⎛ ⎞ε⎢ ⎥⎜ ⎟⎜ ⎟ε⎢ ⎥⎝ ⎠−⎢ ⎥⎛ ⎞⎢ ⎥−⎜ ⎟⎢ ⎥λ λ⎝ ⎠⎣ ⎦
= =⎡ ⎤⎛ ⎞
⎜ ⎟⎢ ⎥⎝ ⎠⎢ ⎥−⎛ ⎞⎢ ⎥× −⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Sub Module 2.6
Measurement of transient temperature
Many processes of engineering relevance involve variations with
respect to time. The system properties like temperature, pressure and flow
rate vary with time. These are referred to as transients and the measurement
of these transients is an important issue while designing or choosing the
proper measurement technique and the probe. Here we look at the
measurement of temperature transients.
Temperature sensor as a first order system - Electrical analogy
Let us look at a typical temperature measurement situation. We
visualize the temperature probe as a system that is subject to the temperature
transient. The probe is exposed to the environment whose temperature
changes with time and it is desired to follow the temperature change as
closely as possible. In Figure 28 we show the schematic of the thermal model
appropriate for this study.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 28 Schematic of a temperature probe placed in a flowing medium
The model assumes that the probe is at a uniform temperature within it
at any time t. This means that the probe is considered to be thermally
lumped. The medium that flows over the probe is at a temperature that may
vary with respect to time. Initially the probe is assumed to be at temperature
T0. Let us assume that the probe is characterized by the following physical
parameters:
Density of the probe material = ρ kg/m3, Volume of the probe = V m3,
Surface area of the probe that is exposed to the flowing fluid = S m2, Specific
heat of the probe material = C J/kg°C, Heat transfer coefficient for heat
transfer between the probe and the surrounding medium = h W/m2°C. By
conservation of energy, we have
Rate of change of rate of heat transfer betweenint ernal energy of probe the probe and the fluid
⎡ ⎤ ⎡ ⎤=⎢ ⎥ ⎢ ⎥
⎣ ⎦ ⎣ ⎦ (30)
Probe at uniform temperature T(t), Characteristic length Lch.
Fluid stream at T∞(t)°C and h W/m2 °C
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
If we assume that the probe is at a higher temperature as compared to
the fluid heat transfer will be from the probe to the fluid and the internal
energy of the probe will reduce with time. Using the properties of the probe
introduced above, the left hand side of Equation 30 is given by dTVCdt
−ρ . The
right hand side of Equation 30 is given by hS(T T )∞− . With these, after some
rearrangement, Equation 30 takes the form
dT hS hST Tdt VC VC ∞+ =
ρ ρ (31)
Note that this equation holds even when the probe temperature is
lower than the fluid temperature. The quantity VChS
ρ has the unit of time and
is referred to as the time constant τ of the first order system (first order since
the governing differential equation is a first order ordinary differential
equation). The first order time constant involves thermal and geometric
properties. The volume to surface area ratio is a characteristic length
dimension and is indicated as Lch in Figure 28. Noting that the product of
density and volume is the mass M of the probe, the first order time constant
may also be written as MChS
τ = . The time constant may be interpreted in a
different way also, using electrical analogy. The quantity MC represents the
thermal capacity and the quantity 1hS
represents the thermal resistance.
Based on this interpretation an electric analog may be made as shown in
Figure 29.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 29 Electrical analog of a first order thermal system
In the electric circuit shown in Figure 29 the input voltage represents
the temperature of the fluid, the output voltage represents the temperature of
the probe, the resistance R represents the thermal resistance and the
capacitance C represents the thermal mass (mass specific heat product) of
the probe. Equation 31 may be rewritten as
TdT Tdt
∞+ =τ τ
(32)
Note that Equation 32 may be simplified using the integrating factor τt
e to write
it as
t td Te T edt
τ τ∞
⎛ ⎞⎜ ⎟ =⎜ ⎟⎝ ⎠
(33)
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
This may be integrated to get t tt
0
Te T e dt Aτ τ∞= +∫ where A is a constant of
integration. Using the initial condition 0T(t 0) T= = we get, after minor
simplification
t t tt
00
T T e e T e dt− −
τ τ τ∞= + ∫ (34)
This is the general solution to the problem. If the variation of fluid
temperature with time is given, we may perform the indicated integration to
obtain the response of the probe as a function of time.
Response to step input
If the fluid temperature is constant but different from the initial
temperature of the probe, the solution is easily shown to be represented by
t
0
T T eT T
−∞ τ
∞
−= φ =
− (35)
The temperature difference between the probe and the fluid
exponentially decreases with time. The variation is indicated in Figure 30. At
the end of one time constant the temperature difference is some 37% of the
initial temperature difference. After about 5 time constants the temperature
difference is quite negligible.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1 2 3 4 5
t/τ
φ
0.367
Figure 30 Response of a first order system to a step input
A step input may be experimentally realized by heating the probe to an
initial temperature in excess of the fluid temperature and then exposing it
quickly to the fluid environment. The probe temperature is recorded as a
function of time. If it is plotted in the form )ln(φ as a function of t, the slope of
the line is negative reciprocal of the time constant. In fact, this is one method
of measuring the time constant. Example 12 shows how this is done.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 12
A temperature probe was heated by immersing it in boiling water and is
then quickly transferred in to a fluid medium at a temperature of Tamb =
25°C. The temperature difference between the probe and the medium
in which it is immersed is recorded as given below:
t (s) 0.35 0.6 0.937 1.438 2.175 3.25 T-Tamb 60 50 40 30 20 10
What is the time constant of the probe in this situation?
o The data is plotted on a semi-log graph as shown here. It is seen that
it is well represented by a straight line whose equation is given as an
inset in the plot. EXCEL was used to obtain the best fit.
Semi-log plot
ln(T-Tamb) = -0.6065t + 4.2837R2 = 0.9989
00.5
11.5
22.5
33.5
44.5
0 0.5 1 1.5 2 2.5 3 3.5
t,s
ln(T
-Tam
b)
ln(T - Tamb) Linear (ln(T - Tamb))
o The slope of the line is -0.6065 and hence the time constant is
1 1 1.6487 1.65 sslope 0.6065
τ = − = − = ≈−
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
o The correlation coefficient of the linear fit is -0.9994. This shows that
the data has been collected carefully.
A note on time constant
It is clear from our discussion above that the time constant of a system
(in this case the temperature probe) is not s property of the system. It
depends on parameters that relate to the system as well as the parameters
that define the interaction between the system and the surrounding medium
(whose temperature we are trying to measure, as it changes with time). The
time constant is the ratio of thermal mass of the system to the conductance
(reciprocal of the thermal resistance) between the system and the medium. It
is also clear now how we can manipulate the time constant. Thermal mass
reduction is one possibility. The other possibility is the reduction of the
thermal resistance. This may be achieved by increasing the interface area
between the system and the medium. In general this means a reduction in
the characteristic dimension Lch of the system. A thermocouple attached to a
thin foil will accomplish this. The characteristic dimension is equal to half the
foil thickness, if heat transfer takes place from both sides of the foil. Another
way of accomplishing this is to use very thin thermocouple wires so that the
bead at the junction has very small volume and hence the thermal mass.
Indeed these are the methods used in practice and thin film sensors are
commercially available.
Response to a ramp input
In applications involving material characterization heating rate is
controlled to follow a predetermined program heating. The measurement of
the corresponding temperature is to be made so that the temperature sensor
follows the temperature very closely. Consider the case of linear heating and
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
possibly linear temperature rise of a medium. Imagine an oven being turned
on with a constant amount of electrical heat input. We would like to measure
the temperature of the oven given by
( ) 0T t T R t∞ = + (36)
The general solution to the problem is given by (using Equation 34)
t tt t t0
0 0
T RT e A e dt te dtτ τ τ= + +τ τ∫ ∫ (37)
where A is a constant of integration. The first integral on the right hand side is
easily obtained as ( )te 1 .ττ − Second integral on the right hand side is
obtained by integration by parts, as follows.
( )t ttt t t t 2 t
00 0
t e dt t e e dt t e e 1τ τ τ τ τ= τ − τ = τ − τ −∫ ∫ (38)
If the initial temperature of the first order system is iT , then iTA = , since both
the integrals vanish for t = 0 (the lower and upper limit will be the same). On
rearrangement, the solution is
( ) ( ) ( )t
i 0 0T t T T R e T R t R− τ= − + τ + + − τ (39)
We notice that as t → ∞ the transient part tends to zero (transient part is the
exponential decaying part) and the steady part (this part survives for t >> )
yields
oT Rt T(t) R+ − = τ . (40)
The steady state response has a lag equal to R with respect to the input.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
2025303540455055
0 20 40 60 80 100
Time t, s
Te
mp
era
ture
T,
oC
Response
Input
Figure 31 Typical response of a first order system to ramp input
Figure 31 shows the response of first order system to a ramp input.
The case shown corresponds to Ti= 20°C, T0 = 35°C, R = 0.15°C/s and τ = 10
s. For t →∞ (t > 5τ = 50 s) the probe follows the linear temperature rise with a
lag of Rτ = 0.15×10 = 1.5°C. In this case it is advisable to treat this as a
systematic error and add it to the indicated temperature to get the correct
oven temperature.
Response to a periodic input
There are many applications that involve periodic variations in
temperature. For example, the walls of an internal combustion engine
cylinder are exposed to periodic heating and hence will show periodic
temperature variation. Of course, the waveform representing the periodic
temperature variation may be of a complex shape (non sinusoidal). In that
case the waveform may be split up in to its Fourier components. The
response of the probe can also be studied as that due to a typical Fourier
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
component and combine such responses to get the actual response. Hence
we look at a periodic sinusoidal input given by
( )aT T cos t∞ = ω (41)
In the above expression Ta is the amplitude of the input wave and is
the circular frequency. We may use the general solution given by Equation 34
and perform the indicated integration to get the response of the probe. The
steps are left as exercise to the student. Finally the response is given by
( )( )( )
t1
at ao 2 2 2 2
Steady state responseTransient response
T cos t tanT eT T e1 1
− −τ
− τω − ωτ
= − ++ ω τ + ω τ
(42)
Again for large t, the transient terms drop off and the steady sate
response survives. There is a reduction in the amplitude of the response and
also a time lag with respect to the input wave. Amplitude reduction and the
time lag (or phase lag) depend on the product of the circular frequency and
the time constant. The variations are as shown in Figure 32.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.01 0.1 1 10 100Circular frequency time constant product, ωτ
Amplitude reduction factor Phase lag/ (90 degree)
Figure 32 Response of a first order system to periodic input
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
In order to bring out the features of the response of the probe, we make a plot
(Figure 33) that shows both the input and output responses, for a typical case.
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
0 5 10 15 20
Time, s
Tem
pera
ture
ratio
T/T
0
Output response
Input
Time lag
Figure 33 Response of a first order system to periodic input
The case shown in Figure 33 corresponds
to a
o
T 0.25; 1rad s and 1 sT
= ω = τ = . The output response has an initial
transient that adjusts the initial mismatch between the probe temperature and
the imposed temperature. By about 4 to 5 time constants (4 to 5 seconds
since the time constant has been taken as 1 second) the probe response has
settled down to a response that follows the input but with a time lag and an
amplitude reduction as is clear from Figure 33.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 13
The time constant of a first order thermal system is given as 0.55 min. The
uncertainty in the value of the time constant is given to be ± 0.01 min. The
initial temperature excess of the system over and above the ambient
temperature is 45oC. It is desired to determine the system temperature
excess and its uncertainty at the end of 50 s from the start.
Hint: It is known that the temperature excess follows the formula
( )( )
tT te
T 0−
τ= where T(t) is the temperature excess at any time t, T(0) is
the temperature excess at t = 0 and τ is the time constant.
o We shall convert all times given to s so that things are consistent. The
time constant is 0.55 min 0.55 60 s 33 sτ = = × =
o We need the temperature excess at s50t = from the start. Hence
( ) ( )t 50
33T 50 T 0 e 45e 9.89 C− −
τ= = = °
o We would like to calculate the uncertainty in this value. We shall
assume that this is due to the error in the time constant alone.
0.1 min 0.1 60 0.6 s∆τ = ± = ± × = ±
o The influence coefficient Iτ is given by
5033
2t 50
T 50I 45 e 0.454 C / s33
−τ
=
∂ ⎛ ⎞= = × = °⎜ ⎟∂τ ⎝ ⎠
Hence the uncertainty in the estimated temperature excess is:
( )T 50 I 0.454 0.6 0.272 Cτ∆ = ± ∆τ = ± × = °
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 14
o A certain first order system has the following specifications:
Material: copper shell of wall thickness 1 mm, outer
radius 6 mm
Fluid: Air at 30°C
Initial temperature of shell: 50°C
o How long should one wait for the temperature of the shell to reach
40°C? Assume that heat transfer is by free convection. Use suitable
correlation (from a heat transfer text) to solve the problem.
o Heat transfer coefficient calculation:
Heat transfer between the shell and the air is by natural convection. The
appropriate correlation for the Nusselt number is given by 4/1Ra43.02Nu +=
where Ra is the Rayleigh number. The characteristic length scale is the
sphere diameter. The air properties are calculated at the mean temperature
at t = 0.
From the given data, we have
mD 12 mm 0.012 m, T (50 30) / 2 40 C= = = + = °
The air properties required are read off a table of properties:
6 216.96 10 m / s, Pr 0.71, k 0.027 W / m C−ν = × = = °
The isobaric compressibility of air is calculated based on ideal gas
assumption. Thus 3 1
amb
1 1 3.3 10 KT 30 273
− −β = = = ×+
The temperature difference for calculating the Rayleigh number is taken
as the mean shell temperature during the cooling process minus the ambient
temperature. We are interested in determining the time to cool from 50 to
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
40°C. Hence the mean shell temperature is Shell50 40T 45 C
2+
= = ° . The
temperature difference is Shell ambT T T 45 30 15 C∆ = − = − = ° . The value of the
acceleration due to gravity is taken as 2g 9.8 m / s= . The Rayleigh number is
then calculated as
( )3 3 3
2 26
g TD 9.8 3.3 10 15 0.012Ra Pr 0.71 206916.96 10
−
−
β∆ × × × ×= = × =
ν ×
o The Nusselt number is then calculated as
1/ 4 1/ 4Nu 2 0.43Ra 2 0.43 2069 4.9= + = + × =
The heat transfer coefficient is then calculated as
2Nu k 4.9 0.027h 11.07 W / m KD 0.012
×= = =
o Time constant calculation:
Copper shell properties are
38954 kg / m , C 383.1 J / kg Cρ = = °
Copper shell thickness is 0.001 mδ =
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Mass of the copper shell is calculated as
2 2 3M D 8954 0.012 0.001 4.051 10 kg−= ρπ δ = × π× × = ×
Surface area of shell exposed to the fluid is
2 2 4 2S D 0.012 4.524 10 m−= π = π× = ×
The time constant is then estimated as
3
4MC 4.051 10 383.1 310 shS 11.07 4.524 10
−
−× ×
τ = = =× ×
o Cooling follows an exponential process. Hence we have, the time t40 at
which the shell temperature is 40°C,
4040 30t 310ln 214.9 s50 30
−⎛ ⎞= − =⎜ ⎟−⎝ ⎠
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Sub Module 2.7
Systematic errors in temperature measurement
Systematic errors are situation dependent. We look at typical
temperature measurement situations and discuss qualitatively the errors
before we look at the estimation of these. The situations of interest are:
Measurement of temperature
o at a surface
o inside a solid
o of a flowing fluid
Surface temperature measurement using a compensated probe:
Consider the measurement of the temperature of a surface by
attaching a thermocouple sensor normal to it, as shown in Figure 34.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 34 Temperature measurement of a surface
Lead wires conduct heat away from the surface and this is
compensated by heat transfer to the surface as shown. This sets up a
temperature field within the solid such that the temperature of the surface
where the thermocouple is attached is depressed and hence less than the
surface temperature elsewhere on the surface. This introduces an error in the
surface temperature measurement. One way of reducing or altogether
eliminating the conduction error is by the use of a compensated sensor as
indicated in Figure 35.
Lead Wire Conduction
Ts
Heat flux paths
Tt < Ts
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 35 Schematic of a compensated probe
This figure is taken from “Industrial measurements with very short
immersion & surface temperature measurements” by Tavener et al. The
surface temperature, in the absence of the probe is at an equilibrium
temperature under the influence of steady heat loss H3 to an environment.
The probe would involve as additional heat loss due to conduction. If we
supply heat H2 by heating the probe such that there is no temperature
gradient along the thermocouple probe then H1-H2 =H3 and the probe
temperature is the same as the surface temperature. Figure 3 (taken from the
same reference) demonstrates that the compensated probe indicates the
actual surface temperature with negligible error.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 36 Comparison between the thermally compensated probe
and two standard probes
Compensated probes as described above are commercially available
from ISOTECH (Isothermal Technology Limited, Pine Grove, Southport,
Merseyside, England) and described as 944 True Surface temperature
measurement systems.
Figure 37 shows how one can arrange a thermocouple to measure the
temperature inside a solid. The thermocouple junction is placed at the bottom
of a blind hole drilled into the solid. The gap between the thermocouple lead
wires and the hole is filled with a heat conducting cement. The lead wire is
exposed to the ambient as it emerges from the hole. It is easy to visualize
that the lead wire conduction must be compensated by heat conduction into
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
the junction from within the solid. Hence we expect the solid temperature to
be greater than the junction temperature (this is the temperature that is
indicated) which is greater than the ambient temperature. This assumes that
the solid is at a temperature higher than the ambient.
Figure 37 Measurement of temperature within a solid
Often it is necessary to measure the temperature of a fluid flowing through
a duct. In order to prevent leakage of the fluid or prevent direct contact
between the fluid and the temperature sensor, a thermometer well is provided
as shown in Figure 38. The sensor is attached to the bottom of the well as
indicated. The measured temperature is the temperature of the bottom of the
well and what is desired to be measured is the fluid temperature.
I
Thermocouple junction
IIL
Solid at Ts
h, Tamb Lead Wire
Heat conducting cement
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 38 Measurement of temperature of a moving fluid
If the duct wall is different from the fluid temperature, heat transfer takes place
by conduction between the fluid and the duct wall and hence the well bottom
temperature will be at a value in between that of the fluid and the wall. There
may also be radiation heat transfer between surfaces, further introducing
errors. If the fluid flows at high speed (typical of supersonic flow of air)
viscous dissipation – conversion of kinetic to internal energy – may also be
important. With this background we generalize the thermometer error
problem in the case of measurement of temperature of a gas flow as indicated
in Figure 39.
Measuring junction
Flow
Lead wireDuct wall
Thermometer well
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 39 Heat transfer paths for a sensor in gas flow
The temperature of the sensor is determined, under the steady state, by a
balance of the different heat transfer processes that take place, as indicated
in Figure 39. Not all the heat transfer processes may be active in a particular
case. The thermometer error is simply the difference between the gas
temperature and the sensor temperature. Estimation of the error will be
made later on.
Summary of sources of error in temperature measurement:
• Sensor interferes with the process
– Conduction error in surface temperature measurement
• Sensor interferes with the process as well as other environments
– Radiation error
• While measuring temperature of moving fluids convection and
conduction processes interact and lead to error
Gas
Conduction
Radiation
Gas convection + radiation Sensor
Lead/Support
Wall Visible surfaces
Gas
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
• In case of high speed flow, viscous dissipation effects may be
important
Conduction error in thermocouple temperature measurement:
Lead wire model
Figure 40 Single wire equivalent of a thermocouple
Heat transfer through the lead wires of a thermocouple leads to error in
the measured temperature. Since a thermocouple consists of two wires of
different materials covered with insulation, and since the error estimation
should involve a simple procedure, we replace the actual thermocouple by a
single wire thermal equivalent. How this is accomplished is indicated by
referring to Figure 40. The cross section of an actual thermocouple is shown
at the left in Figure 40. It consists of two wires of different materials with the
indicated radii and thermal conductivity values. The insulation layer encloses
L1
Insulation
Wire 1 rw1,k1
Wire 2 rw2,k2
L2 r2 r1
1 21 22,
4wL Lr r r +
= =
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
the two wires as indicated. We replace the two wires and the insulation by
a single wire of radius r1 and a coaxial insulation layer of outer radius r2.
The single wire model:
1) The area thermal conductivity product must be the same for the two
wires and the single wire. Thus
( ) ( ) 1 w1 2 w2two wires one wirekA kA k A k A= = + . If the two wires have the
same diameter (this is usually the case) we may replace this
by ( ) ( ) ( )1 21 w1 2 w2 w1one wire
k kkA k A k A 2A
2+
= + = . Thus the thermal
conductivity of the single wire equivalent is equal to the mean of the
thermal conductivities of the two wires and the area of cross section of
the single wire is twice the area of cross section of either wire. Hence
the radius of the single wire equivalent is given by 1 wr 2r= as
indicated in Figure 40.
2) The insulation layer is to be replaced by a coaxial cylinder of inner
radius r1 and the outer radius r2. The outer radius is taken
as 1 22
L Lr4+
= . Note that if 1 2L L 2r= = (true for a circle of radius r),
this formula gives 2r r= , as it should.
3) Since the insulation layer is of low thermal conductivity while the wire
materials have high thermal conductivities, it is adequate to consider
heat conduction to take place along the single wire and radially across
the insulation, as indicated in Figure 41.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Figure 41 Heat flow directions
Figure 34 shows a typical application where the temperature of a
surface exposed to a moving fluid is being measured. The solid is made of a
low thermal conductivity plastic.
Figure 42 Surface temperature measurement
The thermocouple lead wires conduct away some heat that is gathered
by the thermocouple in contact with the solid. This will tend to depress the
Heat flow direction in insulation
Heat flow direction in wire
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
temperature of the junction. In order to reduce the effect of this
thermocouple lead wire conduction, the junction is attached to a heat
collecting pad of copper as indicated in the figure.
Now consider the typical application presented in Figure 42. Figure 43
explains the nomenclature employed for the analysis of this case. The heat
conducting pad receives heat from the front face of area S and loses heat
only through the thermocouple due to lead wire conduction. The appropriate
thermal parameters are as shown in Figure 43.
Figure 43 Nomenclature for lead wire conduction analysis
Heat loss through the lead wire is modeled by using fin type analysis,
familiar to us from the study of heat transfer. Since the wire is usually very
long, it may be assumed to infinitely long. The heat loss from the wire to the
ambient is modeled as that due to an overall heat transfer coefficient given by
Overall2 2
i 1
1hr r1 ln
h k r
=⎛ ⎞
+ ⎜ ⎟⎝ ⎠
(43)
hf, Tf
h, Tamb
Area, S
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The perimeter of the wire is 1P 2 r= π and the area thermal conductivity product
for the wire is ( ) ( ) ( )1 2 2 21 1 1 2
k kkA 2 r r k k
2+
= π = π + . The appropriate fin
parameter m is then given by
( ) ( )Overall Overall 1 Overall
21 2 11 2 1
h P h 2 r 2hmkA k k rk k r
π= = =
++ π (44)
Assuming the lead wire to be infinitely long, the heat loss through the lead
wire is given by
( )Lead wire t ambQ kAm T T= − (45)
Under steady conditions this must equal the heat gained from the fluid by the
pad given by
( )Gain by pad f f tQ h S T T= − (46)
Equating (3) and (4) we solve for the sensor indicated temperature as
f f ambt
f
h ST kAmTTh S kAm
+=
+ (47)
Equation 47 shows that the sensor temperature is a weighted mean of the
fluid temperature and the ambient temperature. It is clear that the smaller the
weight on the ambient side better it is from the point of view of temperature
measurement. This is a general feature, as we shall see later, in all cases
involving temperature measurement. The thermometric error is then given by
Error t fT T T= − (48)
Example 15 below demonstrates the use of the above analysis in a typical
situation.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 15
A copper constantan thermocouple of wire diameter each of 0.25 mm
is used for measuring the temperature of a surface which is
convectively heated by a fluid with a heat transfer coefficient of 67
W/m2°C. The area of the surface exposed to the fluid is 10 cm
2. The
thermocouple has an insulation of thickness 1 mm all round and the
overall size is 5 mm x 2.5 mm. The thermal conductivity of the
insulation is 1 W/moC. The thermocouple is exposed to an ambient at
a temperature of 30°C subject to a heat transfer coefficient of 5
W/m2°C. If the fluid temperature is 200°C what is the temperature
indicated by the thermocouple? Take thermal conductivity of copper as
386 W/moC and the thermal conductivity of constantan as 22.7 W/m°C.
Wire side calculation:
Diameter of each thermocouple wire d 0.00025 m=
Thermal conductivities of the thermocouple wires
1 2 tan tan386 / , 22.7 /= = ° = = °Copper Consk k W m C k k W m C
The area of cross section of each wire is
2 28 2d 0.00025A 4.909 10 m
4 4−= π = π = ×
Effective thermal conductivity area product for the thermocouple pair is
( ) ( )1 2 8 52 (386 22.7) 4.909 10 2.006 10 /2
− −+= = + × × = × °
k kkA A W m C
Overall heat transfer coefficient is now calculated:
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The overall heat transfer coefficient is calculated by combining the
insulation and film resistances. We have 2h 5 W / m C= ° . The radius of
the single wire equivalent is 10.00025 0.000177
2 2dr m= = = . The outer
radius of effective insulation layer is 20.005 0.0025 0.001875
4r m+
= = .
Thermal conductivity of insulation material is ik 1 W / m C= ° . The
overall heat transfer coefficient is
2Overakl
2 2
i 1
1 1h 4.892 W / m C1 0.001875 0.001875r r1 lnln 5 1 0.000177h k r
= = = °⎛ ⎞ ⎛ ⎞++ ⎜ ⎟⎜ ⎟ ⎝ ⎠⎝ ⎠
The overall heat transfer coefficient perimeter product is thus given by
( )Overall 1 Overallh P 2 r h 2 0.000177 4.892 0.005433 W / m C= π = × π× × = °
The fin parameter is calculated as
1Overall5
h P 4.892m 16.457 mkA 2.006 10
−−= = =
×
The surface temperature may now be calculated by equating the heat
transfer from the fluid to surface to that lost through the thermocouple
insulation. The appropriate data is:
2 2 3 2f f ambh 100 W / m C, S 10 cm 10 m , T 200 C and T 30 C−= ° = = = ° = °
From the material presented earlier, assuming the thermocouple wires
to be very long, the surface temperature is given by
f f ambt
f5
5
h ST kAm TTh S kAm
100 0.001 200 2.006 10 16.457 30 199.2 C100 0.001 2.006 10 16.457
−
−
+=
+
× × + × × ×= = °
× + × ×
The thermometer error is t fT T 199.2 200 0.8 C− = − = − °
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Temperature error due to radiation:
Errors in temperature measurement may occur due to surface
radiation, especially at elevated temperatures. We consider the same
example that was considered above. Assume that the copper disk has a
surface emissivity of . Let it also view a cold background at Tbkg. The heat
loss is now due to lead wire conduction along with radiation to the ambient.
Heat loss due to radiation is given by
( )4 4Radiation t bkgQ S T T= εσ −
Note that the temperatures are to be expressed in Kelvin in Equation 49 and
is the Stefan Boltzmann constant. The temperature of the sensor is
determined by equating heat gain by convection to heat loss by conduction
and radiation. Thus
Lead wire Radiation Gain by padQ Q Q+ = (50)
Using Equations 45, 46 and 49 we then have
( ) ( ) ( )4 4t amb t bkg f f tkAm T T S T T h S T T− + εσ − = − (51)
The above non-linear algebraic equation needs to be solved to arrive at the
value of the measured temperature.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 16
Reconsider Example 1 with the following additional data:
The copper pad has a surface emissivity of 0.05 and views a cooler
background at a temperature of 450 K. What is the thermometric error
in this case?
In addition to the heat loss by lead wire conduction we have to include
that due to radiation. This is given by
8 4 4 12 4 4Radiation t tQ 0.05 5.67 10 (T 450 ) 2.84 10 (T 450 )− −= × × − = × −
Using the material already available in Example 1 the equation that
governs the sensor temperature is given by
( )12 4 4t t t0.00033(T 30) 2.84 10 (T 450 ) 0.067 473 T−− + × − = −
This equation may be solved by Newton Raphson method. Alternately the
solution may be obtained by making a plot of the difference between the
left hand side and right hand side of this equation and locate the point
where it crosses the temperature axis. Such a plot is shown below.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
-0.15
-0.1
-0.05
0
0.05
0.1
466 468 470 472 474
Tt
LHS
-RH
S
It is clear that the sensor temperature is now 470.2 K or 197.2°C. The
temperature error has changed to -2.8°C! Error due to radiation is, in fact,
more than that due to lead wire conduction.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Measurement of temperature within a solid:
Now we shall look at the situation depicted in Figure 37. Temperature
error is essentially due to conduction along the lead wires. However, the
portion embedded within the solid (II) has a different environment as
compared to the part that is outside (I). Both of these may be treated by the
single wire model introduced earlier. Assume that the solid is a temperature
higher than the ambient. The thermocouple junction will then be at an
intermediate temperature between that of the solid and the ambient. Heat
transfer to the embedded thermocouple is basically by conduction while the
heat transfer away from the part outside the solid is by conduction and
convection. The embedded part is of finite length L while the portion outside
may be treated as having an infinite length.
Represent the temperature of the single wire equivalent as Ti in a plane
coinciding with the surface of the solid. Let Tt be the temperature of the
junction while Ts is the temperature of the solid. Let the ambient temperature
be Tamb. The fin parameter for the embedded part may be calculated based
on the overall heat transfer coefficient given by
Overall,II3 3 2 2
c 2 i 1
1hr r r rln lnk r k r
=⎛ ⎞ ⎛ ⎞
+⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠
(52)
In the above, kc is the thermal conductivity of the heat conducting cement, r3 is
the radius of the hole and the other symbols have the earlier meanings. Note
that expression 10 is based on two conductive resistances in series. The
corresponding fin parameter is Overall,IIII
h Pm
kA= . The overall heat transfer
coefficient for the exposed part of the thermocouple is given by the expression
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
given earlier, viz. Overakl,I2 2
i 1
1hr r1 ln
h k r
=⎛ ⎞
+ ⎜ ⎟⎝ ⎠
. The corresponding fin parameter
value is Overall,II
h Pm
kA= .
Figure 44 Nomenclature for thermal analysis
Referring now to Figure 44 we see that the heat transfer across the surface
through the thermocouple should be the same i.e. II IQ Q= . Using familiar fin
analysis, we have
( ) ( )IIII Overall,II s i
II
tanh m LQ h P T T
m= − (53)
For the exposed part, we have
( )I I i ambQ kAm T T= − (54)
Equating the above two expressions we solve for the unknown temperature Ti.
Thus
( )
1 s 2 ambi
1 2
II1 Overall,II 2 I
II
w T w TT wherew w
tanh m Lw h P and w kAm
m
+=
+
= = (55)
I
Ts
Tamb
QII
Ti
Tt
II
QI Surface
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Having found the unknown temperature Ti, we make use of fin analysis for the
embedded part to get the temperature Tt. Using familiar fin analysis, we have
i st s
II
(T T )T Tcosh(m L)
−= + (56)
Note that the fin analysis assumes negligible heat transfer near the bottom of
the hole!
Following points may be made in summary:
1) The longer the depth of embedding smaller the thermometric error
2) Higher the thermal conductivity of the epoxy filling the gap between the
thermocouple and the hole the smaller the thermometric error
3) The smaller the diameter of the thermocouple wires smaller is the
thermometric error
4) Smaller the thermal conductivity of the thermocouple wires smaller the
thermometric error
5) If it is possible the insulation over the thermocouple wires should be as
thin as possible in the embedded portion and as thick as possible in the
portion that is outside the hole
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 17
Thermocouple described in Example 1 is used to measure the
temperature of a solid by embedding it in a 6 mm diameter hole that is
15 mm deep. The space between the thermocouple and the hole is
filled with a heat conducting epoxy that has a thermal conductivity of 10
W/m°C. The lead wires coming out of the hole are exposed to an
ambient at 30°C with a heat transfer coefficient of 5 W/m2°C. If the
temperature of the solid is 80°C, estimate the temperature indicated by
the thermocouple.
From the results in Example 1, the following are available:
5kA 2.006 10 W m / C−= × ° , 1Im 16.457 m−=
The weight w2 is then given by
52 Iw kAm 2.006 10 16.457 0.00033 W m / C−= = × × = °
For the embedded part, the following calculations are made:
30.006r 0.003 m
2= = , ck 10 W / m C= ° ,
2Overall,II
1h 218.88 W / m C0.003 0.003 0.001875 0.001875ln ln
10 0.001875 1 0.00177
= = °⎛ ⎞ ⎛ ⎞+⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠
The fin parameter is then calculated as
1II 5
218.8 2 0.00177m 110.08 m2.006 10
−−
× × π×= =
×
With L = 0.015 m, we have IIm L 110.08 0.015 1.651= × =
The weight w1 is then given by
( )1
tanh 1.651w 218.8 2 0.00177 0.002052
110.08= × × π× × =
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The unknown temperature Ti is now calculated as
0.002052 80 0.00033 30 73.070.002052 0.00033iT C× + ×
= = °+
The sensor temperature is then calculated as
(73.07 80)80 77.4cosh(1.651)tT C−
= + = °
The thermometric error is thus equal to -2.6°C.
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The thermometer well problem
This is a fairly common situation as has been mentioned earlier. The
well acts as a protection for the temperature sensor but leads to error due to
axial conduction along the well. It is easily recognized that the well may be
treated as a fin and the analysis made earlier will be adequate to estimate the
thermometric error.
Figure 45 Nomenclature for the thermometer well problem
Assumptions:
1) Since the thermometer well has a much larger cross section area than the
thermocouple wires conduction along the wire is ignored.
2) The thermometer well is heated by the gas while it cools by radiation to the
walls of the duct (based on g t wT T T> > ).
Velocity U Temperature Tg
Temperature Tw
Measuring junction
Flow
Lead wireDuct wall
Thermometer well
Temperature Tt
ID = di, OD = do, kw L
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
3) Well is treated as a cylinder in cross flow for determining the convection
heat transfer coefficient between the gas and the well surface.
The heat transfer coefficient is calculated based on the Zhukaskas correlation
given by
m nNu C Re Pr= (57)
In this relation Re, the Reynolds number is based on the outside
diameter of the well and all the properties are evaluated at a suitable mean
temperature. The constants C, m and n are given in Table 6.
Radiation heat transfer may be based on a linearised model if the gas
and wall temperatures are close to each other. In that case the well
temperature variation along its length is also not too big. Thus we
approximate the radiant flux ( )4 4R wq T T= εσ − by the relation
( ) ( )3R w w R wq 4 T T T h T T≈ εσ − = − where 3
R wh 4 T= εσ is referred to as the
radiation heat transfer coefficient.
Table 6 Constants in the Zhukaskas correlation
Re C m 1-40 0.75 0.4 40-103 0.51 0.5 103-2×105 0.26 0.6 2×105-106 0.076 0.7 m Pr < 10 0.36 Pr > 10 0.37
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Analysis:
Figure 46 Thermometer well analysis schematic
Refer to Figure 46 and the inset that shows an expanded view of an elemental
length of the well. Various fluxes crossing the boundaries of the element are:
1) ( )Con gQ hP x T T= ∆ − 2) ( )R R wQ h P x T T= ∆ − 3) , =Cond in wx
dTQ k Adx
and
4) Cond,in wx x
dTQ k Adx −∆
=
In the above the perimeter P is given by oP dπ= and area of cross section A is
given by ( )2 2
o id dA
4
−= π . Energy balance requires
that Cond,in con Cond,out RQ Q Q Q+ = + . Substituting the expressions for the fluxes
and using Taylor expansion of the derivative, we have
( )2
w g R w w w 2x x x
dT dT d Tk A hP x T T h P x(T T ) k A k A xdx dx dx
+ ∆ − = ∆ − + − ∆
This equation may be rearranged as
Tt
Tw
U,Tg
QCon
QCond,out
QCond,in
QR
x
∆x
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
2R R
g w2w w w
(h h )P h Pd T hPT T T 0k A k A k Adx+
− + + = (58)
Let g R wref
R
hT h TT
h h+
=+
be a reference temperature. Then Equation 58 is
rewritten as
22eff2
d m 0dx
θ− θ = (59)
where refT Tθ = − and ( )Reff
w
h h Pm
k A+
= is the effective fin parameter.
Equation 59 is the familiar fin equation whose solution is well known.
Assuming insulated boundary condition at the sensor location, the indicated
sensor temperature is given by
( ) ( )( )
w reft t ref
eff
T TT T
cosh m L−
θ = − = (60)
The thermometric error is thus given by
( ) ( ) ( )( )
w reft g ref g
eff
T TT T T T
cosh m L−
− = − + (61)
Following points may be made in summary:
1) The longer the depth of immersion L smaller the thermometric error
2) Lower the thermal conductivity of the well material the smaller the
thermometric error
3) Smaller the emissivity of the well and hence the hR smaller the
thermometric error
4) Larger the fluid velocity and hence the h smaller the thermometric error
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
Example 18
Air at a temperature of 373 K is flowing in a tube of diameter 10 cm at
an average velocity of 0.5 m/s. The tube walls are at a temperature of
353 K. A thermometer well of outer diameter 4 mm and wall thickness
1 mm made of iron is immersed to a depth of 5 cm, perpendicular to
the axis. The iron tube is dirty because of usage and has a surface
emissivity of 0.85. What will be the temperature indicated by a
thermocouple that is attached to the bottom of the thermometer well?
What is the consequence of ignoring radiation?
Step wise calculations are shown below:
Step 1.Well outside convective heat transfer coefficient:
Given data:
0 f wd 0.004 m, U 0.5 m / s, T 373 K, T 353 K= = = =
The fluid properties are taken at the mean temperature given
by m373 353T 363 K
2+
= =
From table of properties for air the desired properties are:
6 223.02 10 m / s, k 0.0313 W / m K, Pr 0.7−ν = × = =
The Reynolds number based on outside diameter of thermometer well
is
06
Ud 0.5 0.004Re 86.923.02 10−
×= = =
ν ×
Zhukaskas correlation is used now. For the above Reynolds number
the appropriate constants in the Zhukaskas correlation are
C 0.51, m 0.5 and n 0.37= = = .
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
The convection Nusselt number is
m n 0.5 0.37Nu C Re Pr 0.51 86.9 0.7 4.17= = × × =
Hence the convective heat transfer coefficient is
2
0
Nu k 4.17 0.0313h 32.6 W / m Kd 0.004
×= = =
Step 2 Radiation heat transfer coefficient:
Linear radiation is used since the fluid and wall temperatures are close
to each other.
The pertinent data is: 8 2 45.67 10 W / m K , 0.85−σ = × ε =
The radiation heat transfer coefficient is
3 8 3 2r wh 4 T 4 0.85 5.67 10 353 8.5 W / m K−= εσ = × × × × =
Step 3 Calculation of reference temperature
f r wref
r
hT h T 32.6 373 8.5 353T 368.9 Kh h 32.6 8.5
+ × + ×= = =
+ +
Step 4 Well treated as a fin:
Well material has a thermal conductivity of wk 45 W / m K=
Internal diameter of well is equal to outside diameter minus twice the
wall thickness and is given by m002.0001.02004.0t2dd 0i =×−=−=
The fin parameter
( )( )
( )10 r
f 2 2 2 20 i
w
0.004 32.6 8.5d (h h )m 34.87 md d 0.004 0.002
k 454 4
−× +π += = =
− −π ×
Since the well length is L = 0.05 m the non-dimensional fin parameter
is
Mechanical Measurements Prof. S.P.Venkatesan
Indian Institute of Technology Madras
f fm L 34.87 0.05 1.74µ = = × =
Step 5 Non-dimensional well bottom temperature
It is given by tf
1 1 0.339cosh( ) cosh(1.74)
θ = = =µ
Hence the temperature indicated by the sensor attached to the well is
( ) ( )t ref t w refT T T T 368.9 0.339 353 368.9 363.5 K= + θ − = + × − =
The thermometric error is some 9.5°C.
If radiation is ignored the above calculations should be done by taking
hr= 0 and Tref = Tw. This is left as an exercise to the student.