Temperature Measurement

Chapter 16
16.1 Thermocouple Principles
16.2 Thermocouple Types
16.3 Thermocouple Installation
16.4 Resistance Temperature Devices
16.5 RTD Construction
16.6 RTD Installation
16.7 Thermowells
16.8 Comment

Temperature, obviously, is a common and impor- tion and interfacing requirements are described
tant measurement in the process industries. Vir- in sufficient detail to enable meaningful specifi-
tually all measurements from −250 to +650◦ C can cation of requirements. Finally, temperature probe
be made using either thermocouples or resistance assemblies and their installation are discussed. A
temperature devices (RTD). Thermocouples are good overview of the operation of thermocouples
cheaper than RTDs but not as accurate. Thermis- is given in BS1041 shown in Table 16.1.
tors, which are semiconductor devices, may also
be used within this range. Although they are much
more sensitive than either thermocouples or RTDs,
16.1 Thermocouple Principles
their response is very non-linear: for this reason When wires of two dissimilar metals are joined to-
they are seldom used. Above 650◦ C thermocou- gether to form a circuit, as shown in Figure 16.1,
ples, subject to materials of construction consider- and one of the junctions is heated to a higher tem-
ations, and radiation pyrometers are used. perature than the other, an emf is generated and
This chapter considers thermocouples and a current flows. This is known as the Seebeck ef-
RTDs only.Their principles of operation,construc- fect. The emf generated is small (mV) and has to
be measured on open circuit: any current drawn
through the thermocouple will cause loading ef-
Table 16.1 Standards on temperature measurement fects and distort the measurement. For this reason
BS 1041 Temperature measurement Published
a very high input impedance transducer must be
used, as shown in Figure 16.2.
Part 3 Guide to selection and 1989
use of industrial resistance Metal A
thermometers
Part 4 Guide to the selection and 1992 Hot Cold
use of thermocouples
Part 5 Guide to the use and selec- 1989
Metal B
tion of radiation pyrome-
ters Fig. 16.1 Depiction of the Seebeck effect
 100 16 Temperature Measurement

A A tolerances quoted,either as a percentage of the true

Hot Reference value or on an absolute basis, give an indication of
the potential accuracy of thermocouple measure-
ments. In practice, the accuracy achieved also de-
B pends upon errors due to faulty installation and
Fig. 16.2 Open circuit measurement of emf signal processing.

For a given pair of dissimilar metals, the emf gen-

erated is dependant on the difference in temper- 16.3 Thermocouple Installation
ature between the two junctions. For temperature A third dissimilar metal may be introduced at a
measurement it is usual to use the hot junction junction without affecting the emf generated.Thus,
as the sensor and to measure relative to the cold for example, the two dissimilar metals at the hot
junction. Thus the temperature of the cold junc- junction may be brazed or soldered together. In-
tion, referred to as the reference junction, must be troducing copper at the reference junction enables
known.In the laboratory a reference of 0 ◦ C is com- a more practical arrangement for measuring the
monly established with melting ice.However,this is emf, as depicted in Figure 16.3, the copper leads
not normally practicable for plant purposes, so the being the connections to the transducer. The ref-
reference is invariably taken as ambient tempera-
erence junction is effectively split into two parts
ture which is measured independently by means of which must be kept close together to ensure that
an RTD. they are at the same temperature.
Not only is emf a function of temperature dif-
ference – it also depends on the metals concerned.
In principle,a thermocouple can be made from any Reference
pair of dissimilar metals.However,in practice,only A
Cu
a limited number of combinations of metals and
Hot
alloys are used because of their high sensitivity, in
terms of mV/ ◦ C, and chemical stability. The stan- Cu
dard IEC 60584 provides values of emf tabulated B
against temperature for all commonly used types Brazed junction Reference
of thermocouple, together with their tolerances on Fig. 16.3 Use of third metal for split reference junction
accuracy, as shown in Table 16.2.

Table 16.2 Thermocouple reference data It is good practice for the reference junction to be
IEC 60584 Published
situated away from any potential source of heat. In
a typical installation involving multiple thermo-
Part 1 Thermocouples: 1995 couples it is normal for all the reference junctions
reference tables to be located in a common area, such as an input
Part 2 Thermocouples: 1982 signal termination cabinet, and to have a single
tolerances shared RTD measurement of cabinet temperature
for reference purposes.
Because thermocouple wire is of a very light
gauge, it is common practice to use extension leads
to extend back to the cabinet. These are of a gauge
16.2 Thermocouple Types heavy enough for wiring purposes. Ideally, exten-
They are normally referred to by letter type, the sion leads are of the same materials as the ther-
more common ones being listed in Table 16.3. The mocouple itself. However, on the grounds of cost,
 16.3 Thermocouple Installation 101

Table 16.3 Characteristics of different types of thermocouple

Type Metals Range and accuracy Comments
Continuous usage Short term usage

E Chromel (10%Cr, 0 to 1100◦ C −270 to 1300◦C Most sensitive type.Resistant to ox-

90%Ni)/Constantan idation and to moisture corrosion.
(55%Cu, 45%Ni) Must use compensating cable
J Iron/Constantan 20 to 700◦ C −180 to 750◦ C Can be used in vacuum, reduc-
ing and inert atmospheres. Oxi-
1% or 3◦ C dises above 540◦ C. Unsuitable for
moist or sulphurous conditions.
Must use compensating cable
K Chromel/Alumel 0 to 1100◦ C −180 to 1350◦C Resistant to oxidation, especially
(5%Al, 95%Ni) above 500◦ C. Unsuitable for reduc-
0.75% or 3◦ C ing or sulphurous conditions.
Must use compensating cable
R Platinum/ 0 to 1600◦ C −50 to 1750◦ C Suitable for oxidising and inert
Platinum- conditions.Rapidly poisoned by re-
Rhodium 0.15% or 1◦ C ducing atmosphere. Contaminated
(87%Pt, 13%Rh) by metal vapours so non-metallic
sheaths required, e.g. alumina.
Must use compensating cable
T Copper/Constantan −185 to 300◦C −250 to 400◦ C Can be used in vacuum, oxidis-
ing, reducing, moist and inert at-
1% or 1◦ C mospheres. Not resistant to acid
fumes. Only type to have defined
limits of error below 0◦ C.
Can use copper extension leads

it is usually necessary to use extension leads of ap- junctions are at the same temperature, variations
propriate proprietary “compensating cable”, as de- along the length of the extension leads and/or dif-
picted in Figure 16.4.Such cable has thermoelectric ferences at the other split junction do not matter.
properties similar to thermocouple wire so the emf To provide both chemical and physical protec-
measured is not distorted. Also, the use of com- tion for thermocouples, it is standard practice to
pensating cable minimises corrosion effects due use them in sheaths as depicted in Figure 16.5. The
to the formation of electrochemical cells between sheath is normally of stainless steel and up to 6 mm
the dissimilar metals. Provided the split reference in diameter. The mineral packing is typically of

Reference
A
Cu
Hot Comp. cable

Cu
B
Fig. 16.4 Use of compensating cable for remote measurement
 102 16 Temperature Measurement

Similarly, thermocouples may be used to measure

temperature difference directly by wiring them up
“back to back” as depicted in Figure 16.7.

Ref.

+
T1 –
–
T2 +

Fig. 16.7 Thermocouples wired in parallel for difference mea-

Fig. 16.5 Standard thermocouple sheath surement

magnesia (MgO) and provides good heat transfer

and electrical insulation. Practices differ, the hot 16.4 Resistance Temperature
junction may be bonded to the inside end of the
sheath or else insulated. Clearly a bonded junction Devices
gives a faster response, but the bond is difficult Electrical resistance is a function of temperature.
to make and can become detached. A plastic or, Thus, by measuring the change in resistance of
for high temperatures, a glazed ceramic terminal a given resistor, temperature may be determined.
block is used at the external end of the sheath. The The only RTD of any consequence in process au-
polarity should be clearly marked for wiring pur- tomation is the platinum resistance thermometer
poses. as described in BS1904:
Thermocouples can be used for measuring av-
erage temperatures. Suppose an average of three Above 0◦ C
measurements is required. Each may be measured R = R0 1 + aT − bT2 (16.1)
independently and the results averaged. Alterna-
tively, the three thermocouples may be wired in se- Below 0◦ C
ries, as depicted in Figure 16.6. The resultant emf, R = R0 1 + aT − bT2 − c(T − 100)T3 (16.2)
scaled by a factor of 3 prior to signal processing,
yields a temperature averaged at source which is where R is the resistance §
more accurate. T temperature ◦ C
and subscript 0 denotes the zero condition. The
Ref. values of the coefficients are as follows:
◦ −1
T1
+
Cu a = 3.9083 × 10−3 C
–
−7 ◦ −2
+
b = 5.775 × 10 C
T2 ◦ −4
– c = 4.183 × 10−12 C
+ Comp. cable
T3 – It is standard practice for R0 to be 100§ giving a
value for R of 138.5 § at 100◦ C. This is referred to
as the Pt100 sensor. To satisfy BS1904, the output of
Fig. 16.6 Thermocouples wired in series for average measure- a Pt100 sensor must fall within a defined range of
ment accuracy about Equations 16.1 and 16.2. There are
 16.5 RTD Construction 103

two classes of tolerance, A and B, of which Class A are insulated with silica tubing, to prevent short
is the more stringent: circuiting, terminating in a glazed ceramic block
at the external end of the sheath.
For Class A
dT = 0.15 + 0.002. |T|

For Class B 16.6 RTD Installation

dT = 0.15 + 0.005. |T|
RTD resistance is almost universally measured by
It is normal practice to use the Pt100 sensor for means of an integrated circuit equivalent to a
measurements up to 600◦C. From 600 to 850◦ C it Wheatstone bridge, the resistors being specified
is usual to use the Pt10 sensor. This is of thicker such that the bridge balances at 0◦ C. The two wire
gauge, for more reliable service, with an Ro value system of Figure 16.9 is adequate if the length of
of 10 §. the leads from the RTD to the bridge is short. How-
ever, if the leads are long, as is invariably the case
with process plant, their resistance would distort
the measurement since both leads are on the same
16.5 RTD Construction side of the bridge,in series with the RTD.It is there-
There are two types of platinum RTD: the film type fore common practice to use a three wire system,
and the wire wound. The film type is formed by the as depicted in Figure 16.10.
deposition of platinum on a ceramic substrate. Be-
cause of their flat structure film type RTDs are suit-
able for surface temperature measurements. How-
ever, they are not as reliable as wire wound RTDs
V1 V0
and are unsuitable for use in a conventional sheath.
The wire wound RTD, as depicted in Figure 16.8, is
used almost universally for accurate temperature RTD
measurement.

Fig. 16.9 Bridge circuit for 2-wire RTD arrangement

R0 R0
L
V1 V0
RTD
L

R0

Fig. 16.10 Bridge circuit for 3-wire RTD arrangement

Fig. 16.8 Wire wound type of RTD
In a three wire system the leads to the RTD are
It consists of the platinum resistance wire wound on adjacent sides of the bridge so their resistances
around a glass or, for high temperatures, ceramic L, and any changes in such, effectively cancel out.
former and sealed with a coating of glass or ce- Note that since the output Vo is measured by a high
ramic.As with thermocouples,it is normal practice impedance transducer,which draws negligible cur-
to use RTDs in mineral packed sheaths of stainless rent, the resistance L due to the third lead across
steel up to 6 mm in diameter. The platinum leads the bridge is of no consequence.Any imbalance be-
 104 16 Temperature Measurement

tween the resistances of the leads will nevertheless

lead to some inaccuracy. RTD1
For high accuracy measurements a four wire V1 V0
system is required utilising either null balance or
constant current source techniques: RTD2

• Null balance involves switching between two

field wire configurations and null balancing the Fig. 16.12 Bridge circuit for RTDs used for difference measure-
bridge. One configuration involves wires A, B ment
and D, the other involves wires A, C and D as de- In using a bridge circuit to measure the resistance
picted in Figure 16.11.The resulting null balance of an RTD, it is very important that the other re-
equations are used to derive the RTD resistance sistances are such that the voltage drop across the
value. There is still scope for some minor in- RTD is very low. Otherwise the current through the
accuracy due to differences introduced by the RTD will cause self heating of the RTD which will
switching circuit and its contacts. significantly distort the signal.

R0 R0
A
V1 V0 16.7 Thermowells
B
C RTD A temperature probe assembly consists of a ther-
R0
D mowell, insert and head cap as illustrated in Fig-
ure 16.13.
R0 R0
Often referred to as a thermopocket, the ther-
mowell is effectively part of the plant in which it is
A
V1 V0 installed. It is in direct contact with the process
B medium and has to be capable of withstanding
C RTD
R0 whatever process conditions exist in the plant. The
D thermowell has several functions. It enables ther-
Fig. 16.11 Bridge circuit for 4-wire RTD arrangement mal contact between the process and the sheath
containing the thermocouple or RTD. The ther-
• Constant current source involves passing a small mowell protects the sheath from the process. It en-
constant current through two connecting wires ables the sheath to be withdrawn for maintenance
and the RTD.Another two wires are then used to without having to shut the plant down. In addition,
measure the voltage across, and hence the resis- during maintenance, it prevents contamination of
tance of, the RTD using a high impedance trans- the process by entry of air or dirt.
ducer. The voltage measured is compensated for The thermowell may be either screw fitted,
offset due to thermocouple junction effects at as in Figure 16.13, or flanged. The dimensions of
the RTD. The amount of offset is established by a themowell and its fittings generally conform
switching off the constant current source for a to BS2765. Of particular importance is the inter-
short period and measuring the offset voltage nal diameter, normally 7 mm, which is consistent
directly. There is still scope for some minor in- with the 6 mm maximum outside diameter of
accuracy due to the self heating effects of the thermocouple and RTD sheaths. The length of a
constant current within the RTD. thermowell is determined by the application. The
RTDs may be used to measure temperature differ- choice materials of construction is important, a
ence directly by wiring them into adjacent sides of comprehensive listing of appropriate materials is
a bridge, as depicted in Figure 16.12. given by Pitt (1990).
 16.8 Comment 105

The insert consists of the sheath, containing the

thermocouple or RTD, and the terminal block as-
sembly, as shown in Figure 16.14. The head cap is
typically a steel casting with a screw cap that pro-
tects the terminals from dust and moisture and
provides access for maintenance.
Although a thermocouple or an RTD has an in-
herently fast response, by the time it is packed into
a sheath and inserted into a thermowell, there is an
appreciable thermal capacity. Also, the air film be-
tween the sheath and thermowell is a significant re-
sistance to heat transfer. The net effect is a sluggish
response. If the sheath and thermowell are lumped
together as a first order system, the time constant
is typically 0.5–1.0 min but, in extreme cases, can
be up to 5 min. Better thermal contact may be
achieved by spring loading the sheath, such that its
tip is in contact with the bottom of the thermowell.
Heat transfer may also be enhanced by pouring a
small amount of proprietary heat transfer oil into
the annulus between the sheath and thermowell.

16.8 Comment
For effective temperature measurement it is essen-
tial that the probe assembly is properly located. It
does not matter how accurate the sensor is if, for
example, the probe does not reach down into the
liquid whose temperature is being measured. It is
much cheaper, and less embarrassing, to address
these issues at the design and specification stage
than when the plant is operational.
It is increasingly common practice to find head
mounted transmitters in use with both thermo-
couples and RTDs. In essence, an intrinsically safe
integrated circuit is mounted in the head cap which
enables the temperature to be transmitted directly
as a 1–5-V or 4–20-mA signal, or otherwise. For
thermocouples the circuit provides amplification,
filtering, linearisation and scaling, but the temper-
Fig. 16.13 Thermowell, insert and head cap assembly ature measurement is only relative to the local am-
bient temperature conditions. Likewise for RTDs
for which the circuit provides the bridge type of
measurement.
 106 16 Temperature Measurement

Fig. 16.14 Sheath and terminal block assembly