SECTION 4

Instrumentation
Instrumentation in a gas processing plant is usually com- The basic purposes of this section are to provide:
prised of a system of pneumatic and electronic devices for mea- • A ready reference of definitions and symbols associated
surement and control of all the process variables (pressure, with gas plant instrumentation.
flow, temperature, etc.) which are pertinent to the operation • Guidelines and design information for good process
of the plant. In some cases a computer is included in the in- measurement, signal transmission, signal indication,
strumentation system to handle functions such as data gath- and control response.
ering and transmission, bulk data storage, display, alarms,
logging, computations, and control. Since the advent of inte- • A reference of terminology which pertains to the instru-
mentation of gas plants and related facilities.
grated circuit electronics, specifically the microprocessor,
many types of instruments are becoming more intelligent or
"computerized."
FIG. 4-1
Nomenclature

Controller Symbol Description Output Expression

P Proportional CO = Kp (PV – SP) + MO
I Integral (Reset) CO = Ki ∫ (PV − SP) dt
D Derivative (Rate) CO = Kd[d(PV – SP)/dt]

Cv = valve flow coefficient N1, N2...Nx = numerical constants for units of measurement
CO = controller output used
d = valve inlet diameter Pc = absolute thermodynamic critical pressure
D = internal diameter of the pipe Pu = ultimate time period
Fd = valve style modifier Pv = vapor pressure of liquid at valve inlet tempera-
FF = liquid critical pressure ratio factor, ture, kPa (abs)
dimensionless P1 = upstream absolute static pressure, measured
Fk = ratio of specific heats factor, dimensionless two nominal pipe diameters upstream of valve-
FL = liquid pressure recovery factor of a valve without fitting assembly, kPa (abs)
attached fittings, dimensionless P2 = downstream absolute static pressure, measured
Fp = piping geometry factor, dimensionless six nominal pipe diameters downstream of valve-
FR = Reynolds number factor, dimensionless fitting assembly, kPa (abs)
Gf = liquid relative density at upstream conditions, ∆P = pressure differential, P1 − P2
ratio of density of liquid at flowing temperature PB = proportional band setting
to density of water at 15.6 °C, dimensionless PBu = ultimate proportional band setting
Gg = gas relative density (ratio of density of flowing PV = process variable measurement
gas to density of air with both at standard condi- (PV − SP) = error signal
tions, which is equal to the ratio of the molecular q = volumetric flow rate
weight of gas to the molecular weight of air), di- Qg = gas or vapor flow rate, kg/h or m3/h
mensionless QL = liquid flow rate, m3/h
k = ratio of specific heats, dimensionless SP = setpoint
K = gain T = absolute temperature of gas at inlet, K
Kc = cavitation index, dimensionless Tc = time constant
Kd = derivative gain constant Td = derivative mode time constant
Ki = integral mode gain constant Ti = integral mode time constant
Kn = normalization constant Tt = absolute upstream temperature, K
Kp = proportional mode gain constant w = weight or mass flow rate
Ku = ultimate sensitivity X = ratio of pressure drop to absolute inlet pressure
M = molecular weight, atomic mass units (∆P/P1), dimensionless
MO = manual-mode controller output Xc = pressure drop ratio for the subject valve at critical
flow, with Fk = 1.0, dimensionless

4-1
 FIG. 4-1 (Cont'd)
Nomenclature

Y = expansion factor, ratio of flow coefficient for a gas Z = compressibility factor, dimensionless
to that for a liquid at the same Reynolds number, γ1 = specific weight, upstream conditions
dimensionless ξ = damping factor
A/D: Analog-to-digital. DDC: Direct Digital Control. A control technique in which
Actuator: A device which accepts the output from a control a digital computer is used as the sole controller and its out-
system and moves a final control element (usually a valve) put is used to set the final control element. This is in contrast
to change a process condition. See also "Final Control Ele- to supervisory control.
ment." Dead Band: The range through which an input may vary
Adaptive Control: Method of control whereby tuning (re- without changing the output. In a mechanical instrument
sponse) of the control system is varied with the process con- such as a meter movement or strip-chart recorder, the dead
ditions, unlike other control where tuning is manual and band is caused by friction and slack or "play" in the readout
remains constant. mechanism. In a controller, dead band is a zone around the
Algorithm: Mathematical representation of the action per- setpoint in which the measurement may vary without initi-
formed by a controller such as proportional, integral, deriva- ating a compensating controller response.
tive, or combinations of those modes. Dead Time: The interval of time lag between the initiation
Alphanumeric: A character set that contains both letters and of a controller output or stimulus and the start of the result-
digits and perhaps other characters such as punctuation ing observable process response.
marks. Dedicated Control: Using one controller to control one proc-
Analog Computer: A computing device comprised of func- ess variable.
tional modules such as amplifiers, multipliers, dividers,
Derivative Control: A mode of control using an algorithm
etc., interconnected in such a way as to facilitate the solu-
which anticipates when a process variable will reach its de-
tion of a set of mathematical expressions or to implement
sired control point by sensing its rate of change. This allows
some control strategy. The input to and the output from an
a control change to take place before the process variable
analog computer are continuous signals as contrasted with
overshoots the desired control point. See also "Control Ac-
a digital computer which updates an output every scan.
tion, Derivative (Rate)."
Attenuation: An increase or decrease in signal magnitude
between two points or between two frequencies. Derivative Time: The time difference by which the output of
Baud Rate: The number of bits or discrete pieces of informa- a proportional-derivative (PD) controller leads the controller
tion transmitted per second. input when the input changes linearly with time.
Bit: Abbreviation for "binary digit." A single character in a Digital Computer: An electronic machine for performing cal-
binary number, represented by zero (0) or one (1). culations on discrete quantities of data. Usually includes
Byte: The number of adjacent binary digits operated upon as bulk storage devices such as disks, tape units, etc., in addi-
a unit. tion to internal memory. Also includes devices for printing
Cascade Control: Controllers arranged such that the output and/or displaying output data.
of one controller manipulates the setpoint input of a second Distributed Control System: Any control system in which
controller instead of manipulating a process variable di- the degradation or failure of any single element will affect
rectly. only the control loop, or related loops, in which it operates.
Control Action, Derivative (Rate): Control action in which EPROM (Erasable Programmable Read-Only Memory):
the controller output is proportional to the rate of change of A memory device with information placed into it during
the input. manufacture that cannot be altered by the computer. It can
Control Action, Integral (Reset): Control action in which only be erased and reprogrammed with special equipment.
the controller output is proportional to the time integral of Error Signal: The signal resulting from the difference be-
the error signal. tween the setpoint reference signal and the process variable
Control Action, Proportional: Control action in which the feedback signal in a controller.
controller output has a linear relationship to the error signal. Feedback Control: A type of control whereby the controller
Controller: A device which receives a measurement of the receives a feedback signal representing the condition of the
process variable, compares that measurement with a set- controlled process variable, compares it to the setpoint, and
point representing the desired control point, and adjusts its adjusts the controller output accordingly.
output based on the selected control algorithm to minimize
Feedforward Control: A type of control which takes correc-
the error between the measurement and the setpoint. If an
tive action based on disturbances before the process variable
increase in the measured process variable above the setpoint
is upset.
causes an increase in the magnitude of the controller output,
the controller is said to be "direct acting." If a process vari- Final Control Element: That component of a control system
able increase above the setpoint causes a decrease in the (such as a valve) which directly changes the manipulated
magnitude of the controller output, the controller is "reverse variable.
acting." Gain: The ratio of change in output divided by the change in
Data Base: A collection of values for process variables, set- input that caused it. Both input and output must be in the
points, scaling factors, control parameters, limits, constants, same units; hence gain is a dimensionless number.
identifiers, etc. for access by the application programs in a Hierarchy: The ranking or precedence of the elements in a
computer-based control system. supervisory system. For example, a lower ranking element
Data Highway: A high-speed serial or parallel data path such as a local controller affects only one variable while a
which connects several units of a distributed control or data higher ranking element such as a computer might affect
collection system. many variables.

4-2
 FIG. 4-1 (Cont’d)
Nomenclature
Hysteresis: Difference between upscale and downscale out- PROM (Programmable Read-Only Memory): A device
put in instrument response when subjected to the same input with information placed into it during manufacture that can-
approached from opposite directions. not be altered by the computer. It can, however, be repro-
Integral (Reset) Control: A control algorithm which at- grammed using special equipment.
tempts to eliminate the offset (caused by proportional con- Proportional Band: The change in the controller error sig-
trol) between the measurement and setpoint of the nal required to produce a full range change in output due to
controlled process variable. See also "Control Action, Inte- proportional control action. It is the reciprocal of gain ex-
gral (Reset)." pressed as a percentage: PB(%) = 100/k.
Integral (Reset) Time: The proportionality constant in the Proportional Control: A mode of control using an algorithm
equation relating the controller output to the error for inte- which causes the output of a controller to change in a linear
gral control CO = Ki∫(PV – SP) dt. fashion as the error signal (process variable – setpoint dif-
Where: Ki = Kp/Ti. Kp is the integral gain of the controller. ference) changes. See also "Control Action, Proportional."
Ti is the time required to produce a change in controller out- RAM (Random Access Memory): Memory which contains
put equal to the change in error input. no pre-programmed information but is loaded and/or altered
Integral Windup/Windown: Also called "controller by the computer system. It is of a "volatile" nature in that
windup/windown" or "reset windup." Saturation of the con- all the contents are lost when electrical power is removed.
troller output at its maximum positive or negative value due RAM memories are usually provided with battery backup
to an error signal existing for an excessive period of time. power systems, making it "non volatile."
Can be caused by the controller being left on "automatic" Ramp: An increase or decrease of a variable at a constant rate
when the measurement transmitter is out of service. of change with respect to time.
Intrinsically Safe: Refers to equipment or wiring which is Reset Rate: The inverse of integral time; usually expressed
incapable of releasing sufficient electrical or thermal energy as "repeats per minute."
under either abnormal or normal conditions to cause ignition ROM (Read-Only Memory): Memory with information
of a specific hazardous atmospheric mixture in its most eas- placed into it during manufacture that cannot be altered.
ily ignited concentration. Serial Communications: Sending bits of information in suc-
I/P Transducer: (Current-to-pneumatic). A device which con- cession along a single circuit (pair of wires).
verts an electrical current signal to a proportional pneumatic Setpoint: The desired value at which a process variable is to
signal for the purpose of interfacing electronic and pneu- be controlled.
matic parts of a control system. A typical I/P transducer
Software: A set of programs and associated data tables which
might convert a 4-20 milliamp signal from an electronic con-
causes the hardware components of a computer system to
troller to a 20-100 kPa signal to actuate a pneumatic valve.
perform the desired tasks.
I/O Devices: Input/output devices used to enter data into and
receive data from a computer or control system. Examples Split-Ranging: Action in which two or more final control ele-
are analog and digital input and output devices for handling ments are actuated by a single controller output. For exam-
process measurements and conditions as well as "business" ple, in a heating circuit, 0-50% of the controller output
type devices such as terminals, printers, plotters, etc. operates a primary heat source and the 50-100% portion of
Noise: In process instrumentation, an unwanted component the controller output operates a secondary heat source.
of a signal or variable. Noise may be expressed in units of Steady-State: The condition when all process properties are
the output or in percent of output span. constant with time, transient responses having died out.
Offset: The steady-state deviation of the controlled variable Supervisory Control: A method of computer control
from the set-point, usually caused by a disturbance or a load whereby a computer or master station provides setpoints to
change in a system employing a proportional-only controller individual controllers which independently perform the ac-
such as a level controller. Offset will eventually be reduced tual control algorithms.
to zero by the integral action in a PI or PID controller. Telemetry: A technique which permits a measured quantity
P Controller: A controller which produces proportional con- to be transmitted and interpreted at a distance from the
trol action only. measuring location. Form, or types of telemetry include ana-
Parallel Data: Data transmission where all data bits of a log, digital, frequency, and pulse.
data word are processed at once. Transmitter: A device that converts a process measurement
PD Controller: A controller which produces proportional plus (pressure, flow, level, temperature, etc.) into an electrical or
derivative (rate) control action. pneumatic signal suitable for use by an indicating or control
PI Controller: A controller which produces proportional plus system.
integral (reset) control action. Word, Computer: A group of bits treated as a unit and capa-
PID Controller: A controller which produces proportional ble of being stored in one computer location. Some common
plus integral (reset) plus derivative (rate) control action. word lengths are 8 bits, 16 bits, and 32 bits.

4-3
 FIG. 4-2
Instrumentation Symbols

4-4
 FIG. 4-2 (Cont’d)
Instrumentation Symbols

4-5
 GENERAL INSTRUMENTATION maintain this rate for about five minutes or such time as
CONSIDERATIONS is considered adequate to perform an emergency shut-
downof the plant or to switch over to a backup air system.
Type Selection Also theair storage tankcapacity should be large enough
to prevent excessive cycling of the compressor.
Often the type selection of an instrument is pre-determined
2. Filtering and Regulation: Instrument air systems are
by whatever is available, or what will be compatible with the
normally designed for pressures up to 875 kPa (ga) and
rest of a system. There are cases, however, where the choice
should be protected by relief valves. Instrument air
to install pneumatic or electronic instrumentation must be
should be free from all contamination such as oil,
made by comparing the features of each type. Fig. 4-3 lists
water, and any hazardous or corrosive gases. Non-lu-
some of the attributes of each type to aid in this comparison.
bricated compressors should be used if possible. Where
lubricated compressors are used, an oil removal sepa-
FIG. 4-3
rator is required. The presence of oil may cause instru-
Instrument Type Features ment contamination and possibly create a combustible
mixture. After being compressed, instrument air must
Pneumatic Electronic be cooled to remove the major portion of the contained
Advantages water. A final drying system must be used to reduce
1. Intrinsically safe, no electrical 1. Greater accuracy. the water dewpoint to at least 6°C below the ambient
circuits. 2. More compatible with temperature at line pressure. An afterfilter may be re-
2. Compatible with valves. computers. quired to remove particulate carryover from the dehy-
3. Reliable during power outage 3. Fast signal transit time. drators.
for short period of time, 4. No signal integrity loss if
dependent on size of air surge current loop is used and signal 3. Proper Distribution: The air distribution system should
vessel. is segregated from A.C. current. be free of any "pockets" where liquid could accumulate.
Disadvantages If this is not possible, drain valves should be installed.
1. Subject to air system 1. Contacts subject to corrosion. All supply lines should connect to the top of the air mani-
contaminants. 2. Must be air purged, explosion fold or "header." Instrument air filter-regulators should
2. Subject to air leaks. proof, or intrinsically safe to be be provided at each air-consuming device to reduce the
used in hazardous areas.
3. Mechanical parts may fail due line pressure to the supply pressure recommended by the
to dirt, sand, water, etc. 3. Subject to electrical instrument manufacturer. This also provides one more
4. Signal boosters often needed on interference.
stage of protection from contaminants. Instrument Soci-
transmission lines of over 90 m. 4. More difficult to provide for
positive fail-safe operation. ety of America Standards ISA-S7.3 and ISA-S7.4 are ref-
5. Subject to freezing with
moisture present. 5. Requires consideration of
erences for additional information.
6. Control speed is limited to installation details to minimize
points 1, 2, 3, and 4. 4. Non-Air Systems: Natural gas has been used instead of
velocity of sound.
instrument air in some remote installations where com-
pressed air was not available. This practice should be
avoided if at all possible due to safety and pollution prob-
lems and the additional filtering and clean-up of the gas
Identification which must be done to protect the instruments. The user
An instrument may perform a single function such as a tem- must be cognizant of all applicable regulations when con-
perature indicator (TI), or a combination of functions such as sidering the use of any combustible gas in instrumenta-
a flow recording controller (FRC). Fig. 4-2 covers the common tion service. Some small-scale systems have used bottled
symbols on process and mechanical flow sheets. Table 1 in nitrogen for instrument gas. This is quite acceptable, but
Fig. 4-2 shows the accepted Instrument Society of America non-bleed type instruments should be used to keep the
(ISA) letter designations and their meanings when used in consumption to a minimum.
instrument identifications. Special identification require-
ments may be encountered in certain applications, e.g., off- 5. Hydraulic Powered Devices: Hydraulic actuators are
shore requirements of ISO 10480 (API RP 14C). For more sometimes used on valves or rams where very high
detailed information, refer to Instrument Society of America thrusts (up to 3500 kgf) are required for operation. Due
publication ISA-S5.1, "Instrumentation Symbols and Identifi- to the problems of transmitting very high pressure sig-
cation," or the respective ISO standard ISO 3511. nals, a local pump powered by an electric motor is often
used to form what is commonly known as an "electro-hy-
draulic actuator."
PNEUMATIC POWER SUPPLIES
The pneumatic power supply is more commonly known as
the instrument air system. The main considerations of an in- ELECTRONIC POWER SUPPLIES
strument air system are:
1. Adequate Capacity: The minimum capacity of the sys- Installation and interconnection wiring requirements are
tem should be the sum of the individual requirements regulated by the National Electrical Code (NEC) Article 500
of each air-consuming instrument in the system, plus (Hazardous Locations) and Article 725 (Remote Control & Sig-
a supplemental volume for purges, leaks, additions, nal Circuits). Special attention should be given to Article 725.
etc. If accurate consumption figures are not available, The requirements pertaining to physical protection of wiring,
an estimated consumption volume of 0.014m3 per min- isolation and spacing of conductors depending upon class, and
ute for each air-consuming device is usually adequate. minimum wire sizes are often overlooked in an instrumenta-
The air storage tank should have sufficient capacity to tion installation.

4-6
Power Outages and Interruptions Uninterruptible Power Supplies
It is usually the responsibility of the consumer, not the elec- Uninterruptible Power Supplies (UPS), often referred to as
tric utility company, to provide protection for connected elec- battery back-up systems, should be sized to span the third
tronic equipment against upsets such as voltage spikes caused reclosure time if power supply levels must be maintained to
by lightning, high or low voltage surges, etc. The frequency of minimize erratic plant behavior. However, if the instrument
power outages and average time for service to be restored air compressor is driven by an electric motor, it usually is not
should be determined to assist in the design of electronic beneficial to maintain battery power past the time the air sup-
power supply protection and battery backup systems. The ply is depleted. All instrument needs required for an orderly
power company should be able to provide data about their re- shutdown should be considered. Back-up power is most eco-
closure gear (equipment which attempts to restore service af- nomically provided by "floating" the batteries across the out-
ter a current surge has tripped the substation or sectionalizing put of the DC power supplies. The circuits should be designed
breakers). Also, a record of power outages in the local substa- to prevent over-charging or under-charging of the batteries as
tion area and storm frequency charts will be very useful. A well as to prevent damage to the regulator circuits when the
typical reclosure operation description is shown in Fig. 4-4. AC input is disrupted. If the AC input power must be backed
up, batteries are used to feed an inverter which transforms
FIG. 4-4 DC power into AC power of the proper voltage and frequency.
Typical Reclosure Gear Operation for Power Outages of Static switches are available to automatically switch the
Commercial Utilities power supply input from the normal AC line to the inverter
when AC line power is lost. Manufacturer recommendations
for environmental requirements must be observed to assure
Reclosure Attempt Time % Successful reliability of electronic power supplies, static switches, etc.
First Reclosure 0.1 s or less 84 Regular maintenance of battery systems is mandatory since
batteries have a limited life-span compared to other electronic
Second Reclosure 15- 45 s 10 components and terminal corrosion may cause problems.
Third Reclosure 120 s 1.5 Standby generators may be required in some installations to
Manual Intervention – 4.5 permit instrument operations to continue beyond the time
limit of the battery system. Additional information on standby
power systems is included in Section 16.
Example 4-1 — If a plant can tolerate loss of power to its elec-
tronic equipment for six seconds, and an average of 50 power
outages per year is expected, then, according to Fig. 4-4, 84% SENSING DEVICES
of those outages will be restored on the first reclosure attempt,
and the remaining 16% or approximately eight power outages Some of the more common types of sensing devices for the
per year can be expected to disrupt plant operations. Note: measurement of process variables are described as follows:
Storm frequency charts are often available from manufactur-
ers of surge arresting devices. These charts may be used in
Pressure Sensors
case power outage records are not available from the power Manometer (Fig. 4-5) — Two different pressures are ap-
company. Climatic data for a particular area may be obtained plied to two separate openings in a transparent vessel contain-
from the National Oceanic and Atmospheric Administration ing a liquid. The difference in the heights of the liquid is used
(NOAA) in Asheville, North Carolina, or from the respective as a measure of the differential pressure. This difference
national authorities for areas outside the U.S. should be corrected for temperature and gravity of the liquid
in the manometer (usually either water or mercury). Pres-
Power Supply Specifications sures are often expressed in units such as "inches of water" or
Three manufacturer specifications which should be care- "millimeters of mercury."
fully noted are: (1) regulation, (2) ripple, and (3) short-circuit Bourdon tubes (Fig. 4-6) — A Bourdon tube is a metal-
protection. Regulation is an indication of how well power sup- lic coil constructed from a metal tube having the desired elas-
ply output voltage remains constant as the electrical load is tic quality and corrosion resistance. The tendency of the tube
removed and re-connected. Good regulation implies no inter- to straighten under pressure causes a mechanical linkage to
action between connected devices on the same power supply. move a pointer or initiate pneumatic or electronic transmis-
Ripple is the amount of AC variation on the DC output with a sion of the measured pressure. Dampners should be used
constant load on the power supply. This is especially critical where pulsation is a problem. Condensate traps should be
when the outputs of transmitters are connected to analog-to- used upstream of the device in steam service. The pressure
digital (A/D) converters in a computer or microprocessor based indicated is "gauge" pressure which is relative to that of the
installation. For example, if the A/D precision allows surroundings. Bourdon gauges are also available as "compound"
resolution to the nearest 10 millivolts, power supply ripple types which indicate vacuum as well as positive pressure.
should be less than 1/3 of this or 3.3 millivolts, unless the noise
can be rejected by the converter. Short-circuit protection is a Bellows (Fig. 4-7) — A tubular device with pleated sec-
means by which the power supply current is limited at a safe tions somewhat like an accordion. It is flexible along its axis
maximum in case the output is accidentally shorted-out at and lengthens or shortens according to the applied pressure.
some point. All power supplies should include short circuit pro- The bellows is usually used in low pressure or vacuum service
tection to prevent serious damage. Power supplies should al- but types are available for use with high pressures. Typical
ways have the common side of the output separate from diameters range from 10 to 300 mm. They are often used in
chassis ground to permit the common to be grounded at a sin- force-balance type transmitters and other applications where
gle point to a "high quality" instrument ground. small displacements are required. Like the Bourdon tube, it
indicates pressures as "gauge" or relative to its surroundings.

4-7
 FIG. 4-5
Types of Manometers

FIG. 4-6
Types of Bourdon Tubes

Bourdon Tube Process

Pressure
Pointer

Motion
Pinion Sector

Link
Process
Pressure
Moving
Up
Socket

(a) "C" Tube (b) Spiral (c) Helical

Diaphragm (Fig. 4-8) — A flat or curved seal with a link classified as either transparent or reflex types. A transparent
attached to an indicator or transmission device. A diaphragm gauge glass consists of either a glass tube or an arrangement
may have its own deflection properties such as with a metallic of flat glass plates in some type of holder. Since the process
type or it may be attached to a spring or other elastic member fluid level is viewed directly, the transparent gauge glass is
such as with non-metallic diaphragms. normally used with opaque fluids. The reflex type has reflect-
ing prisms to aid in viewing transparent fluids. Caution
Electrical Pressure Transducers should be observed when handling and installing these and/or
The primary sensing element of many electrical pressure any tempered glass instrument. Scratches or chips can reduce
transducers usually takes the form of a Bourdon tube, bellows, the strength of the glass and cause safety problems.
or diaphragm to generate a movement which is transmitted Chain and tape float gauges (Fig. 4-10) — Used in
to a strain gauge. A strain gauge is a device using resistance large, unpressurized storage tanks where the entire full-to-
wire connected in a Wheatstone bridge configuration to gen- empty range must be measured.
erate an electrical signal proportional to the movement and
hence proportional to the process variable being measured. Lever and shaft float gauges (Fig. 4-11) — Used on
Other types of electrical pressure transducers use properties either unpressurized or pressurized vessels where only a
of inductance, capacitance, or magnetic coupling to convert a small range of level must be measured. The range of measure-
pressure measurement to an electrical signal. ment is determined by the length of the float arm.

Level Sensors Displacer level measuring device (Fig. 4-12) —

One of the most frequently used level measuring devices is the
Gauge glass (Fig. 4-9) — This is the most commonly torque tube displacer. It is attached to the free end of a torque
used visual process-level device. Gauge glasses are generally tube which has elastic properties that permit it to twist as the

4-8
 FIG. 4-7
Types of Bellows

Process
Pressure

Evacuated
Reference
Bellows

Pressure Pressure

(a) Unopposed (b) Spring-loaded (c) Beam Balance Sensor

FIG. 4-8
Diaphragm Pressure Elements

Force Bar
Opposing
Force
Flexure
Seal
Pressure Restraining
Spring

Corrugated
Diaphragm

High
Pressure
Housing

Slack Diaphragm High Low

Pressure

(a) Slack Diaphragm (b) Force-Balance

Differential Pressure

displacer tries to float. This slight turning of the free end of Fig. 4-14. Slight tension on the tape reel permits the follower
the torque tube is connected to an indicator or transmitter. magnet to track the float at the liquid level in the device in
Torque tube displacement gauges are normally limited to level Fig. 4-14a. The position of the reel represents the level and is
spans of three meters. either connected to an indicating device or a transmitter. The
device shown in Fig. 4-14b illustrates the use of a conductive
Head-pressure level gauges (Fig. 4-13) — The true fluid for high and low level alarm indication.
level of a liquid can be determined by dividing the measured
hydrostatic head by the density of the liquid. This method re- Capacitance probes (not illustrated) — A continu-
quires a knowledge of the densities of all phases of the liquid. ous method of level measurement based on electrical proper-
Some of these methods are: pressure gauge, bubble tube, and ties. This method uses an electrode placed inside a vessel (or
differential pressure measurement. The bubbler (Fig. 4-13a) in a protective shell inside the vessel). The capacitance be-
is used at vacuum and low pressures and is especially good for tween the electrode and the wall of the vessel or shell varies
services such as molten sulfur and dirty liquids. In as the dielectric constant varies. The dielectric in this case is
"boiling-liquid" service (Fig. 4-13b), a condensate trap must be the fluid, hence the capacitance varies in proportion to the
used on the vapor leg. The level of trapped condensate in the liquid level. This capacitance is then measured, and converted
vapor leg will usually be different than the vessel liquid level, to a level measurement to be indicated or transmitted.
requiring compensation of the transmitter.
Other methods — Some of the other level detection
Electrical type level gauges and switches (Fig. methods use principles of:
4-14) — Two common types of level gauges are the float-mag-
netic gauge configuration and the conductive type shown in 1. Absorption of nuclear radiation as in a gamma gauge.

4-9
 FIG. 4-9 FIG. 4-11
Flat Glass Gauge Glasses Lever and Shaft Float Gauge

U-Bolt

Packed Bearing
Chambers Glass
Float
Float
Glass
Glass (a) Internal Float
Covers
Packed Shaft

(b) External Float

Reflex Gauge Transparent Gauge

FIG. 4-10 FIG. 4-12

Chain and Tape Float Gauge Displacer Level Measuring Device
Tape
Sheaves Torque Tube

Displacer Rod Rotary Shaft

Guide
Wires

Float
Gauge
Head

Displacer

Gauge Board Indicator Dials Liquid Level

(a) Gauge-Board Indicator (b) Ground-Reading Gauge

2. Absorption of ultrasonic waves. cal grounding of the thermocouple at the measurement point
must be avoided.
3. Viscosity of the liquid.
In applications other than a laboratory environment, main-
4. Thermal conductivity and expansion.
taining a reference junction temperature is not practical.
Therefore, the reference junction is usually at or a part of the
Temperature Sensors measuring instrument where the reference temperature can
Thermocouples — An ordinary thermocouple consists of be measured and compensated for the measurement circuits.
two different kinds of wires (dissimilar metals) joined together Various series arrangements of thermocouples may be made
at one end to form the measuring or "hot" junction. Where the to obtain differential temperatures or temperature averages.
free ends are connected to the measuring instrument, a refer- Qualified personnel may check indicating or recording tem-
ence or "cold" junction is formed. The millivolt readings meas- perature devices measuring thermocouple potentials using
ured by the instrument represent the difference in the portable equipment compatible with the thermocouple and
temperatures of the two junctions and can be converted to with compensating circuitry identical to the primary device.
temperature by various methods using conversion data from The use of incompatible equipment could result in erroneous
thermocouple tables. The reference temperatures normally used results, especially in low temperature applications. At low
to generate thermocouple tables are 0°C and 20°C. Fig. 4-15 temperatures, extreme care must be taken to eliminate
shows some of the common thermocouple types, their usable sources of moisture in thermocouple installations. Some gen-
temperature ranges, and the materials of construction. eral properties for thermocouples are given in Fig. 4-15. Con-
version tables for converting millivolts to temperatures can be
Thermocouples used for process measurements are usually found in NBS Circular #561, or obtained from thermocouple
protected by a thermowell. The mass of the thermowell should suppliers for common types.
be kept to a minimum in the interest of faster response. The
thermocouple must be in thermal contact with the thermowell. Resistance thermometers — These are often called
This is accomplished by the use of a thermally conductive lu- RTD’s for "Resistance Temperature Devices." Since the resis-
bricant or physical contact between the thermocouple and the tance of metals changes as the temperature changes, a resis-
well. In many measurement and control applications, electri- tance thermometer can be constructed using this principle.

4-10
 FIG. 4-13 electrical output which is suitable for direct connection to in-
dicators, controllers, recorders, etc. The use of RTD’s may also
Head Pressure Level Gauges
be more economical in some installations since the extension
Pressure Measuring wires may be of copper rather than the more expensive ther-
Element
mocouple extension wire. A reference temperature source is
Air Or Gas not required for calibration. A special class of resistance ther-
Supply
mometer is the thermistor device. It is low in cost, has fast
Constant Flow response, and is very stable, but is limited to use at tempera-
Regulator tures below 320°C.
Filled-system thermometers — These are simple, re-
liable, low cost devices. A bulb is attached to a capillary tube
Bubble Tube
which is connected to a measuring element (bellows, Bourdon
Condensate Pot tube, etc.) in an indicating or transmitting device. The system
is filled with a liquid or gas which changes in volume or pres-
sure as the temperature of the bulb changes. The length of the
capillary run is normally less than 80 m.
(a) Bubbler Glass stem thermometers — These devices are nor-
mally used in the office, laboratory, or other non-process areas.
Breakage is a problem; accuracy is from 0.05 to 1.0 °C depend-
ing upon the range.
(b) Boiling Liquid
Bimetallic thermometers — The sensing element con-
Low Pressure High Pressure sists of two metals with different coefficients of expansion
Differential-Pressure bonded together and attached to an indicator. These are inex-
Measuring Element
pensive, but not very accurate and are normally used in on-off
temperature thermostats where precise control is not re-
The metals that fit a near linear resistance temperature rela- quired, or in process applications where relative changes are
tionship requirement best are platinum, copper, and nickel. to be monitored. They should be calibrated at or near the nor-
An accurate resistance measuring device utilizing a Wheat- mal operating point of the temperature being monitored.
stone bridge is calibrated in units of temperature rather than
resistance. RTD’s are used in applications where faster re- Flow Sensors
sponses and greater accuracies are required than may be ob- Variable head flow meters — Flow meters in this class
tained with thermocouples. Also RTD’s have a fairly high detect a pressure difference across a flow element specially

FIG. 4-14
Electrical Level Gauges/Switches

Tape
Reel
To To
High Low
Level Level
Tape Alarm Electrodes Alarm

Non-Magnetic Load
Guide Tube Load
Contact Contact

Relay Relay

Float Conductive Fluid

A.C. A.C. Supply

Follower Supply
Magnet Transformer Transformer

(a) Magnetic Gauge (b) Conductivity Level Switch

4-11
 FIG. 4-15
Properties of Thermocouples
Thermocouple Wires
Limits of Error of Thermocouples for Standard Wire Sizes

Temperature Limits of Error

Type
Range, °C Standard Special
Chromel/Alumel (K) 0 to + 280 ± 2°C ± 1°C
+ 280 to + 1260 ± 3⁄4% ± 3⁄8%
Iron/Constantan (J) 0 to +280 ± 2°C ± 1°C
+280 to +760 ± 3⁄4% ± 3⁄8%
Copper/Constantan (T) –180 to –60 — ± 1%
–100 to –60 ± 2% ± 1%
–60 to + 90 ± 11⁄2% ± 3⁄4%
+90 to + 370 ± 3⁄4% ± 3⁄8%

Platinum/Platinum - 10% Rhodium (S) 0 to +540 ± 3°C —

Platinum/Platinum - 13% Rhodium (R) +540 to +1480 ± 1⁄2% —
Chromel/Constantan (E) 0 to +320 ± 2°C —
+320 to +870 ± 1⁄2% —

designed to create that pressure difference. The most common Turbine meters — These use a small permanent magnet
flow element is the orifice plate, but other elements also in use mounted on the meter tube to create a magnetic field. A small
are flow nozzles, venturi tubes, pitot tubes, averaging pitot turbine is mounted inside the tube and turns with a speed
tubes, target plates, and pipe elbows. Refer to Section 3 for proportional to the flow rate. As each vane of the turbine
information relating to orifices and orifice plates. passes through the magnetic field the magnetic flux is dis-
turbed which induces a pulse in a pickup coil mounted on the
Variable area flow meters (Fig. 4-16) — This type outside of the meter. The pulse rate is proportional to the flow
includes the familiar rotameter. The differential pressure rate. Pulses are then counted and converted to standard flow
across the device is held constant, and the area through which units.
the fluid passes changes due to the movement of the float up
and down the tapered tube. These are usually limited to use Positive displacement meters — Positive displace-
with relatively small flows where visual indication is suffi- ment meters and metering pumps measure discrete quantities
cient. of the flowing fluid. The rotating element is mechanically cou-
pled to a transmitter or counter which integrates or totals the
FIG. 4-16 counts to provide an indication in units of gallons, liters, cubic
Rotameter feet, etc. Some common types are: rotating vane, bi-rotor, ro-
tating paddle, oscillating piston, and oval gear meters. They
are used for custody transfer devices such as gas meters or
gasoline pumps.

Other flowmeters — Some other flowmeter types occa-

sionally encountered are:
• Magnetic flowmeters
• Doppler effect or ultrasonic flowmeters
• Vortex shedding flowmeters
• Laser velocimeters
• Thermal meters
• Nuclear Magnetic Resonance meters
• Gas ionization meters
• Cross-correlation devices
• Mass flowmeters
All flow meters should be calibrated using the fluid being
measured, or, if a different fluid is used for calibration, the
properties of the calibrating fluid must be related to the fluid
of measurement.

4-12
 SIGNAL TRANSMITTERS FIG. 4-17
Connection Methods
Pneumatic Transmitters
A pneumatic transmitter is a device that senses some proc-
ess variable and translates the measured value into an air
pressure which is transmitted to various receiver devices for
indication, recording, alarm, and control. The signal range of
20-100 kPa is the accepted industry standard; however, other
ranges may be encountered. This signal is proportional to the
range of measurement of the process variable.
The prime function of a transmitter is to reproduce the low-
energy measurement signal with sufficient energy that it may
be transmitted over an appreciable distance or used as a power
source to a control device. The low-energy measurement signal
is that position or movement associated with the action of the
process variable on the sensing element (bellows, diaphragm,
Bourdon tube, etc.). Pneumatic transmitters operate in a man-
ner similar to proportional controllers.
Electronic Transmitters
Electronic transmitters perform the same function as pneu-
matic transmitters: a low energy process-related signal is con-
verted into a higher energy signal suitable to connect to other
instruments in the system. The output signal of most elec-
tronic transmitters is a 4-20 mA, 10-50 mA, or 1-5 Vdc signal.
Other ranges often encountered are: 0-10 Vdc, 2-10 Vdc, and
0.25-1.25 Vdc. Electronic transmitters are also classified as
force balance or motion balance types.
Connection Methods — Of great concern to the instru-
ment engineer is the method by which electronic transmitters
are connected in the instrumentation system. The "two-wire,"
"three-wire," and "four-wire" classifications are often used to
describe the method of connection.
Two-wire transmitters (Fig. 4-17a) — These are the
simplest and most economical and should be used wherever
load conditions will permit. In a two-wire system the only
source of power to the transmitter is from the signal loop. Re-
ferring to Fig. 4-17a, the 4 mA "zero-end" current is sufficient
to drive the internal circuitry of the transmitter and the cur-
rent from 4 to 20 mA represents the range of the measured
process variable. The power supply and the instruments are
usually mounted in the control room.
Three-wire transmitters (Fig. 4-17b) — Some
transmitters require more power than the signal loop (4-
20 mA, etc.) can supply to support their internal circuitry. A
DC common wire is run from the instrument to the transmit-
ter. This permits the transmitter to draw whatever power it
needs from the power supply and produce the desired signal
current at the transmitter output.
Four-wire transmitters (Fig. 4-17c) — Some trans-
mitters have their own internal power supply and require no
connection to the DC power supply. A 120 Vac source is con-
nected directly to the transmitter and its output signal loop is
connected only to the receiving instrument. These are often
used where an instrument is "added on" to an existing instru-
mentation installation to avoid adding to the load of the DC
supplies. The disadvantage is the need for AC power at the
instrument site.
Signal Converters
Signal converters are used either to achieve compatibility
between different types of instruments or for isolation pur-
poses. Some common forms of signal converters are:

4-13
 Pneumatic-to-electronic (P/I) — These are electronic scale. This movement is due to the action of the measured
pressure transmitters designed for 20-100 kPa input range quantity on a diaphragm, bellows, electromagnetic coil, or
and the desired output range (4-20 mA, etc.). other sensing device which is mechanically linked to the
pointer. This includes pressure gauges, filled tube dial ther-
Electronic-to-pneumatic (I/P) — I/P converters are mometers, voltage and current meters, level gauges, etc.
pneumatic transmitters with an electro-magnetic device
connected to a nozzle-baffle arrangement which generates a Electronic analog type — These are analog indicators
pneumatic output signal which is proportional to the input with no moving parts. A signal from the sensing device acti-
signal. vates an optical display attached to graduated scale. A
common type uses a bank of 200 tiny gas filled tubes which
Isolators — These are usually electronic current-to-cur- are illuminated additively in proportion to the magnitude of
rent or voltage-to-voltage converters which provide electrical the process signal. This results in resolution of 0.5% of full
isolation to eliminate unwanted ground loop currents or com- scale.
mon mode voltages.
Electric signal converters — These fit the same cate- Digital type — Digital indicators include an analog-to-
gory as I/Ps and P/Is in that they change the signal from one digital converter which changes the electrical process signal
range to another. Examples are 4-20 mA to 0-10 Vdc, 1-5 Vdc to binary format which is then displayed in numerical form.
to 10-50 mA, etc. Typical displays consist of light emitting diodes (LED’s), liquid
crystal displays (LCD’s), gas filled tubes, etc.
Frequency converters — Frequency to DC converters
typically receive pulse inputs from turbines or positive dis-
placement flowmeters and provide a proportional 4-20 mA,
10-50 mA or voltage output. Voltage output converters are
CONTROL CONCEPTS
often referred to as F/V (frequency-to-voltage) converters or
transmitters. V/F (voltage-to-frequency) converters are often Control Loops
used to interface standard "current-loop" type instrumenta-
tion to control devices requiring frequency or pulse-train set- A control circuit is commonly referred to as a "loop." A con-
point inputs. These are commonly used in speed indicators for trol loop may be classified as either "open" or "closed" depend-
high speed centrifugal equipment. ing upon whether the control adjustments are manual settings
(open loop) or automatically determined by some type of feed-
back controller (closed-loop).
RECORDERS AND INDICATORS
Open loop (Fig. 4-18a) — In an open-loop control sys-
Recorders tem, an operator makes a manual adjustment to a device
(valve) which controls the flow of a manipulated variable
A recorder is a device used to plot the value of one or more (steam) to attempt to achieve some set-point (desired tempera-
measured variables, generally against time, but in some cases ture) value of a controlled variable (hot water). However, this
against another associated variable or variables. Recorders adjustment is only valid for the conditions under which the
are often classified in the following ways: operator made the adjustment. Any disturbance such as a
1. According to use, i.e., whether the recorder is an integral change in inlet water temperature, steam temperature, heat
part of the measuring/controlling system or is a general loss to the surroundings, or throughput will cause the outlet
purpose type such as would be used in a laboratory or temperature to change.
with a chromatograph.
2. According to method used to drive the pen(s). This refers Closed loop (Fig. 4-18b) — If appropriate measuring
to whether the pen is directly connected to the sensing and controlling elements are added to the system, the loop is
element or to some type of pen positioning mechanism closed by the inclusion of an automatic feedback controller.
activated by the measuring signal. The controller detects any difference between the set-point
and measurement signals (error signal) and produces an out-
3. According to chart type. This primarily refers to whether put signal to drive the valve in the proper direction to adjust
the recorder is of the circular or strip chart type and the heat input to cause the measurement to reach the set-point
whether the time-axis drive is powered by a mechanical value.
spring, electrical motor, or pneumatic drive.
4. Analog or Digital. Analog recorders are the more familiar Feedback control (Fig. 4-18c) — The basic compo-
strip chart and circular types. Digital recorders include nents of a feedback control loop are shown in block diagram
such things as strip printers, data loggers, electronic to- form in the figure. The "comparator" actually represents the
talizers, and computer-related devices such as data ter- entire controller and any associated signal converters. The
minals and printers. "control element" is the valve, the "feedback element" is the
transmitter, and the "process" is the mixing of the steam and
Indicators cold water inside the water heater.
An indicator is any device which presents a visual display
of a measured quantity such as temperature, pressure, humid- Feedforward control (Fig. 4-18d) — Feed forward
ity, voltage, etc. Indicators are included in an instrumentation control (often called "Predictive Control") is actually a form of
system either as independent devices (denoted as TI, PI, FI, open-loop control. An input variable (cold water temperature)
etc.) or as a part of a controlling device (TIC, PIC, etc.). Indi- is monitored and the manipulated variable (steam flow) is ad-
cators may be classified in the following groups: justed accordingly to compensate for changes in the input vari-
able. Feedforward control is almost always used in conjunction
Mechanical type — In these indicators the measured with feedback control to overcome the effects of some expected
quantity causes the movement of a pointer along a graduated disturbance.

4-14
 FIG. 4-18 CONTROL MODES AND CONTROLLERS
Control Concepts
Basic forms of control action or "modes" used in most process
control are: two-position or "on-off" control, proportional con-
trol, integral or "reset" control, and derivative or "rate" control.
The latter three modes are often used in various combinations
Desired with each other.
Temperature = T o
Temperature Two-Position (on-off) Controllers
Indicator The simplest form of control action is "on-off" control, in
which the controller output either energizes or de-energizes
Hot Water some two-state device such as a relay or an open-shut type
valve. The two-position controller is used extensively in home
heating and cooling systems, refrigerators, hot water tanks,
Steam Water air compressors, and other applications where the cost of more
Heater precise control is not justified. Most two-position controllers
Steam are reverse-acting, i.e., when the measured variable is above
Cool Water Valve
the set-point, the controller turns the manipulated variable
OFF, and when the measured variable is below the setpoint,
(a) Manual Feedback Control (Open Loop) the controller turns the manipulated variable ON. A "dead-
band" or differential gap exists around the zero error condition
to minimize cycling. This is often implemented as a pair of
control points: one where the controller will "kick-on" and the
Set Point
(Desired Temperature T o) Temperature other where the controller will "kick-off" as opposed to a single
setpoint.
Indicating Transmitter
Temperature Temperature Proportional, Integral, and Derivative
Controller Sensor Control Modes
Hot Water
When the cycling nature of "on-off" control cannot be toler-
ated, a controller using some combination of the proportional,
integral, and derivative modes is normally used. The normally
Steam Water used combinations of these modes are: (P) Proportional only;
Heater
Control (PI) Proportional plus integral; (PD) Proportional plus deriva-
Cold Water Valve tive; and (PID) Proportional plus integral plus derivative. In-
tegral mode is rarely used alone since it adds lag to the system
(b) Automatic Feedback Control (Closed Loop) without the benefits of proportional mode. Derivative mode is
never used alone since its output is zero except when the error
signal (PV – SP) is changing. Fig. 4-20 shows the response of
Desired Disturbance each mode to illustrated error signal.
Value Inputs
Direct and Reverse Acting Controllers
Error Manipulated Controlled The error signal term (PV – SP) indicates "reverse" control-
Comparator Control Process
Signal Element Variable Variable ler action (the controller output decreases when the measure-
ment increases). Most controllers may be switched from one
Feedback type of action to the other, and the choice depends upon the
Element action of the final control element and the direction that the
measured controlled variable responds to the manipulated
variable.
(c) Basic Components Of A Feedback Control Loop
Proportional Mode (P)
The "proportional-only" mode is the simplest of the three
modes of control. It is characterized by a continuous linear
relationship between the error signal (PV – SP) and the con-
troller output.
Hot Water
The proportional controller is simple, inexpensive, and does
not introduce any additional time lags into the control system.
Its chief disadvantage is its inability to cope with load changes.
Steam Water
Heater In the expression:
Steam
Cool Water Valve Temperature CO = Kp (PV − SP) + MO Eq 4-1
Indicator
the term MO represents the controller output while on "man-
ual" (just before the controller is switched to "automatic").
(d) Concept Of Feedforward Control Thus MO is the controller output required to maintain the
process variable at the setpoint value with the present load on
the system. The terms PV and SP in the equation are always

4-15
 FIG. 4-19 Combining the expressions for proportional and integral
Responses of Proportional, Integral, and Derivative Con- modes gives the expression for the PI controller:
trol Modes to Various Process Inputs and Disturbances CO = Kp (PV − SP) + Ki ∫ (PV − SP) dt + MO Eq 4-3

Because of the relationship between the controller gain and

the process gain, the integral mode gain Ki can be expressed
as Kp/Ti, where Ti is the integral mode time constant. This
gives the more familiar form of the PI controller equation:
CO = Kp [(PV − SP) + 1/Ti ∫ (PV − SP) dt] + MO Eq 4-4

The integral term will cause the valve (or other final control
element) to move at a speed proportional to the amount of the
offset until all offset is removed. The integral time constant Ti
is also known as reset time, and since it represents the time
for the integral action to "repeat" the amount of the propor-
tional action, the reciprocal of Ti (called "reset rate") is often
given in units of "repeats per minute".
Proportional Plus Derivative Mode (PD)
Feedback control has a deficiency in that no corrective action
is taken until some error signal exists. Problems are often en-
countered in the control of processes with large time constants
and/or dead times. In controlling these slow processes, correc-
tive action needs to be taken as soon as possible or the time to
recover will be too long. Derivative action adds to the control-
ler output based on the rate of change of the error signal
(PV – SP). This is also called "rate" or "preact" control action,
and control using derivative (rate) action is known as "antici-
patory" control. Combining the proportional and derivative
terms gives the expression for PD control action:
CO = Kp (PV − SP) + Kd [d(PV − SP)/dt] + MO Eq 4-5
in the same units, but not in the same units as CO and MO. The derivative gain constant Kd may be expressed as
The gain constant Kp can include a conversion factor to make Kd = KpTd. The derivative time constant Td represents the time
the units compatible. required for the contribution from the derivative term to equal
Proportional gain is often expressed in terms of Proportional the contribution from the proportional term with a steadily
changing or "ramp" error signal. Derivative is, by its nature,
Band, where:
sensitive to signal noise and can cause excessive wear on the
Proportional Band (PB %) = 100/Gain Eq 4-2 control valve. Sufficient signal filtering should be done to re-
duce the noise when derivative action is used in the controller.
The proportional band represents the percentage of full
range change of the error signal (PV – SP) that will cause full Proportional Plus Integral Plus Derivative
range change in the output of the controller. Mode (PID)
Offset Many controllers include circuitry for proportional, integral,
and derivative modes which permits the user to select the de-
"Offset" is the deviation of the controlled variable from the sired control action. Combining terms as before gives the ex-
setpoint expressed in process units. Offset will occur whenever pression for PID control action:
process load conditions have changed or a setpoint change has
been made such that the original manual output, MO, of the CO = Kp (PV − SP) + Ki ∫ (PV − SP) dt Eq 4-6
controller is not sufficient to maintain the controlled process
variable at the setpoint, SP. Offset may be minimized by using + Kd [d(PV − SP)/dt] + MO
high controller gains (narrow proportional band); however, As with the PI controller, the integral gain may be expressed
high gain may be impractical or may cause instability in some as Kp/(Ti) and the derivative gain may be expressed as Kp/(Td).
processes. Some controllers are equipped with a manual reset This gives the more familiar form:
control which can be used to "zero-out" the offset while in the
automatic mode. Controllers not so equipped are often CO = Kp [(PV − SP) + 1/Ti ∫ (PV − SP) dt Eq 4-7
switched back to "manual," adjusted for zero offset, then
switched back to "automatic." Since this manual resetting + Td [d(PV − SP)/dt]] + MO
would need to be done after every upset, integral mode is often An alternate form used in many digital computer systems
added to give automatic reset action. uses a normalization constant which "lumps" the units conver-
sion factors in kn:
Proportional Plus Integral Mode (PI)
CO = Kn [Kp(PV − SP) − 1/Ti ∫ (PV − SP) dt Eq 4-8
The "PI controller" is the most widely used of the various
controller types (70-90% of the controllers in a typical plant). + Td [d(PV − SP)/dt]] + MO

4-16
 Fig. 4-21 compares the advantages and disadvantages of the
various control modes for most desirable control. FIG. 4-21
Typical Response Curve
FIG. 4-20
Control Mode Comparisons

Mode Advantages Disadvantages

On-Off Simple, inexpensive Constant cycling
Proportional Does not add lag Almost always has offset
Integral Eliminates offset Adds time lag to system
Derivative Speeds up response Responds to noise

Controller Tuning
Controller tuning has been referred to as the most impor-
tant, least understood, and most poorly practiced aspect of
process control. It is not an exact science and considerable
"trial and error" must be used to achieve an acceptable combi-
nation of the tuning parameters for a particular process.
"Good" control is a matter of definition and depends upon such
factors as individual preference, process disturbances and in-
teractions, product specifications, etc. Fig. 4-21a shows typical
controller responses for different damping factors (ξ) to a step-
change in load or setpoint. The damping factor may be defined
as the ratio of actual damping to critical damping. The reader
is referred to Perry’s Chemical Engineers’ Handbook, Section
22 for a more complete definition of damping factors. The
"over-damped" responses (ξ> 1}) cause no overshoot. The "un-
der-damped" responses (ξ < 1) cause considerable oscillation
about the setpoint, but the controlled variable eventually
"lines-out" on the setpoint. The "critically-damped" response
(ξ = 1) would seem ideal but often the time required to reach
the setpoint is excessive. The quarter-decay response allows
overshoot but each "swing" above and below the set-point is
one-fourth that of the previous swing. This is the response that
is usually considered acceptable for most processes. See Fig.
4-21b for an example of quarter-decay response.
Techniques for tuning controllers generally fall into one of
two classes. First, there are some methods based on parame-
ters determined from the closed-loop response of the system.
Second, some methods are based on parameters determined
from the open-loop response curve commonly known as the
process reaction curve. The popular Ziegler-Nichols method
falls into the second category.
Ziegler-Nichols Method1
This method was first reported by Ziegler and Nichols in
1942 and is also known as the "ultimate" method. The method
determines the ultimate or maximum controller gain (propor-
tional action only) for which the system is stable. The ultimate
gain and ultimate period (time for one cycle at ultimate gain)
are then used to calculate initial controller settings for quar- 4. If the response is stable (curve C in Fig. 4-22), increase
ter-decay response. The following steps may be used to deter- the gain and repeat the upset.
mine ultimate gain and period: 5. If the response is unstable (curve A in Fig. 4-22), decrease
1. Switch the controller to automatic. the gain and repeat the upset.
2. Tune out all integral (reset) and derivative (rate) action. 6. When the response exhibits sustained cycling (curve B
(set Ti = max., set Td = zero) in Fig. 4-22), note the controller gain setting and the ul-
timate period (Pu).
3. Set the proportional gain Kp at some arbitrary value, im-
pose an upset on the process, and observe the response. Fig. 4-23 gives relative controller gain, integral time, and
(Moving the setpoint briefly and returning will create an derivative time for the various control mode combinations for
upset.) quarter-decay response as related to ultimate controller gain

4-17
 FIG. 4-23
FIG. 4-22
Ziegler-Nichols Settings for 1/4 Decay Response1
Typical Responses Obtained When Determining Ultimate
Gain and Ultimate Period
Mode Kp or PB(%) Ti Td
(P) 0.5 Ku 2(PBu) max. zero
(PI) 0.45 Ku 2.2(PBu) Pu /1.2 zero
(PD) 0.6 Ku 1.65(PBu) max. Pu/8.0
(PID) 0.6 Ku 1.65(PBu) Pu /2.0 Pu/8.0

FIG. 4-24
Typical Controller Settings

Integral Derivative
Process Gain PB(%)
Ti (s) min/repeat Td (s)
Flow 0.6-0.8 167-125 3.0-1.8 0.05-0.03 0.0
Pressure 5.0 20.0 120-60 2.0-1.0 0.0
Temp. 1.0-2.0 100-50 120-30 2.0-0.5 6.0-12
Level 0.8-1.2 125-83 600-300 10.0-5.0 0.6-1.2

Special considerations should be made in applying a "split-

range" controller. A common example is a column temperature
controller on a cryogenic demethanizer. In this system the first
half (0-50%) of the controller output actuates the "free" heat
exchange with the incoming feed, and the second half (50-
100%) of the controller output actuates the supplemental heat
from the hot oil system. Adaptive gain control may be required
since the heating value of the hot oil is much greater than that
of the gas used in the heat exchange.

setting, Ku, and ultimate period Pu. Gain settings are also CONTROL VALVES
shown in units of proportional band, PB.
Selecting the proper control valve for each application in-
Fig. 4-24 shows some typical settings for various types of volves many factors. The valve body design, actuator style,
process controllers. and plug characteristic are critical items for selection. Proper
Example 4-2 — An example using the Ziegler-Nichols method valve sizing is necessary for accurate, efficient, economical
is given below: process control. In areas where personnel will be affected,
noise prediction and control becomes a significant factor.
For a certain temperature control system, the ultimate sen-
sitivity Ku was found to be 1.5 kPa per °C, and the ultimate Engineering application guidelines, nomographs, and equa-
period Pu was found to be two minutes. A three mode PID con- tions presented in the following pages may be used to deter-
troller is required. mine the correct control valve configuration, size and flow
Using Fig. 4-24: characteristics, and to predict noise levels for most applica-
tions. The material presented here may also be used to evalu-
Proportional gain Kp: ate the performance of valves installed in existing plants.
Kp = 0.6 Ku = 0.6 (1.5 kPa/°C) = 0.9 kPa/°C The nomographs and equations given in this section are
used to calculate the flow coefficient (Cv or Cg) required for a
Integral time constant Ti: valve to pass the required flow. Most valve manufacturers pub-
Ti = Pu/2, Ti = 2/2 = 1.0 minute lish flow coefficients for each valve style and size.

Derivative time constant Td: A brief description of the two major components of a control
valve, the valve body and the actuator, is presented in Fig. 4-
Td = Pu/8, Td = 2/8 = 0.25 minutes 26.
Control Mode Considerations Control-Valve Bodies
The process control engineer has the responsibility for The control-valve body (see Fig. 4-27) regulates the rate of
matching the many and variable characteristics of the process fluid flow as the position of the valve plug is changed by force
to be controlled with the most effective control hardware avail- from the actuator. Therefore, the valve body must permit ac-
able.6 Fig. 4-26 provides guidelines for choosing the mode of tuator thrust transmission, resist chemical and physical
control for various types of applications based upon the proc- effects of the process, and provide the appropriate end connec-
ess reaction rate and size and speed of load changes. tions to mate with the adjacent piping. It must do all of this

4-18
 FIG. 4-25
Control Mode vs. Application

Process Load Changes

Control Mode Reaction Applications
Rate Size Speed
On-off: Two-position Slow Any Any Large-capacity temperature and level installations.
with differential gap Storage tanks, hot-water supply tanks, room heating,
compressor suction scrubbers.
Floating, Single-speed Fast Any Small Processes with small dead time. Industrial furnaces
with adjustable neutral and air conditioning.
zone
Proportional Slow to Small Moderate Pressure, temperature, and level where offset is not
Moderate objectionable. Kettle reboiler level, drying-oven
temperature, pressure-reducing stations.
Proportional-plus- Moderate Small Any Where increased stability with minimum offset and
derivative (rate) lack of reset wind-up is required. Compressor
discharge pressure.
Proportional-plus- Any Large Slow to Most applications, including flow. Not suitable for
integral (reset) Moderate batch operations unless overpeaking is allowed.
Proportional-plus- Any Large Fast Batch control; processes with sudden upsets; tem-
integral-plus-derivative perature control.

without external leakage. Most valve body designs are of the Control-Valve Actuators
globe style, but other configurations such as ball and butterfly
styles are available. Final selection depends upon detailed re- Pneumatically operated control-valve actuators are the
view of the engineering application. most popular type in use, but electric, hydraulic, and manual
actuators are also widely used. The spring-and-diaphragm
pneumatic actuator (see Fig. 4-28) is commonly specified, due
to its dependability and its simplicity of design. Pneumatically
FIG. 4-26 operated piston actuators provide integral positioner capabil-
ity and high stem-force output for demanding service condi-
Relationship of Major Components
tions, such as high differential pressure or long valve stem
travel distance.

FIG. 4-27
Push-Down-to-Close Valve Body Assembly

4-19
 FIG. 4-28
Typical Spring-and-Diaphragm Actuator Assemblies

Loading Pressure
Connection Vent

Diaphram
Diaphragm Case
Diaphragm Diaphram Plate
Casings Actuator
Diaphragm Spring Loading Pressure O-Rings
Diaphragm Plate Connection

Actuator
Stem Seal Bushing
Spring Seat Actuator Spring Actuator Stem
Spring
Adjustor
Stem Spring Seat Spring Adjustor
Connector Yoke
Travel Stem Connector Travel Indicator
Indicator Scale
Indicator Travel Indicator
Scale
Yoke
Valve
Plug Stem

Direct Acting Reverse Acting

Discussion of Flow Characteristics and existing flow. The change in flow rate is always proportional
to the flow rate just before the change in position is made for
Valve Selection a valve plug, disc, or ball position. When the valve plug, disc,
The flow characteristic of a control valve is the relationship or ball is near its seat and the flow is small, the change in flow
between the flow rate through the valve and the valve travel rate will be small; with a large flow, the change in flow rate
as the travel is varied from 0 to 100%. "Inherent flow charac- will be large. Valves with an equal-percentage flow charac-
teristic" refers to the characteristic observed during flow with
a constant pressure drop across the valve. "Installed flow char- FIG. 4-29
acteristic" refers to the characteristic obtained in service when
Example Flow Characteristic Curves
the pressure drop varies with flow and other changes in the
system.
Fig. 4-29 illustrates typical flow-characteristic curves. The
quick-opening flow characteristic provides for maximum
change in flow rate at low valve travel with a fairly linear
relationship. Additional increases in valve travel give sharply
reduced changes in flow rate. When the valve plug nears the
wide open position, the change in flow rate approaches zero.
In a control valve, the quick-opening valve plug is used pri-
marily for on-off service; however, it is also suitable for many
applications where a linear valve plug would normally be
specified.
The linear flow-characteristic curve shows that the flow rate
is directly proportional to the valve travel. This proportional
relationship produces a characteristic with a constant slope so
that with constant pressure drop (∆P), the valve gain will be
the same at all flows. (Valve gain is the ratio of an incremental
change in flow rate to an incremental change in valve plug
position. Gain is a function of valve size and configuration,
system operating conditions, and valve plug characteristic.)
The linear-valve plug is commonly specified for liquid level
control and for certain flow control applications requiring con-
stant gain.
In the equal-percentage flow characteristic, equal incre-
ments of valve travel produce equal percentage changes in the

4-20
teristic are generally used for pressure control applications. FIG. 4-30
They are also used for other applications where a large per-
centage of the total system pressure drop is normally absorbed Valve Sizing Equations
by the system itself, with only a relatively small percentage by
the control valve. Valves with an equal-percentage charac-
teristic should also be considered where highly varying pres- Flow Basis and Units Equation
sure drop conditions can be expected. Mass Flow with Specific w = N6FpCvY√XP1γ1

The modified parabolic-flow characteristic curve falls be- Weight, γ1
tween the linear and the equal-percentage curve.
Volumetric Flow with Spe-

√
Note: Where detailed process knowledge is lacking, as a rule cific Gravity, Gg X
of thumb, use equal-percentage characteristics at 70% opening q = N7FpCvP1Y
G TZ g
for the valve sizing.
Mass Flow with Molecu-
FUNDAMENTALS OF CONTROL VALVE
SIZING AND NOISE PREDICTION
lar Weight, M w = N8FpCvP1Y √XM
 TZ

Volumetric Flow with Mo-

Gas Service
Critical Pressure Drop — Critical flow limitation is a
lecular Weight, M q = N9FpCvP1Y 
√ X
MTZ
significant problem with sizing valves for gaseous service.
Critical flow is a choked flow condition caused by increasing
gas velocity at the vena contracta. The vena contracta is the 2. Calculate an initial, approximate required Cv based on
point of minimum cross-sectional area of the flow stream an assumed Rated Pressure Drop Ratio Factor, Xc. Initial
which occurs just downstream of the actual physical restric- assumptions for the value of Xc can be based on the gen-
tion. When the velocity at the vena contracta reaches sonic eral style of valve. See Fig. 4-32.
velocity, additional increases in pressure drop, ∆P, (by reducing 3. From the valve manufacturer’s sizing data select a spe-
downstream pressure) produce no increase in flow. cific valve type and size such that the listed Cv is equal
In the ISA sizing procedure critical flow limitations are ad- to or greater than the calculated Cv. The Xc associated
dressed by calculating Y, the expansion factor, for utilization with the listed Cv should then be used in the chosen siz-
within the actual sizing equation. ing equation to calculate a revised, required Cv. This it-
X eration process continues until the calculated Cv equals
Υ=1− Eq 4-9 the manufacturer’s listed Cv.
3Fk Xc
4. For a new valve selection a valve size is typically chosen
where, such that the maximum, calculated Cv is close to 75% to
k 85% of valve travel. This allows for process variability
Fk = Eq 4-10 while maintaining flow capability. The minimum, calcu-
1.4 lated Cv should typically occur at or about 10% of valve
Critical pressure drop, and thus critical flow, is realized travel.
when X ≥ FkXc. Therefore, since the flow can’t exceed that pro- FIG. 4-31
duced at the critical pressure drop the value of Υ in the follow-
ing sizing equations should never be less than 0.67. Numerical Constants for Gas and Vapor Flow Equations

Constant Units Used in Equations

X 1 N w q* p, ∆p γ1 τ d,D
Υ=1− = 1 − = 0.67 Eq 4-11
3Fk Xc 3 N5 0.00241 – – – – – mm
1000 – – – – – in
Likewise the value of X in the equations should never exceed N6 2.73 kg/h – kPa kg/m3 – –
FkXc. 27.3 kg/h – bar kg/m3 – –
63.3 lb/h – psia lb/ft3 – –
Sizing Calculation Procedure — The compressible
fluid sizing equations (see Fig. 4-30) can be used to determine N7 4.17 – m3/h kPa – K –
417 – m3/h bar – K –
the flow of gas or vapor through any style of valve. Absolute 1360 – scfh psia – °R –
units of temperature and pressure must be used in the equa-
tion. Most commonly the equations are used to calculate the N8 0.948 kg/h – kPa – K –
required Cv and thus valve size for a given set of service con- 94.8 kg/h – bar K –
ditions. The equations can likewise be rearranged to calculate 19.3 lb/h – psia – °R –
the flow or pressure drop for a given valve and set of service N9 22.5 – m3/h kPa – K –
3
conditions. The steps are: 2250 – m /h bar – K –
7320 – scfh psia – °R –
1. Select the appropriate sizing equations based on the
stated inlet conditions and units of measurement. The *q is in cubic feet per hour measured at 14.73 psia and 60°F,
or cubic meters per hour measured at 101.3 kPa and 15.6°C.
limitations on Y and X as discussed above must be ob-
served in all the sizing equations.

4-21
 FIG. 4-32
Typical Cv, Xc and FL Values for Valves*

Valve Normal Flow Characteristic

Style Body Size,
mm Equal Percentage Linear

Cv Xc FL Cv Xc FL
25 8 0.74 0.88 17 0.61 0.84
38 17 0.69 0.84 30 0.70 .082
50 25 0.70 0.85 62 0.68 0.77
Globe
63 49 0.66 0.84 84 0.71 0.81
75 66 0.66 0.82 118 0.70 0.82
100 125 0.67 0.82 181 0.74 0.82
150 239 0.74 0.85 367 0.78 0.84
200 268 0.60 0.85 526 0.74 0.87
25 16 0.53 0.86 – – –
50 59 0.53 0.81 – – –
75 120 0.50 0.80 – – –
Ball 100 195 0.52 0.80 – – –
150 340 0.52 0.80 – – –
200 518 0.54 0.82 – – –
250 1000 0.47 0.80 – – –
300 1530 0.49 0.78 – – –
50 60 0.37 0.69 – – –
75 111 0.40 0.69 – – –
100 238 0.40 0.69 – – –
150 635 .040 0.69 – – –
Butterfly
200 1020 0.40 0.69 – – –
250 1430 0.40 0.69 – – –
300 2220 0.40 0.69 – – –
350 2840 0.40 0.69 – – –
400 3870 0.40 0.69 – – –
*At approximately 70% of valve travel. Maximum valve capacity may be estimated using the values given in this
figure in conjunction with fig. 4-29. For a more detailed analysis of capacity capabilities of a given valve at other
percentages of travel, consult the valve manufacturer’s data.

5. Fp is the Piping Geometry Factor. It corrects the sizing Other valve configurations, such as ball and butterfly
equations for the effects of fittings such as reducers and valves, can be sized in a similar manner using the unique Xc
expanders that are attached to the valve body ends. Fp and Cv values derived by the manufacturers.
values can be determined via test or calculated per the
ANSI/ISA S75.01 standard. If the valve has no such fit- Aerodynamic Noise Prediction — Aerodynamic
tings attached, e.g., the nominal value size and nominal noise, the most common type of control valve noise, is the re-
pipe size are the same, then Fp = 1.0. Refer to the full sult of Reynolds stresses and shear forces that are the results
standard for the Fp calculations in cases where fittings of turbulent flow. Noise from turbulent flow is more common
do exist. in valves handling compressible gases than in those control-
ling liquids.

4-22
 FIG. 4-33 ously, as the velocity increases, an interchange of energy be-
tween the velocity and pressure heads forces a reduction in
Critical Pressure Ratios for All Liquids
the pressure.
If the velocity increases sufficiently, the pressure at the vena
contracta will be reduced to the vapor pressure of the liquid.
At this point, vapor cavities or bubbles, the first stage in cavi-
tation, appear in the fluid stream. Downstream from the vena
contracta, the fluid stream undergoes a deceleration process
resulting in a reversal of the energy interchange which raises
the pressure above the liquid vapor pressure.
The vapor cavities, or bubbles, cannot exist at the increased
pressure and are forced to collapse or implode. These implo-
sions are the final stage in the cavitation process. They pro-
duce noise, vibration, physical damage, and other
performance problems. In order to avoid cavitation completely,
the pressure at all points within the valve must remain above
the vapor pressure of the liquid. Cavitation can occur as the
result of changes in the mean pressures through the valve, but
also from localized changes due to flow separations and other
local disturbances that are not indicated by examining just the
mean inlet, vena contracta, and outlet pressures. Determining
when a problem-causing level of cavitation is present repre-
sents a considerable challenge. The reader is referred to ISA
RP75.23, "Considerations for Evaluating Control Valve Cavi-
The valve manufacturer should provide noise predictions or tation." This recommended practice provides more informa-
furnish adequate data to calculate expected noise levels since tion on the cavitation process as well as suggesting a common
noise characteristics vary greatly with the type and model of terminology and methodology for making safe valve selections
valve being considered or in use. Most control valve manufac- in cavitating applications. That recommended practice estab-
turers have valve sizing software available that include the lishes the definition of a cavitation index, Kc, as follows:
valve sizing and noise prediction routines.
P1 − Pv
Kc = Eq 4-12
Liquid Service P1 − P2
The procedure used to size control valves for liquid service The evaluation of Kc at any given set of service conditions
should consider the possibility of cavitation and flashing since can then be compared to the manufacturer’s valve operating
they can limit the capacity and produce physical damage to limit. As discussed in the recommended practice, the
the valve. In order to understand the problems more thor- selection of the appropriate operating limit for a given
oughly, a brief discussion of the cavitation and flashing process situation is dependent on the service conditions but should
is presented below. also consider other influences such as duty cycle, location,
desired life, and past experience. All of these point to the
Cavitation — In a control valve, the fluid stream is accel- need to consult the valve manufacturer when selecting a
erated as it flows through the restricted area of the orifice, valve for cavitation control.
reaching maximum velocity at the vena contracta. Simultane-
Flashing — The first stages of cavitation and flashing are
FIG. 4-34 identical; that is, vapor forms as the vena contracta pressure
Critical Pressure of Various Liquids is reduced to the vapor pressure of the liquid. In the second
stage of the flashing process, a portion of the vapor formed at
the vena contracta remains in the vapor state because the
kPa
downstream pressure is equal to or less than the vapor pres-
Ammonia 11 280 Isobutane 3648 sure of the liquid.
Argon 4865 Isobutylene 3999
After the first vapor cavities are formed, the increase in flow
n-Butane 3797 Methane 4604
rate will no longer be proportional to an increase in the square
Carbon Dioxide 7382 Nitrogen 3399 root of the body differential pressure. When sufficient vapor
Carbon Monoxide 3499 Nitrous Oxide 7223 has been formed, the flow will become completely choked. As
Chlorine 7711 Oxygen 5081 long as the inlet pressure (P1) remains constant, an increase
Dowtherm A 3206 Phosgene 5676 in pressure drop (∆P) will not cause flow to increase.
Ethane 4880 Propane 4249 Sizing Information — The following section is based on
Ethylene 5041 Propylene 4600 IEC-534, "Flow Equations for Sizing Control Valves." The
Fluorine 5574 Refrigerant 11 4378 reader is referred to that standard for more complete discus-
Helium 228 Refrigerant 12 4115 sion of these equations and methods. As that standard points
Hydrogen 1297 Refrigerant 22 4937
out, these equations are not intended for situations involving
mixed-phase fluids, dense slurries, dry solids, or non-Newto-
Hydrogen Chloride 8260 Water 22 118 nian liquids. In these cases the valve manufacturer should be
consulted for sizing assistance.

4-23
 2. Calculate the Cv required using the equation for nonva-
FIG. 4-35 porizing flow.
Liquid Valve Sizing Equations 3. Calculate the Cv using the equation for vaporizing flow.
Use Fig. 4-36 for value of Numerical Constants, N An initial assumed value of F1 can be taken from Fig.
4-32 or the manufacturer’s literature. FF, the liquid criti-
Flow Basis and Units Equation cal pressure ratio factor, can be found from Fig. 4-33
based on the critical pressure and inlet vapor pressure
Nonvaporizing Mass Flow w = N6FpCv√

(P1 − P2)γ1 for subject liquid. Fig. 4-34 lists critical pressures for
with Specific Weight, γ1 some common fluids. The user must at this point iterate
through this calculation accounting for the variation in
Nonvaporizing Volumetric
Flow with Relative q = N1FpCv√
P −P
G
 1 2
F1 and valve-rated Cv due to valve style, size, trim, flow
f
Density, Gf direction, etc.
4. Select the higher of the two calculated Cv’s as the re-
Vaporizing Mass Flow with w = N6FLCv√(P1 − FFPv)γ1
 quired Cv.
Specific Weight, γ1
5. From the valve manufacturer’s sizing data, select a spe-
Vaporizing Volumetric Flow
with Relative Density, Gf q = N1FLCv √
G

P −F P
1 F v cific valve type and size such that the listed Cv is equal
to or greater than the calculated Cv.
f
6. See the previous section on Cavitation and consult the
FIG. 4-36 manufacturer’s data for appropriate valve cavitation op-
erating limits.
Numerical Constants for Liquid Flow Equations
Constant Units Used in Equations
INSTALLATION, TROUBLESHOOTING, AND
CALIBRATION
N w q p, ∆p d, D γ1 v
N1 0.0865 – m3/h kPa – – –
Installation and Troubleshooting
3
0.865 – m /h bar – – – Control system troubleshooting logically falls into two cate-
1.00 – gpm psia – – – gories: (1) the repair of control systems that previously func-
N6 2.73 kg/h – kPa – kg/m3 – tioned well, and (2) the successful modification of poorly
27.3 kg/h – bar – kg/m3 – commissioned systems that have never worked properly due
63.3 lb/h – psia – lb/ft3 – to improper application, poor design, faulty hardware, or im-
proper operating procedures. Different techniques are em-
The ISA methodology recognizes the impact of service con- ployed for each category.
ditions that will cause the liquid to vaporize at some point Failed Systems
between the inlet and outlet of the valve. This vaporization
results in either cavitation or flashing, causing a breakdown • Control system malfunctions normally are reported by
in the normal relationship between Cv and √ 
∆P and ulti- the process operator. A discussion with the operator
mately a limit to the flow through the valve regardless of an should yield some clues as to the source of the problem,
increasing pressure drop caused by decreasing P2. The recog- since he has probably been observing it for several hours,
nition of this comes in the form of a separate sizing equation or days.
for each regime, nonvaporizing and vaporizing. Each must be • The next step is to use the "process of elimination" to
solved and then the larger calculated Cv chosen as the required localize the problem. If replacement of an element with
value. a known good one causes the problem to disappear, this
is usually conclusive! Often this simple approach of
This discussion of liquid sizing will be further restricted to: parts changing will save time by avoiding a detailed sys-
1. Turbulent flow streams: There are usually flow streams tem analysis. However, if the situation permits, the "bad"
that are not either high viscosity or low velocity. The ma- part should be temporarily re-installed to verify a "hard"
jority of process plant control valves do operate in the failure rather than a "hung-up" condition which is often
turbulent regime, however if the Reynolds number for a reset by the procedure of substitution.
process is less than 4000 the reader is referred to the ISA • A "detailed system analysis" may be required if a control
standard where a non-turbulent flow correction method system has a number of interactive or serially dependent
can be found. components and especially if more than one component
2. Valve installed without fittings attached to the valve is faulty. The process-of-elimination tests may have
ends: When fittings are present there are, as with the shown some conflicting results in this case. A complete
previous gas sizing discussion, necessary modifications control system diagram should be used to help isolate
to the sizing equations to accommodate the additional possible problem areas, separate cascade loops, etc. This
disturbance to flow. This discussion will be limited to the step usually requires the services of the control engineer
case where there are no fittings attached, therefore the or someone familiar with all the control loop components
valve size and pipe size are the same, Fp = 1.0. Refer to and their functions. Caution should be taken to assure
the full ISA standard for the proper methods if fittings that control system response is observed over a sufficient
are present. length of time to detect problems in slow changing proc-
esses. Strip chart recorders are very useful in this analy-
Sizing Calculation Procedure — sis. Conversely, sequential event recorders may be
1. Select the appropriate sizing equations based on the needed to diagnose intermittent problems which occur
stated inlet conditions and units of measurement from only for very brief periods at irregular intervals. Re-
Fig. 4-35. corders are available which can resolve events to within

4-24
 FIG. 4-37
Common Measurement Problems

Variable Symptom Problem Source Solution

Zero shift, air leaks in Excessive vibration from posi- Use independent transmitter mtg., flexible pro-
signal lines. tive displacement equipment. cess connection lines. Use liquid filled gauge.
Variable energy con- Change in atmospheric pressure. Use absolute pressure transmitter.
Pressure

sumption under tem-

perature control.
Unpredictable trans- Wet instrument air. Mount local dryer, use regulator with sump,
mitter output. slope air line away from transmitter.
Permanent zero shift. Overpressure. Install pressure snubber for spikes.
Transmitter does not Liquid gravity change. Gravity compensate measurement, or recali-
Level (diff. press.)

agree with level. brate.

Zero shift, high level Water in process absorbed by Use transmitter with integral remote seals.
indicated. glycol seal liquid.
Zero shift, low level Condensable gas above liquid. Heat trace vapor leg. Mount transmitter
indicated. above connections and slope vapor line away
from transmitter.
Noisy measurements, Liquid boils at ambient Insulate liquid leg.
high level indicated temperature.
Low mass flow Liquid droplets in gas. Install demister upstream; heat gas upstream
indicated. of sensor.
Flow

Mass flow error. Static pressure change in gas. Add pressure recording pen.
Transmitter zero shift. Free water in fluid. Mount transmitter above taps.
Measurement is high. Pulsation in flow. Add process pulsation dampner.
Measurement error. Non-standard pipe runs. Estimate limits of error.
Temperature

Measurement shift. Ambient temperature change. Increase immersion length. Insulate surface.
Measurement not repre- Fast changing process tempera- Use quick response or low thermal time con-
sentative of process. ture. stant device.
Indicator reading Electrical power wires near Use shielded, twisted pair thermocouple exten-
varies second to second. thermocouple extension wires. sion wire, and/or install in conduit.

one millisecond for use in troubleshooting the fastest ations such as compressor surge control are too fast for
control loops. human reaction times and must be verified by other
In a well designed control loop, the process is the slowest means. Many troublesome control loops are mistakenly
responding element. Thus the rate of change of a disturbance declared to be faulty because it is not always realized
as it initially propagates around the loop should point to the that there is a trade-off between speed of response and
origin of the problem. In some cases, where multiple symp- loop stability. Also, there is an inherent interaction be-
toms are present, answering the question: "What is common tween certain control loops. An example is the interac-
to all symptoms?" will locate the problem. It is often necessary tion between temperature and pressure control loops in
to make observations over a period of time plus taking parallel a distillation column.
measurements to completely document the problem and point • If the process variable is controllable with the loop in the
to the source of the trouble. manual mode, note which measurement must be ob-
served to make the decision to open or close the valve;
Poorly Commissioned Systems then determine whether the automatic control system
makes adjustments based upon the same variables. For
Some control systems may not have been properly commis- example, it may be determined that a single variable
sioned or were not adequately designed. In this case it is nec- controller is insufficient, if the controlled variable devi-
essary to determine how the system is meant to function ates too much due to disturbances, feed composition
before proper control can be implemented. changes, etc. In this case it may be necessary to change
• The first step, after determining how a particular control the control scheme to include override control arrange-
loop is meant to function, is to put the final control ele- ments or two or more controllers in cascade to achieve
ment (valve, usually) on manual control and adjust it for the desired automatic control.
the desired response of the controlled variable. If manual
adjustment of the manipulated variable does not cause
Poor Performance
some response in, or controlling action on, the controlled Two major sources of poor performance in control loops are:
variable then the control strategy is not suitable and (1) excessive time delay between the actuation of the control
some redesign is in order. However, some control situ- element (valve) and the resulting change in the measurement

4-25
of the controlled variable, and (2) a variation in loop gain due calibrators for generating process temperatures are of three
to a change in process conditions. varieties: (1) electrically heated direct air, (2) hot oil bath, and
• Excessive time delay can result from process lag and (3) a gas fluidized bed using a thermally conductive powder.
dead-time and from instrument dead-time (such as the All these require a considerable amount of time (20-40 min-
cycle time of a chromatograph) and should be compen- utes avg.) to reach a steady state temperature. Electronic tem-
sated for in the control system. To achieve stability in a perature transmitters which use resistance temperature
system with excessive time delay, the controller gain devices (RTD) have an advantage in that the sensing element
must be low and reset (integral) time must be long. Re- calibration needs only a single point check and the transmitter
covery from process upsets will be slow. When dead-time can be calibrated electronically using the known intrinsic
is present, loop performance will deteriorate proportion- properties of the sensor. Field checking of these transmitters
ally to the square of the dead-time. is practical.
• Variable loop gain is a common problem in cascade con- Thermocouple transmitter calibration requires an accurate
trol loops where the secondary controller is a flow con- millivolt source to simulate the thermocouple signal at the
troller without square root extraction on the measure- upper and lower limits of the temperature range of the trans-
ment signal. Since an orifice plate differential pressure mitter being calibrated. However, since all thermocouple ta-
signal is proportional to the square of the flow, the set- bles show millivolt readings based on some reference
point to the secondary controller is also a "flow-squared" temperature, usually 0°C, some sort of reference temperature
signal. This setpoint signal is provided by the output of compensation must be done. Historically, a reference junction
the primary controller in a cascade configuration. There- in an ice bath was used to provide ice point compensation. Also
fore, the primary loop gain increases rapidly as flow is a reference junction at ambient temperature was often used
reduced; if that controller was tuned at high flows it will in conjunction with a thermometer to read the ambient tem-
become unstable at low flows. perature and provide a reference millivoltage. Modern devices
make temperature transmitter calibration much simpler be-
Calibration cause of built-in reference compensation. Some digital devices
Calibration of measurement transducers is a vital part of merely require the entering of the desired temperature and
instrument maintenance and should be performed on a regu- they provide the correct millivoltage to simulate a thermocou-
lar basis. ple with automatic reference compensation. Also, compact
electronic modules are available for ice point compensation
Pressure transmitters — Moderate to high pressure and thermocouple linearization. Some electronic calibrators
units are usually calibrated with a "dead weight tester." This supply the millivolt signal as well as power for two-wire trans-
is a hydraulic device in which weights are added to one side of mitters and indicators. The millivolt source is set to the values
a hydraulic circuit to generate a known pressure which is ap- which correspond to the upper and lower temperature limits
plied to the input of the transducer under calibration. If the of the range of the transmitter and the transmitter is adjusted
transducer is to be calibrated in units of absolute pressure, a so that it produces the desired zero-scale (20 kPa, 4 mA, etc.)
barometer should be used to measure atmospheric pressure and full-scale (100 kPa, 20 mA, etc.) outputs. When field
and an adjustment of the added weights made. The dead checking electronic devices, all safety codes must be observed.
weight tester is normally a test bench device. Low pressure Calibrators which are battery powered and intrinsically safe
units are calibrated with a pneumatic calibrator which sup- are recommended.
plies a precise air pressure. Since they are easily adjustable,
20-100 kPa devices are usually calibrated by this method.
Pneumatic calibrators are also test bench devices. Field or COMPUTER SYSTEMS
process area checks are often made with a hand-held bulb
pump and an accurate pressure gauge.
Analog Computers
Differential pressure transmitters — Differential A process control analog computer is composed of electronic
pressure level transmitters are usually calibrated with a mer-
modules such as amplifiers, summers, controllers, multipliers,
cury manometer or a low level pneumatic calibrator. If remote
dividers, square and square root devices, lead and lag mod-
seals are used, the seal paddle relative elevation should be the
ules, limiters, and special dead-time devices. These modules
same during calibration as when the transmitter is installed.
are based around the integrated circuit operational amplifier
Zero elevation and suppression can be done with a bench cali-
and are interconnected to implement the desired control strat-
brator, but if the transmitter has a zero shift with changes in
egy. The process control analog computer is very reliable, low
static pressure, this shift must be removed after the transmit-
ter is installed. FIG. 4-38
Calibration of differential pressure transmitters for flow Square Root Input/Output Relationship
measurements is done using the same equipment as for levels.
An additional consideration must be made for those electronic
transmitters which have a square root extractor as an integral Input Output
part of the transmitter. In this case, the transmitter output is % of span Signal level % of span Roots
the square root of the input differential pressure. Fig. 4-39 0 20 kPa /4 mA 0.0 0
shows the square root input-output relationship.
9 — 30.0 3
Temperature transmitters — Temperature transmit- 25 40 kPa /8 mA 50.0 5
ters are by far the most difficult to calibrate because of the 50 60 kPa /12 mA 70.7 7.07
difficulty in generating precise controlled temperatures. Cali- 75 80 kPa /16 mA 86.6 8.66
bration of pneumatic filled-bulb type temperature transmit-
100 100 kPa /20 mA 100.0 10
ters is normally a lengthy test bench procedure. Temperature

4-26
in cost, and easy to use, but has been made obsolete to some Distributed control systems (DCS) — In a DCS in-
degree by the more versatile digital computer. stallation the controllers and measurement circuits are modu-
larized in small groups (e.g. eight controllers per module) for
Digital Computers greater security against failure. The controllers exist in a com-
Digital computers are attractive for many applications in the bination of hardware and software and may be part of a control
process control field because of their speed, accuracy, flexibility, scheme programmed in a master DCS system. The controllers
display and logging capabilities, and ability to perform complex may also receive setpoints from a separate supervisory com-
calculations and store and transmit vast amounts of data. puter system. DCS systems support a variety of communica-
tion methods, such as a high speed "data highway" serial data
Digital computers come in almost any size or shape, ranging
transmission concept which can interface to many different
from tiny single board special purpose devices to large data
computers. Any digital first-level control system must be
processing computers.
backed up by a battery powered uninterruptible power supply
Programmable logic controllers (PLC) — A special (UPS) to prevent loss of control of the process during AC power
purpose class of microcomputers designed to implement a se- line interruptions.
ries of sequential functions such as ladder network diagrams.
They are best suited to batch type operations or machine con-
trol. Some examples are: dehydrator control, compressor load- ANALYTICAL INSTRUMENTS
ing control, engine start-up sequencers, and product blenders.
Some PLC’s provide for data logging and/or display of infor- Cyclic Analyzers
mation on CRT terminals. Many analytical instruments are cyclic, or sampled data de-
vices, such as the chromatograph. These automatically take a
Microcomputers — Microcomputers usually include a
higher level software system and more versatile input/output sample of the process stream, analyze it, and transmit the
hardware than PLC’s. They are often used as the process con- results to the desired device. Since most analysis cycles take
from one to 20 minutes to complete, considerable "dead-time"
trol computer in small to medium sized installations such as
gas plants. is introduced into a control loop using this type of measure-
ment. Dead-time compensation should be included in the con-
Minicomputers — The minicomputer incorporates a trol scheme for proper control response. Process stream
more complex, higher speed arithmetic logic unit as the cen- sampling is an item of vital concern in good chromatography
tral processor. Minicomputers accommodate large amounts of or with any analysis technique. Some points of consideration
memory and may include a high speed disk unit for data and are:
program storage as well as input/output ports to interface with • Sample probe: Must be located at a point in the process
a variety of peripheral equipment. Minicomputers are nor- where the material to be analyzed is in the desired phase
mally used as the process control computer in larger installa- (vapor or liquid) at sufficient pressure and flow.
tions such as refineries, chemical plants, etc. • Sample lines: Should be kept as short as possible for
Process input/output equipment — In addition to minimum transport time. May need to be heat-traced to
the array of printers, CRT terminals, disk units, etc., normally keep vapor samples from condensing. Sample flow must
found with a computer system, there is a process input/output be great enough to completely flush the line between
system. This is a set of electronic modules, usually a "card- analysis cycles.
cage" type subassembly, used to interface the process signals • Sample filters: Used to keep any particulate matter out
to the computer. Most process I/O signals fall into one of the of the analyzer. Knock-out pots or other devices may be
following groups: required to remove liquid condensate.
• Analog (flows, pressures, levels, temperatures, etc.) FIG. 4-39
• Digital (on-off status sense or actuation)
• Pulse (tachometers, counters, etc.) Typical Process Chromatograph System
• Serial (coded data)
Carrier Gas Vent
DIGITAL FIRST-LEVEL CONTROL
SYSTEMS
First-level controllers are those which actually manipulate Analyzer Programmer
the valves or other final control devices to maintain the proc-
ess variable at a desired setpoint.
Individual controllers — Controllers located either on Sample
the control room panel board or in the process area. These Probe
operate independently or may receive a setpoint from another Sample
controller or computer system in a cascade arrangement. Conditioner
Direct digital controllers (DDC) — DDC controllers Readout
exist as algorithms in the software of a digital computer, and, Calibration
through the appropriate transducers, continually sample the Gas
respective process measurements, compare them with their
corresponding setpoints, and manipulate the appropriate Sample Return
valve or other final control device. Reliability of the computer
is essential in a DDC system.

4-27
 FIG. 4-40 2. Instrument Society of America, "Process Instrumentation Termi-
nology," ANSI/ISA-S51-1-Latest Edition.
Continuous Analysis Instruments
3. Anderson, Norman A., "Instrumentation for Process Measure-
Analyzer Detector (Sensing Method) ment and Control," Chilton Company, 1980.

Liquid Density Resonant Mass 4. Liptak, Bela G., "Instrument Engineers Handbook," Chilton Co.,
Liquid Viscosity Viscosimeter 1970, 1982.
Liquid Level Ultrasonic, Gamma Ray 5. Considine, D. M., "Power Instruments and Control Handbook"
Gas Density Torque Measurement McGraw-Hill, Latest Edition.
Sulfur in Oil X-Ray Attenuation 6. Considine, D. M., "Process Instruments and Controls Handbook,"
pH Electrolytic McGraw-Hill, Latest Edition.
Oxygen Paramagnetic, Coulometric 7. Young, William J., "Organization of Instrumentation Guidelines
Trace Moisture Electrolytic Cell for Standard Instruments and Control Systems," PennWell Pub-
Nitrous Oxides Chemiluminescence lishing Co., 1982.
CO, CO2, SO2 Infra-red Absorption 8. Ogata, Katsuhiko, "Modern Control Engineering," Prentice-Hall,
H2S, SO2 Ultra-violet Inc., 1970.
Light Hydrocarbons Chromatograph 9. Zoss, L. M., "Applied Instrumentation in the Process Industries,"
Heating Value Calorimeter, Chromatograph Gulf Publishing Co., 1979.
10. Cheremisinoff, P. N., "Engineering Measurements and Instru-
• Sample pumps: Required for certain low pressure proc- mentation," Marcel Dekker, 1981.
esses. Vacuum aspirators may be used for low sample 11. Perry, R. H., "Chemical Engineers’ Handbook," Section 22,
flows. McGraw-Hill, Latest Edition.
Other cyclic or batch type instruments determine such 12. Cheremisinoff, P. N. and Perlis, H. J., "Analytical Instruments
quantities as vapor pressure, dew point, end point, flash point, and Measurements for Process and Pollution Control," Ann Arbor
Btu content, and sulfur content. Science, 1981.
Continuous Analyzers 13. Hewson, John E., "Process Instrumentation Manifolds," Instru-
ment Society of America, 1981.
Other analytical instruments perform continuous analyses
of a process stream for a specific parameter such as oxygen 14. Whitaker, Norman R., "Process Instrumentation Primer," Petro-
content, gas density, etc. leum Publishing Co., 1980.
Some continuous analyzers pump or flow a sample through 15. Scott, R. W., "Developments in Flow Measurement," Applied Sci-
a detector cell. Others place the detector element directly in ence Publishers, 1982.
the process stream. The methods of detection vary widely and 16. Shinskey, F. G., "Process Control Systems," McGraw-Hill, 1979.
may include the use of nuclear radiation, optics, vibrating
17. Shinskey, F. G., "Distillation Control for Productivity and Energy
reeds, flotation cells, catalytic "burners," and other methods.
Conservation," McGraw-Hill, 1977.
Fig. 4-39 shows a typical block diagram for a process chro- 18. Shinskey, F. G., "Energy Conservation Through Process Control,"
matograph installation. Fig. 4-40 identifies several types of Academic Press, 1978.
continuous process analytical instruments and a type of detec-
tion used by each. 19. Whalen, B. R., "Basic Instrumentation," University of Texas Pe-
troleum Extension Service, 1983.
20. Deshpande, P. B. and Ash, R. H., "Computer Process Control,"
REFERENCES Instrument Society of America, 1981.
1. Ziegler, J. G., and Nichols, N. B., "Optimum Settings for Auto- 21. Omega Engineering, Inc., "Temperature Measurement Hand-
matic Controllers," ASME Transactions, 1942, p.759-768. book," Omega Engineering Inc., Latest Edition.
22. Badavas, Paul C., "Direct Synthesis and Adaptive Controls,"
Chemical Engineering, CE Refresher Part 5, February 6, 1984,
BIBLIOGRAPHY p. 99-103.
1. Instrument Society of America, "Dictionary of Industrial Digital 23. "Practical Process Instrumentation and Control," McGraw Hill
Computer Terminology," 1972. Publications Co., 1980

4-28