2006 Engineering Design Seminar

Instrumentation

Daniel O’Leary
UOP LLC

EDS 2006/Inst-1

UOP Confidential - Do Not Copy 1

 Introduction

„ Instructor
– Name Daniel J. O’Leary
– Background/experience
„ Session Objectives
– Obtain basic understanding of concepts and
fundamentals applied to instrumentation
– Review from simple primary elements to complex
control systems existing in Industry
– Convey UOP’s process control philosophy

EDS 2006/Inst-2

The session objectives start with a review of the basic concepts of the feedback
control loop. The feedback control loop is sectioned into individual components
with a discussion on the theory of the more common elements.

Several factors affect control loop performance. Among these are transmitter
performance, control valve performance, and the speed of response of today’s
Digital Control Systems (DCS). Each of these topics will be covered in detail.

Upon completion of the concepts of the feedback control loop and its individual
components, these concepts will be applied to some of the more common control
systems and applications.

One major point to keep in mind is that typically more than one solution may be
available to solve a particular control application. The contents of this material will
convey UOP’s process control philosophy as well as some of the background behind
the philosophy.

UOP Confidential - Do Not Copy 2

 Session Material

„ Presentation Material
– General Instrumentation
„ Text Material
– Instrumentation
– Glossary of Terms

EDS 2006/Inst-3

As a handout each student will receive a copy of the power point presentation. User
notes have been provided for the majority of the slides. Additional notes should be
taken by the student when specific information is presented as examples not covered
in the original presentation.

Each student also receives text material and a glossary of the more common process
control jargon. The text material provides another level of detail above and beyond
the power point presentation. If a specific topic sparks your interest, this text
material should be reviewed for additional details.

UOP Confidential - Do Not Copy 3

 Session Overview

„ Terminology
„ P&ID Representation
„ Feedback Control Loop
„ Individual Components of Feedback Loop
„ DCS System Requirements
„ Process Control Applications

EDS 2006/Inst-4

We will review UOP’s standard nomenclature for instrument representation on the

P&IDs. UOP follows for the most part ISA S-5 symbology, but this nomenclature
can easily be modified include additional items not covered by ISA.
We will review the FEEDBACK control loop and identify the various components
of the feedback control loop. We will also spend some time looking at the details of
each of the components along with a review of the most commonly-used
components.
In today’s modern world the distributed control system has proven to be a
monumental advancement in process control; but due to the rapid improvements in
system components we will limit our discussion to a few basic system requirements.
UOP process units do not require DCS systems to meet guarantees and UOP does
not mandate the use of them. However taking advantage of their use, can only
improve the overall operation and optimize performance.
Lastly we will apply the knowledge learned about the feedback control loop and
apply it to various process control applications. Starting with the simple feedback
control loop, we will investigate some common process load disturbances and build
on various improvements in the basic control loop to counterbalance these
disturbances.

UOP Confidential - Do Not Copy 4

 Terminology and P&ID Representation

„ Comprehensive Dictionary of Measurement and

Control – Second Edition, Instrument Society of
America
– Glossary of Terms
– P&ID Instrument Signals

EDS 2006/Inst-5

The Glossary of Terms contains definitions of some of the more common control
terms as applied to the Process Industry. For a more comprehensive listing of
process control terminology and definitions, refer to:
Comprehensive Dictionary of Measurement and Control Second
Edition 1991 Instrumentation,
Systems, and Automation Society (formerly called: Instrument
Society of America) 67 Alexander
Drive PO Box 12277
Research Park North
Carolina 27709 USA www.isa.org

The instrument symbols provided are UOP’s typical P&ID representation and are
common industry practice. However these symbols may vary from client to client;
and a basic understanding of these symbols is required in order to interpret the
meaning of any instrumentation/control loop represented on the P&ID’s.

UOP Confidential - Do Not Copy 5

 Terminology

„ What is an Instrument?
– A device used directly or indirectly to measure and/or
control a variable
„ What is Process Control?
– The regulation or manipulation of variables influencing
the conduct of a process in such a way as to obtain a
product of desired quality and quantity in an efficient
manner

EDS 2006/Inst-6

The Process Industry is governed by economics and economics is a major player in

determining whether or not it is an “efficient manner”.

As an example, separation of the isomer para-xylene (P-xylene) from its

counterparts (meta, ortho, and ethyl-benzene) by way of distillation is economically
not justified. The boiling points of the individual isomers are: O-xylene (292 °F),
M-xylene (282 °F), P-xylene (281 °F), and ethyl-benzene (277 °F).

Two competing technologies, other than distillation, exist today for the efficient
separation of P-xylene. These are selective adsorption (UOP’s Sorbex Process), and
crystallization technology (freezing point of P-xylene is 56 °F, next closest isomer
is O-xylene with a freezing point of -13 °F).

UOP Confidential - Do Not Copy 6

 Terminology
(Feedback Control Loop)
Controller

Set
+ Error Signal Function
Point Generator
-

Process
Output
Variable
Measuring Final Control
Means Element
Measured Manipulated
Variable Variable

Process

Load
Variable

EDS 2006/Inst-7

This block diagram outlines the individual components that make-up the feedback
control loop. The glossary of terms should be reviewed for definitions of the
various components.

Examples of the process are flow, pressure, level, temperature, etc. In the process
industry, the primary work horse for automatic control is the three mode PID
feedback controller.
P is proportional action, most commonly known today as gain. I is integral action
and D is derivative action. A more detailed discussion of each type of action, when
to use or not use various combinations, etc will follow when we investigate the
controller component in detail.

UOP Confidential - Do Not Copy 7

 P&ID Instrument Symbols
(Instrument Line Symbols)

CAPILLARY TUBING (FILLED SYSTEM)

ELECTRIC BINARY SIGNAL (ON-OFF)
ELECTRIC SIGNAL
ELECTROMAGNETIC OR SONIC SIGNAL
HYDRAULIC SIGNAL
PNEUMATIC SIGNAL
PROCESS CONNECTION OR MECHANICAL LINK
SOFTWARE LINK
IST-LGND-001A

EDS 2006/Inst-8

In today’s environment the more common symbols used are the electrical binary
signal, electrical signal, pneumatic signal, and software link.

The electrical binary signal represents a discrete input or output signal and generally
is a 24 Vdc, 48 Vdc, 110 Vac, or 220 Vac signal.

The most common input/output electrical signal is the 4-20 ma signal. However
with the advent of digital signals, a variety of proprietary signals exist today. A
brief discussion follows this section.

The pneumatic signal is typically reserved for the final control element and in most
cases is the control signal to the control valve. The industry standard is 3 - 15 psig
(0.2 - 1.0 kg/cm2(g)).

UOP Confidential - Do Not Copy 8

 P&ID Instrument Symbols
(General Instrument Symbols)
INSTRUMENT LOCATED IN FIELD

INSTRUMENT MOUNTED ON LOCAL PANEL

INSTRUMENT MOUNTED ON CONSOLE OR PANEL

INSTRUMENT MOUNTED ON RACK OR BEHIND PANEL

INSTRUMENT WITH DUAL FUNCTION (OR SERVICES)

DIAPHRAGM SEAL
IST-LGND-001B
ORIFICE PLATE

EDS 2006/Inst-9

The circle represents a tangible object, i. e. a physical piece of hardware. The

various lines (double line, single line, dashed line, etc) are used to distinguish the
location of the device.

All in-line devices (primary flow elements, control valves, etc.) are examples of
instruments located in the field. Other examples of instruments located in the field
are temperature and pressure gauges.

Packaged units, such as compressors, instrument air dryers, centrifuges, etc., are
often supplied with local control panels in which many of the instruments may be
installed in the local panel supplied with the packaged unit.

Board-mounted instruments are limited today because of the distributed control

systems (DCS). Examples are hard-wired emergency shutdown switches and hard-
wired annunciator alarms.

UOP Confidential - Do Not Copy 9

 P&ID Instrument Symbols
(Distributed Control-Shared Display Symbols)
SHARED DISPLAY DEVICE WITH LIMITED
ACCESS TO ADJUSTMENTS

SHARED DISPLAY INDICATOR, CONTROLLER, OR

OTHER DEVICE WITH OPERATOR ACCESS TO
ADJUSTMENTS

* AH SOFTWARE ALARMS WITH SHARED DISPLAY DEVICE

* AL ( * IS MEASURED VARIABLE)

* AH CRITICAL SOFTWARE ALARM

( * IS MEASURED VARIABLE)
* AL

CRIT CRITICAL SHUTDOWN ALARM

DATA RECORDING FUNCTION

ACCESSIBLE TO OPERATOR

INSTRUMENTATION FOR ADVANCED PROCESS

CONTROL AND OPTIMIZATION FUNCTION IST-LGND-001C

EDS 2006/Inst-10

The majority of new installations today are equipped with DCS’s. The flexibility of
the DCS allows for easy configuration of many control functions including
controllers, indicators, alarming functions, math functions, etc.

Essentially all main panel instruments have been replaced with a DCS configurable
counterpart. One major advantage of the DCS is that the configuration can be easily
modified compared to recalibration of its hardware counterpart.

As an example a current switch, which was used frequently for detecting say low
process pressure, would have to be recalibrated by a qualified technician if the
alarm point was changed. In the DCS changing the alarm point takes only seconds
to implement.

UOP Confidential - Do Not Copy 10

 P&ID Instrument Symbols
(Miscellaneous Symbols)

LOGIC SYSTEM WITH LIMITED ACCESS TO ADJUSTMENTS

FUNCTION IDENTIFICATION

REFER TO LIST BELOW FOR IDENTIFICATION = INPUT/OUTPUT OF THE FOLLOWING

SUM HIGH LIMITER

DESIGNATION SIGNAL
.
. DIVIDE LOW LIMITER A ANALOG
D DIGITAL
E VOLTAGE (EMF)
EXTRACT SQ RT DIFFERENCE CURRENT
N HYDRAULIC
MULTIPLY O ELECTROMAGNETIC
HIGH SELECTOR
P PNEUMATIC
R RESISTANCE
d
LOW SELECTOR dt DERIVATIVE IST-LGND-001D

EDS 2006/Inst-11

A variety of mathematical functions exist in today’s DCS’s. Among these functions

are summers, multipliers, signal selectors, signal limiters, etc.

Field-mounted signal converters, such as current to pneumatic (I/P) transducers,

convert the 4 -20 ma electrical signal to a 3 -15 psig pneumatic signal compatible
with the operation of the final control element.

Safety, interlock, and sequential logic systems have a variety of analog and discrete
I/O modules. Serial communications are often provided between the logic systems
and the DCS’s used for basic process control.

UOP Confidential - Do Not Copy 11

 P&ID Instrument Symbols
(Functional Identification of Instruments)
FIRST LETTER SUBSEQUENT LETTERS

MEASURED VARIABLE MODIFIER READOUT OUTPUT

A ANALYSIS ALARM

B BURNER FLAME

C COMPENSATED CONTROL

D DIFFERENTIAL

E PRIMARY ELEMENT

F FLOW RATIO (FRACTION)

G GLASS, GAUGE

H HAND (MANUAL)

I CURRENT INDICATE

J POWER SCAN

K TIME RATE OF CHANGE CONTROL STATION

L LEVEL LIGHT

IST-LGND-001E

EDS 2006/Inst-12

The following tables outline the functional identification of the basic

instrumentation loops. UOP will include similar tables on the legend P&ID.

The loop identification and tag number is contingent upon the number of letters, the
letter sequence and the quantity of loops. The table is divided into two primary
sections: 1) first letter and 2) subsequent letters. The typical flow loop consists of a
Flow Element, Flow Transmitter, Flow Indicating Controller, and Flow Valve and
would be shown on the P&ID with the following tag numbers, respectively: FE-001,
FT-001, FIC-001, and FV-001. The numeric, per UOP general practice, is to
number all loops sequentially (001,002, etc), while the alpha-characters are
designated by type (flow, pressure, temperature, etc).

UOP Confidential - Do Not Copy 12

 P&ID Instrument Symbols
(Functional Identification of Instruments cont’d)
FIRST LETTER SUBSEQUENT LETTERS

MEASURED VARIABLE MODIFIER READOUT OUTPUT

O ORIFICE

P PRESSURE, VACUUM POINT (TEST CONN)

Q QUANTITY INTEGRATE, TOTALIZE

R RECORD

S SPEED, FREQUENCY SAFETY SWITCH

T TEMPERATURE TRANSMITTER

U MULTIVARIABLE

V VIBRATION VALVE

W WEIGHT WELL

X SKIN

Y RELAY, COMPUTE

Z POSITION

IST-LGND-001F

EDS 2006/Inst-13

The tables can be modified to include additional items as needed. Therefore each
set of P&ID may have unique legends.

UOP Confidential - Do Not Copy 13

 Feedback Control Loop

„ Signal Transmission And Transmitters

– Analog – Pneumatic/Electronic
– Electronic Analog/Digital
– Transmitter Performance
– Fieldbus
„ Loop Components
– Process
– Measuring Means
• Temperature, Flow, Pressure, Level, Analysis

EDS 2006/Inst-14

Over the past 1/2 century, technology has advanced from pneumatic to
microprocessor-based digital transmitters. Digital transmitters offer various
advantages over its analog counterpart, but the speed of response has become an
issue with respect to control performance and transmitter performance can vary
from vendor to vendor.

UOP Confidential - Do Not Copy 14

 Analog – Pneumatic vs. Electronic

„ Pneumatic
– Dry air used as transmission medium
– Standard range 3-15 psig (0.2-1.0 kg/cm2(g))
(corresponding to 0-100% of signal)
– Transmission response is typically slow
„ Electronic
– Low power level system (0-24 volt DC)
– Standard range 4-20 milli-amperes (mA)
– Transmission response is instantaneous

EDS 2006/Inst-15

Fast responding pneumatic loops were generally limited to local field-mounted

controllers or at best short distances to the control room.

Electronic transmitters proved to be exceptionally better than the pneumatic

transmitters. Transmission response was greatly enhanced with the electronic
analog instruments, the overall loop performance was improved, and transmission
distances were not the limiting factor for the typical process installations.

UOP Confidential - Do Not Copy 15

 Electronic Analog vs. Smart Digital

„ Electronic Analog
– A signal representing a variable that is continuously
being measured/transmitted
– Output is a continuous 4-20 ma signal
„ Digital
– A signal representing a variable that is sampled
– Sampled values are a set of discrete values
– Output is a continuous 4-20 ma signal held at the last
sampled value

EDS 2006/Inst-16

An analog signal parallels the process variable being measured. The process is
continuously being measured and the transmitter signal is analogous to the process
variable.

A digital signal is a discrete sampling of the process variable updated periodically.

UOP Confidential - Do Not Copy 16

 Analog Signal
„ A signal representing a variable that is continuously
being measured and/or being transmitted.
PLC/DCS

4-20mA
(Continuous Signal)
Analog transmitter,
continuous measurement of
process variable (e.g. flow) Analog Transmitter, analog
electronics with analog sensor

EDS 2006/Inst-17

To understand the difference between analog and digital instruments, an

understanding of the difference between an analog signal and a digital signal is
required.

The definition of an analog signal is as follows: “A signal representing a variable

that is continuously measured and/or being transmitted”.

Are digital smart transmitters better than analog transmitters for response and
accuracy?
Many digital transmitters on the market, although more accurate than analog
transmitters, respond poorly compared to analog transmitters..

UOP Confidential - Do Not Copy 17

 Digital Signal
„ A signal representing a variable that is sampled. These
sampled values are a set of discrete values that are
represented by numbers.
PLC/DCS

4-20mA
(Continuous analog signal held at last
Digital (Smart) Transmitter, sampled value)
sampled process measurement
Digital (Smart) Transmitter,
digital electronics with analog
sensor

EDS 2006/Inst-18

The definition of a digital signal is as follows: “A signal representing a variable that

is sampled. These sampled values are a set of discrete values that are represented
by numbers.”.

Even though the 4 - 20 ma signal is an analog signal , the update of the 4 - 20 ma

signal is dependent on the digital sampling rate of the measured variable in the
digital transmitter.

This difference between an analog signal and a digital signal defines the differences
in design between analog and digital instruments.

Many digital transmitters on the market today update the measured variable very
slowly and thus respond worse than analog transmitters. The slow updating of the
process variable adds dead-time to the transmitter performance. This dead-time
degrades the overall transmitter performance.

UOP Confidential - Do Not Copy 18

 Smart Instrument
„ An instrument that digitally measures a variable and can
communicate with a hand held PC or communicator.

PLC/DCS

4-20mA
(Continuous analog signal held at last
Digital (Smart) Transmitter, sampled value)
sampled process measurement
Digital (Smart) Transmitter,
digital electronics with analog
sensor

EDS 2006/Inst-19

One of the advantages of the digital transmitter is that the design incorporates a
microprocessor, i. e. digital electronics.

Therefore a smart digital transmitter by definition is: “An instrument that has a
microprocessor, that can communicate with a hand-held communicator or PC, and
whose output is a 4 -20 ma signal”.

UOP Confidential - Do Not Copy 19

 Fully Digital Smart Instrument
„ An instrument that digitally measures a variable, can
communicate with a hand held communicator or PC
and who's transmitted output of the process variable is
fully digital.
PLC/DCS

Discrete digital
transmitted signal
(Fieldbus type signal:
1 0 1 1 1 1
HART, SP50 Fieldbus,
Foxboro I/A, Profibus,
Honeywell) Fully digital (Smart)
transmitter, digital electronics
with analog sensor

EDS 2006/Inst-20

A fully digital smart instrument is one that is fully digital including it’s output.
Digital instruments with digital outputs for communication are of the fieldbus type.

There are some digital instruments on the market that are both smart and fully
digital smart because the output can be selected to be 4 -20 ma or fully digital.

As with the digital (smart) transmitters updating of the process variable adds dead-
time to the transmitter performance, the digital communication of the fully digital
smart transmitters adds additional dead-time to the transmitter and performance.
Therefore the digital communication of the measured variable needs to be as fast as
possible to limit the effects of the digital dead-time.

UOP Confidential - Do Not Copy 20

 Real Time Versus Deterministic Time
„ Real time in process control:
– Continuous measurement or transmission of a process
variable or control signal.

Process
Variable Real Time Signal

Time

EDS 2006/Inst-21

Real time control or measurement refers to any instrument, control device, or

system that measures or controls continuously.
Virtually all digital measurements, control devices, or systems do not operate in real
time because the sampling rate (deterministic interval) is too large.

UOP Confidential - Do Not Copy 21

 Real Time Versus Deterministic Time

„ Digital Deterministic Time (Dt)

– Number of times per second (rate) that process variable
is measured or signal is updated
„ Rate value is dependent on the vendor
– Considered as real time because the rate value (from 1
to 20 times per sec) is a constant value?
„ Dt is real time when a digital system measures /outputs
a signal an infinite number of times per second

EDS 2006/Inst-22

Any control system or instrument, equipped with a microprocessor, has some time
delay because of the digital sampling rate. Combined with the fact that
microprocessor also cycles through its own diagnostic program, the transmitter is
not capable of transmitting the process variable continuously as does the analog
counterpart.

The digital deterministic interval should be such that the instrument or control
system is able to measure, display, and control the fastest process variations.

UOP Confidential - Do Not Copy 22

 Misused Real Time Definition
„ Would this be considered as real time?

In this example rate is five times per second

True real time Process

Process Variable Deterministic
Variable sampled process
variable
considered as
real time by
many vendors

Time

EDS 2006/Inst-23

Many vendors consider this example as real time and would consider the sampling
rate of five times per second as too fast and not required. For fast loops such as
flow, five times per second is a bare minimum rate.

UOP Confidential - Do Not Copy 23

 Misused Real Time Definition

„ Would this be considered real time?

In this example rate is once per second
Process
Variable True real time Process
Variable
Deterministic
sampled process
variable
considered as
real time by
some vendors
Time

EDS 2006/Inst-24

Many vendors would consider a sampling rate of once per second as real time.

A sampling rate of once per second is often considered as a “standard rate”.

Vendors would further argue that a faster rate is not required for any loop fast or
slow. For many control loops, once a second is a long time to not be controlling.
This is incorrect and misleading.

For control loops in fast responding processes, such as flow loops or liquid pressure
loops, degradation of loop performance is often attributed to the slow sampling rate
of the associated control system or instruments.

UOP Confidential - Do Not Copy 24

 Correct Sampling Rate

„ The sampling rate should be fast enough to see the

fastest changes in the process
– Fast control loops, such as flow, pressure or compressor
anti-surge, require a sampling rate of at least 10 times a
second
„ High rates alone will not improve loop performance on
a fast control loop
„ Controllers with fast execution rates & fast responding
control valves are also required

EDS 2006/Inst-25

Any field device should be able to measure and transmit the fastest changes in the
process.

On fast loops the following combination among the devices should exist:
a very responsive field device sensor;
the field device is sampling the sensor at a high enough rate (10 to 20 times
per second) to “see” the process changes;
the communication medium is as fast at the field device sampling rate;
the loop controller has a high execution rate (e.g. 5 to 10 times per second,
for anti-surge control execution rate can be up to 25 times per second);
the control valve is very responsive (e.g. control valve performance should
not inhibit overall loop performance).

UOP Confidential - Do Not Copy 25

 Analog Transmitter with 4-20mA Output
• Analog instruments operate in real time when little or no
filtering or signal processing is used
- As a consequence they are very responsive
Process
Analog Instrument
Variable
4-20mA output:
Input 20mA Output
Signal Measurement, Signal
Linearization,
Signal Processing and
4-20mA
Transmitted Output. 4mA
Time Time
Analog output update rate is
infinite samples per second

EDS 2006/Inst-26

The main benefit of analog instruments is that they work in real time, unlike digital
instruments that sample the process measurement.

The sampling rate for analog instruments is infinite samples per second. Some
applications, such as blower anti-surge control, are very fast control loops and the
analog instrument is well-suited for these types of applications. Digital instruments
with a slow sampling rate may not detect that the equipment has moved in and out
of surge because of the quickness of the surge phenomenon.

UOP Confidential - Do Not Copy 26

 Smart Transmitter with 4-20mA Output
Smart Transmitter Digitizing Measurement
Process with 4-20mA Output
Variable
Sampling
Input of Input
Signal Analog To Signal
Digital
Conversion
(ADC Digital
Sampling)
1 2 3 4 Time
Time Linearization,
Min. filtered Signal Processing Time
20mA and
Output Signal Digital to Analog Output of ADC
Conversion. 1 0 1 0 1 0 1 Sample 4
Output update rate 1 1 1 0 0 1 0 Sample 3
1 to 20 times per
0 0 1 0 1 0 1 Sample 2
second dependent
4mA 1 1 0 0 1 1 0 Sample 1
on vendor
Time

EDS 2006/Inst-27

Digital sampling essentially prohibits a digital instrument from measuring and

transmitting the process variable continuously. Sampling rates should be as fast (or
faster) than the process changes.

As indicated earlier the digital transmitter is equipped with an analog sensor and
digital electronics (microprocessor). The heart of the microprocessor (and thus the
digital sampling) is the analog to digital converter.

The digital sampling process produces dead-time in the process measurement. In

order to approach real time measurement, dead-time has to be minimized.
Instruments with slower sampling rates create additional dead-time. Instruments
with slow sample rates (i. e., 1 to 3 samples per second) should not be used except
on processes that vary slowly (process variations in seconds not milliseconds).

Due to the action of the digital to analog converter the 4-20mA signal has steps. A
further filtering stage is normally required before the 4-20mA signal is transmitted.

UOP Confidential - Do Not Copy 27

 Digital Filtering (Damping)

• Minimum filtering value typically 0.2 sec. Removes

noise and other unwanted high frequency signals
- Increased filtering (typically up to 32 sec) removes lower
frequency signals (often wanted process variability signals)
- Therefore increased filtering should be used with caution

Process Process
Variable Variable
Min. filtered
20mA 20mA Heavily filtered
Output Signal
User Added Output Signal
Filtering/
Damping
4mA 4mA
Time Time

EDS 2006/Inst-28

Filtering (also called damping, as higher frequencies are filtered out) is used to
“clean-up” the signal before the 4-20mA signal is transmitted.

The minimum filtering value for a differential pressure transmitter is typically 0.2
seconds. The amount of filtering can be increased (increase damping) by
increasing the damping value (e.g from 0.2 to 10 seconds). However too much
filtering/damping of the measured variable can be detrimental for control of fast
processes by filtering-out process variability. The signal is made to look “clean” at
the expense of responsiveness.

UOP Confidential - Do Not Copy 28

 Analog and Digital Comparison

Analog Digital
Accuracy 1.00% 0.10%
Sampling Rate Infinite, (continuous) Discrete (1 - 20 samples/sec)
Measures Real Time Yes No
Responsiveness Faster Slow to Fast (rate dependent)
Calibration Shop DCS or Communicator
Ambient Temp Effect No Compensation Temperature Compensated
Linearization Inferior Superior
Microprocessor No Yes
Built-in Diagnostics No Yes
Dead Time Minimal Yes (varies)

EDS 2006/Inst-29

Some benefits of analog instruments are as follows:

Analog instruments operate in real time and are therefore more responsive
than digital instruments. Analog instruments have the equivalent of
measuring infinite samples per second of the process variable being
measured. For current digital instruments the range of sampling rate is
between 1 to 20 samples per second.
Some of the disadvantages of analog instruments are as follows:
Calibration typically requires a skilled technician; and in the process
industry, calibration usually requires that the instrument be dismantled and
sent to the instrument shop. Measurement accuracy is affected by ambient
conditions and the instrument tends to drift with on-stream time. Periodic
recalibration is required to ensure measurement accuracy.
Some of the advantages of digital instruments are as follows:
The inclusion of the microprocessor allows re-ranging the transmitter with a
hand-held communicator or remotely from the DCS console. Typically a
RTD is “built-in” to allow for compensation of ambient temperature effects
improving the overall accuracy of the measured variable. Superior
linearization and transmitter characterization are some additional advantages
of the digital instruments.

UOP Confidential - Do Not Copy 29

 Transmitter Reference Accuracy

„ Combined effects of linearity, hysteresis, and

repeatability at reference conditions
„ Means of evaluating/selecting transmitter
„ Accuracy ranges from 0.075 to 0.2%
„ Does not predict real operating performance
„ Does not take into account temperature effects, line
pressure, process variations, or transmitter stability
over time

EDS 2006/Inst-30

Reference accuracy alone does not necessarily predict how the transmitter
will perform during normal operation or long term.

UOP Confidential - Do Not Copy 30

 Overall Real Performance

„ Operating Performance
– Includes reference accuracy, ambient temperature
effects, and static pressure effects
– 5-15 times greater than reference accuracy when the
appropriate errors are included
„ Evaluate performance based on
– Operating Performance
– Long-term Stability
– Dynamic Performance

EDS 2006/Inst-31

In addition to reference accuracy, the evaluation and selection of

transmitters should take into account operating performance, long term
stability, and dynamic performance.

Operating performance includes reference accuracy along with ambient

temperature and static pressure effects. The operating performance can be 5
- 15 times greater when taking into account actual operating conditions.

Long term stability and dynamic performance should also be reviewed

when selecting the transmitter. Dynamic performance takes into account
the update rate and digital communication of the transmitter. For fast acting
processes, like flow or liquid pressure control loops, dynamic performance
of the transmitter will play an important role in the overall loop
performance.

UOP Confidential - Do Not Copy 31

 Transmitter Performance
Reference Accuracy Does Not Equal Operating Performance !

Uncertainty = 0.100% Uncertainty = 1.775%

Conditions Conditions
Span = 100 inH2O Span = 100 inH2O
Temp = 72 F Not all transmitters Temp= 122 F
L.P. = 0 psi L.P. = 500 psi
Stability = 0 Yr will perform the same Stability = 1 Yr

Operating
Performance

Reference Accuracy Operating Performance

@ Reference Conditions • Reference Accuracy
• 72 Deg F • Temperature Effects
• Zero Line Pressure • Line Pressure Effects
• Out of Box/No Time In-Service • Long Term Stability
• Dynamic Performance

EDS 2006/Inst-32

Performance is typically characterized by a comparison of reference accuracy

specifications for each transmitter. Published accuracy's are generally 0.075% to
0.100% of span. However reference accuracy alone does not equate to nor predict
actual operating performance of the transmitter. In many cases the actual operating
performance is 2 - 15 times larger than the transmitter reference accuracy when the
appropriate errors are included.

Reference accuracy is the combined effects of linearity, hysteresis, and repeatability

without accounting for ambient temperature changes, line pressure, process
variations, or transmitter stability over time.

Actual operating performance includes reference accuracy, ambient temperature

effects, and static pressure effects. Long term stability and dynamic performance
should also be considered in the evaluation of transmitter performance. A basic
understanding of actual operating performance is required to adequately compare
and select the best transmitter for the application.

UOP Confidential - Do Not Copy 32

 Transmitter Performance
Worst Case Error (WCE) Comparison
% OF SPAN ERROR

4.5 Operating Conditions

Temperature Range (+/- F): 50
4
Static Pressure (psi): 200.0

3.5 Pressure Input (% of span): 75.0

3 Rosemount 3051C
Rosemount 1151E
2.5 Rosemount 1151S

1.5
Digital Characterization
1

0.5
Temp Compensation

0
25

30

40
35

50

0
0

0
0

00
35
10

30
15

20

25

40

45

50

60

70

80

10
in H2O

EDS 2006/Inst-33

Worst Case Error (WCE) as a function of Rosemount transmitter design

For the Rosemount 1151E (electronic analog), which is a purely analog transmitter,
the WCE ranges from 1.25 to almost 3%. The peaks represent a change in capsule
range.

The 1151S (smart) improved the WCE ranging from 0.3 to1%. This improvement
is due mainly to the digital characterization of the pressure sensor.
The latest transmitter, 3051C (fully digital smart), has WCE at less than 0.5%.
Improvement for this transmitter design is due to the addition of a temperature
sensor in the sensor module compensating for ambient temperature changes as well
as an improved pressure sensor (which is also digitally characterized).

UOP Confidential - Do Not Copy 33

 Negative Effect of Digital Devices

„ Digital devices introduce Dead Time (Td)

– Sampling process adds dead time to the loop
– Digital communication adds further dead time
– Dead time worsens the response of a controller
„ To improve the overall loop response, dead time
needs to be reduced – not increased

EDS 2006/Inst-34

A digital instrument measures more accurately the process variable than an analog
instrument, but most of the digital devices do not respond as well as an analog
instrument. The slower response is due to dead-time caused by the sampling
process; and for instruments with digital communications, this digital
communication adds further dead-time to the transmitter response.

For digital instruments the sampling process adds dead-time and the digital
communication adds dead-time. By replacing the analog transmitter with a digital
transmitter, dead-time has been added to the overall control loop. As a result of
adding dead-time, the control loop response will not be improved; in fact, the
response will be worse for most digital devices. The degradation of the control loop
response depends on the speed of the sampling process and the speed at which the
process variable is transmitted to the controller.

If the sampling rate is high (10 - 20 times per second and digital communication is
fast(5 - 10 times per second) for the transmitter, then the negative effects of the
digital device can be minimized. Also if the loop controller is poorly tuned or the
process changes slowly then the negative effects will not influence the overall loop
performance.

UOP Confidential - Do Not Copy 34

 Analog Versus Digital Performance
(Dead Time)

„ Dead Time (Td)

– Dead time is seen as the time lag between the process
variable being measured actually changing and when
the output of the transmitter begins to change
– Sensor signal A/D conversion, sensor processing time,
and digital to analog conversion produce the dead
time

EDS 2006/Inst-35

In many applications the replacement of an analog instrument with a digital

instrument adds dead time, but the dead time is often masked by other effects like:
Slow scanning PLC’s and DCS’s (PLC/DCS dead time due to slow scans is
larger than instrument dead time).
Poorly performing control valves due to hysteresis and deadband which in
itself produces control valve dead-time.
High process dead time (process dead time much larger than instrument dead
time)
Large process time constant (i.e your control loop dead time (Td) to process
time constant (Tc) ratio is much smaller than 1):
Td (control loop) / Tc(process) << 1

UOP Confidential - Do Not Copy 35

 Analog Versus Digital Performance
(Time Constant)

„ Time Constant (Tc)

– Measured by applying a step change to the transmitter
sensor and measuring the time it takes for the
transmitter output to reach 63.2% of the step change
– The time it takes for the transmitter output to reach
63.2% of the change in the process variable

EDS 2006/Inst-36

The time constant is representative of a first order lag system. For an induced step
change the time constant is defined as the time it takes the output to change 62.3%
of its anticipated change.

UOP Confidential - Do Not Copy 36

 Analog Versus Digital Performance

Analog Digital
Response Tc Td + Tc
Time
Settling 4(Tc) Td + 4(Tc)
Time
Update Infinite Varies
Rate

EDS 2006/Inst-37

In an analog transmitter (which has no dead time) the response time is the time
constant (Tc).
The update rate is infinite because the 4-20mA output signal in the instrument
electronics is always transmitting based on a continuous output signal from the
sensor circuit that is always measuring (i.e. no digital sampling). It’s important to
understand that response time does not represent the time to see the full process
change, but rather the time it takes to respond to 63.2% of the initial input step
change.
Settling time represents the time it takes to reach 98% of the step change. For
digital devices it also includes the instrument dead time.

UOP Confidential - Do Not Copy 37

 Analog Versus Digital Performance
(Transmitter Response)

24.0
Pressure
Released Note the stepping response of the digital devices.
20.0
Slow digital smart dP response
16.0
Better digital
12.0 smart dP response
Output in mA
8.0 Reference
Analog response
4.0
Good digital smart dP response
0.0
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Time in Sec

Range of variation in digital transmitter responses, for dP transmitters

currently on the market. Step response is based on doing a full range
step test (i.e. from 20 to 4 mA)

EDS 2006/Inst-38

The above graph shows actual tests done on dP analog and digital instruments with
4-20mA outputs.
The analog response acts as the reference because in theory a digital instrument
replacing an analog one should have at least the same response if not better. In
reality for many instruments this is not the case.
Note the stepping response of the digital instruments.

UOP Confidential - Do Not Copy 38

 Digital Devices in Closed Loop Control

„ Digital devices provide communications digitally,

accurate process measurements, diagnostics, and the
capability of measuring more than one variable in a
single device
„ Negative effects of digital device should be negligible
provided a high enough sampling rate is provided and
the measured process variable is transmitted to the
controller in a small amount of time (minimum dead
time)

EDS 2006/Inst-39

Because optimizing control loop performance is often neglected, the effects of

replacing an analog instrument with a digital one is often missed.
Digital instruments as mentioned before can provide digital communication,
diagnostics, very accurate process measurements, the capability of measuring more
than one variable; and if a high enough sampling rate is used and if the measured
process variable is transmitted to the controller in a small amount of time, then the
negative effects of the digital instrument should be negligible in many control loops.
For fast loops with slow responding digital transmitters, control loop performance is
often degraded.

UOP Confidential - Do Not Copy 39

 Summary of Digital Strengths

„ Microprocessor allows for diagnostics

„ Easy re-ranging, communication with hand held
communicators, PC, and DCS’s
„ More accurate linearization, signal processing and
compensation features
„ Digital devices have the ability to have PID controllers,
mathematical functions, logic functions, etc., built-in

EDS 2006/Inst-40

Having looked at the main weaknesses of a digital instrument when being used in
closed loop control, the strengths need to be explored.
By far the main strength of a digital transmitter is that the in-built microprocessor
allows for a whole new range of features which are just not practical with analog
instruments. Diagnostics, communication, linearization, signal processing and
compensation features are among the current advantages of many digital devices.
Some of the other features to be explored are the ability of a digital instrument to
have the features often found in PLC and DCS systems such as PID controllers,
mathematical functions, and digital functions.

UOP Confidential - Do Not Copy 40

 Future Benefits of Digital Devices
PLC/DCS with Fieldbus Device Level
and Controller Level Communication
Fisher- Fisher Valvelink
Rosemount Smart Valve
Asset Diagnostic
Management Software

Controller Level Bus Bidirectional(two way)

Device Level Bus Fieldbus Communication

Fisher

Fieldbus Smart Fieldbus Smart Control Valve with

Transmitter Transmitter Fieldbus Smart/digital
Valve Positioner

EDS 2006/Inst-41

The future is to bring process measurement, process control, instrument diagnostics,

instrument calibration, and troubleshooting of field devices to one common
platform. This is the basis of the ISA SP50 Foundation Fieldbus standard.
Being on one common platform means that all field and control devices
communicate and are configured using a common software language and common
hardware components.

UOP Confidential - Do Not Copy 41

 Introduction to SP50 Fieldbus

„ SP50 Fieldbus Standard

– Is a digital field device and control bus communication
standard as well as a functional control standard for
the measurement and control of process variables in
hazardous and non hazardous areas to achieve device
and control system inter-operability

EDS 2006/Inst-42

The SP50 Fieldbus standard was initiated by ISA. The next natural step after
developing and implementing the 4-20mA standard, was to design a new digital
communication and functional standard for field devices measuring or controlling
analog variables. This new standard would not only define the digital
communications but also define the functions the control and measurement devices
could perform. SP50 has been specifically designed for analog measurement and
control in both hazardous and non hazardous areas.

SP50 is both a digital device and control bus communication standard as well as a
functional control standard for the measurement and control of process variables in
hazardous and non hazardous areas to achieve device and control system
interoperability.

This standard is now being pursued by the Fieldbus Foundation which is

implementing the SP50 standard.

The contender to the SP50 Fieldbus standard is Profibus which has been in
existence for a few years. However many of the instrument vendors in the US are
supporting the SP50 Fieldbus standard. Only time will tell which will become the
“defacto” bus standard for process control.

UOP Confidential - Do Not Copy 42

 Fieldbus Objectives

„ Maintain best features of 4-20mA system in developing

the SP50 Fieldbus Standard
– Simple two-wire wiring practices
„ Bus powered devices
„ Intrinsic safety
„ Inter-operability of measurement and control devices
between vendors

EDS 2006/Inst-43

The SP50 ISA committee wanted to keep many of the 4-20mA features for the new
SP50 fieldbus standard.
Plant wiring will be basically as per analog loops, i.e. twisted pairs.
Bus systems can supply instrument power.
Bus systems can be IS and use safety barriers.
Bus systems allows for openness in connecting any field device, controller,
or other system that uses the SP50 fieldbus standard.

UOP Confidential - Do Not Copy 43

 Benefits of SP50 Fieldbus

„ More and better information for predictive maintenance,

plant safety, product quality, and regulatory compliance
„ Multi-drop wiring, for lower installation costs
„ Support of new intelligent instrument functions and
migration of control to field devices for better operating
performance

EDS 2006/Inst-44

The anticipated benefits of the SP50 fieldbus standard over the 4-20mA standard are
shown above.
To get the full benefits of the SP50 fieldbus standard more control functions need to
be at the field level. The control system should not be structured like a DCS where
all control action, trending, alarming, reporting are done in the DCS system.
Fieldbus is not a direct replacement for a DCS, it is a radically different “control
system concept” that allows for large amounts of information to pass between all
parts of the process (controllers and field devices), analyze and collect data in a
process control system, and allow for most of the plant control to be done at the
field level.
The field devices will measure variables, set the process alarm points, be able to do
diagnostics and set control philosophies. The host systems will collect data, display
them on MMI’s, operator displays, produce reports, and optimize on overall plant
control schemes.
The bus system will reduce wiring costs which is really a small benefit of fieldbus.

UOP Confidential - Do Not Copy 44

 Feedback Control Loop

„ Signal Transmission and Transmitters

– Analog – Pneumatic/Electronic
– Electronic Analog/Digital
– Transmitter Performance
– Fieldbus
„ Loop Components
– Process
– Measuring Means
• Temperature, Flow, Pressure, Level, Analysis

EDS 2006/Inst-45

UOP Confidential - Do Not Copy 45

 Feedback Control Loop
(Flyball Governor)

Flyball
Weight

Engine
Steam

EDS 2006/Inst-46

The first significant application of closed loop, automatic feedback control was the
use of a fly-ball governor on a steam engine in the late 1700’s. In this application,
the fly-ball governor uses the centrifugal force generated by rotating fly-ball
weights, along with a system of linkages and levers, to operate a valve that controls
the steam flow to the engine.

On start-up the system is at rest and the steam valve would be in the wide-open
position. The operator would introduce steam to the system by manually opening
an isolation valve (not shown) on the steam supply. The steam provides work and
the engine speed would be sensed by the fly-ball governor through the series of
linkages. The fly-ball weights, due to the centrifugal force, would move outward
and upward from the center of rotation. This in turn pulls upward on the valve stem
closing the steam valve, thus limiting the amount of steam supplied to the engine.
The speed set point of the engine could be adjusted manually by adjusting the
length of the linkage connected to the valve stem.

UOP Confidential - Do Not Copy 46

 Feedback Control Loop
(Block Diagram)
Controller

Set
+ Error Signal Function
Point Generator
-

Process
Output
Variable
Measuring Final Control
Means Element
Measured Manipulated
Variable Variable

Process

Load
Variable

EDS 2006/Inst-47

Mechanical feedback control was applied to the process industry at the beginning of
the 20th century. Today the most widely used automatic feedback control is the
basic 3-mode PID controller. Even though other more sophisticated control
techniques are available today (especially with distributed control systems),
feedback control accounts for more applications than any other.

The block diagram illustrates the various components that make-up the basic
feedback control loop. Any feedback loop will consist of a Process (flow, pressure,
temperature, etc), Measuring Means to measure the Process Variable, and some
form of Comparator to compare the value of the Process Variable with the desired
Set Point.

The resulting difference, or Error Signal, is sent to a Function Generator (PID

mathematical algorithm) which operates a Final Control Element (typically a
control valve). The Final Control Element adjusts a Manipulated Variable (flow,
pressure, speed, etc) as necessary to minimize the Error Signal, and bring the
Process Variable closer to the Set Point. How quick this happens is contingent
upon the process itself and the P, I, and D constants (tuning parameters) selected for
the Function Generator (mathematical algorithm).

UOP Confidential - Do Not Copy 47

 Feedback Control Loop
(Examples)

Example Number {1} {2} {3} {4}

Process Bathtub Toilet Tank Shower Water Heater
Type Batch Batch Continuous Continuous
Process Variable Water Level Water Level Water Temperature Water Temperature
Measuring Means Eye Float Touch Thermostat
Set Point Adjustment Hand Linkage Hand Thermostat
Comparator Brain Linkage Brain Thermostat
Function Generator Brain Linkage Brain Thermostat
Final Control Element Water Valve Water Valve Water Valve Gas Valve
Manipulated Variable Water Flow Water Flow Water Flow Gas Flow
Load Variable (See Note) (See Note) Water Pressure Water Flow
Note: Normally no load variable; however, load variable effect could be created by water leakage or variation in water
pressure. A Load Variable is any variable, other than the Set Point, that has an affect on the Process Variable.

EDS 2006/Inst-48

These are just a few examples of feedback control in our daily lives and some
feedback loops contain a human component in the loop. Example 1 illustrates a
person adjusting a water valve to fill a bathtub. The person filling the tub watches
the water level and shuts off the water flow as the level approaches the desired level
in the tub. The level information is fed back visually to the operator, who
eventually takes action as the level approaches the desired level.

Likewise in example 2 when a float ball in a toilet tank moves a valve to fill the
tank to the specified level, feedback control is involved.

Examples 3 and 4 also illustrate feedback control. The third example involves a
human component (similar to example 1), where a person taking a shower will
adjust the flow of hot/cold water to control the temperature of the water. The last
example illustrates the use of a thermostat to operate a gas valve to control the
temperature of a hot water heater.

UOP Confidential - Do Not Copy 48

 Feedback Control Loop Examples
(Comparisons)

„ Bathtub/Shower examples by definition

– manual and open loop control
„ Toilet Tank/Water Heater examples
– automatic and closed loop control
„ First 2 are batch – Last 2 are continuous
„ All are processes having measuring means,
comparator, set point, function generator, and final
control element
„ All examples are subject to load variables

EDS 2006/Inst-49

These examples not only exhibit some similarities, they also exhibit some basic
differences which helps define various types of feedback control loops. Filling of
the bathtub and adjusting shower temperature are by definition manual open loop
feedback control. In both instances operator intervention is required (manual
control). Likewise open loop implies that no mechanism is available to
automatically adjust. In other words, the operator must be present to prevent mis-
operation.
Filling of the toilet tank and temperature control on the hot water heater are by
definition automatic closed loop feedback control. Some mechanical device is
utilized to measure and provide feedback (other than the operator) to manipulate the
process automatically. As long as the loop components are operational, the operator
need not be present at all times. Reliability then becomes an issue.
Another difference is whether the process is batch or continuous. The bathtub and
toilet tank are examples of batch processes, while the other two examples are
examples of continuous processes. The refining and petrochemical industries are
primarily continuous processes.
Also all cases are subject to disturbances in the Process, (these disturbances are
often called Load Variables), and require corrective action by the control system.
Changing water supply pressure is an example of a load variable.

UOP Confidential - Do Not Copy 49

 Process

„ Even though processes are different, the principles of

the feedback control loop are applied to all
„ The laws of physics and dynamics govern all and each
is represented by the same mathematical equations,
differing only by the coefficients used in the equations
„ Many common processes are computer simulated and
dynamically modeled

EDS 2006/Inst-50

Although these four processes are different, they all have feedback loops and the
same principles can be applied to all of these processes. In fact the process can be
almost anything and each will follow the same laws of physics and dynamics.
Therefore each process can be described by the same mathematical equations
differing only in the coefficients used in the equations. We will investigate some of
the simpler first order processes and use some practical approaches to simply the
procedures to model the process and determine optimal tuning parameters for
controlling the process.

UOP Confidential - Do Not Copy 50

 Process
(cont’d)

„ Each process has two external inputs to the feedback

control loop
– Set Point - servomechanism response
– Load Variable- regulator response
„ Load variables play a major role in process control
causing transients from steady state operation
(dynamics)
„ Process and Manipulated Variable may differ leading
to more difficult control

EDS 2006/Inst-51

Examining the previous examples of the feedback control loops, continuous

processes (examples 3 and 4) have two external inputs to the feedback control loop.
Both the Set Point and Load Variables are External Inputs and the control loop must
respond to a change in either of these inputs. Any process can have more than one
Load Variable. A Load Variable is any variable, other than the Set Point, that has
an affect on the Process Variable.
In Example 3 a change in hot water pressure upstream of the valve will cause a
larger differential pressure across the valve. The net effect is an increase in
temperature as a result of more hot water flowing through the valve. This increase
in flow is a direct result of an increase in differential pressure.
For real continuous processes true steady state seldom exists. Load Variable
changes and upsets in the process must be countered and corrected and the
dynamics of the system must be considered. However the majority of the processes
are approximated by the same laws of physics and dynamics, and the instrument
engineer is concerned with “tuning” the loop to match the dynamics of the process.
A third point worth noting about the feedback control loop is that the Process
Variable and the Manipulated Variable may not be the same. See Example 4. The
more indirect the relationship between MV and PV is, the more difficult the
controller’s task may be.

UOP Confidential - Do Not Copy 51

 Process Measurement

„ To control must measure Process Variable

– Primary element converts intrinsic property into a
measurable signal
– Transmitter converts this signal into standard
instrument signal
„ Transmission of standard instrument signals
– Pneumatic, 3 - 15 psig (0.2 - 1.0 kg/cm2(g))
– Electronic, 4 - 20 mA (milli-amperes)
– Digital protocol, proprietary/fieldbus

EDS 2006/Inst-52

A multitude of devices can used to measure Process Variables and transmit the
signal to the control room. In general a primary element is located in the process
unit and converts an intrinsic property of the process into a measurable signal. An
associated transmitter “reads” this measurable signal and converts the signal from
the primary element into a standard instrument signal. Standard instrument signals
used in most applications today are the 4 - 20 ma electronic signal and the 3 - 15
psig pneumatic signal.
For an orifice plate, a mathematical relationship exists between flow through the
orifice and the differential pressure generated across the orifice plate as a function
of flow rate. Once the orifice plate and transmitter are designed and calibrated,
variation in flow across the orifice plate is transmitted to the control room where 4 -
20 ma represents 0 to 100% of design. As long as the flow is less than what the
system has been designed for, the Process Variable is measurable.
Typically the primary element is located in the field; and depending upon the
equipment used, the transmitter may or may not be located in the field.
Temperature transmitters are milli-volt transmitters and are frequently located in the
control room.
In the Process Industry, several main types of Process Variables exist (flow,
pressure, temperature, level, analytical) along with several varieties of primary
elements used to measure these variables.

UOP Confidential - Do Not Copy 52

 Thermocouple
Temperature Devices
Thermocouple
Head Lead Wire
A A + Cu
Millivolt
Transmitter
B B - Cu

Hot Junction (T1) Cold Junction (T2)

ISA Type A(+) B(-)

E Chromel Constantan
J Iron Constantan
K Chromel Alumel
T Copper Constantan

An EMF is developed when opposite junctions of certain dissimilar pairs of

metal are exposed to different temperatures. EMF is proportional to T1 - T2

EDS 2006/Inst-53

The primary element used most often for temperature measurement is the
thermocouple (T/C). T/C operation is based on the Seebeck effect, in which an
electromotive force (EMF) develops when opposite junctions of certain dissimilar
pairs of metals are exposed to different temperatures. This EMF varies as the
temperature difference between the two junctions varies. The magnitude of the
EMF depends upon the material used; however, for a given material combination
(T/C type), the relationship between EMF and temperature difference is predictable.
The cold junction is typically located in the control room; therefore T/C extension
wire is installed from the field termination to the control room termination, and the
circuitry must include a mechanism for cold junction temperature compensation.
Once the temperature difference is determined as a function of the EMF generated
and the cold junction temperature is compensated for, the hot junction temperature
is known.
UOP will typically specify type E for most process applications and will specify
type K for high temperature applications ( fired heaters). Type J was frequently
specified in the 60 - 80’s, but because of iron oxidizing, UOP now specifies type E.
Type T is often used for cold temperature measurements.
Chromel, Alumel, and Constantan are proprietary alloys, but are widely available.
The thermocouple is a relatively simple inexpensive, rugged and reliable device.

UOP Confidential - Do Not Copy 53

 Thermocouple EMF

E K
40 J
30

20

10 T EMF dependent upon material used

but EMF is predictable for a given
0 difference in temperature.

-10
0 400 800 1200 1600 2000
Temperature Difference °F

EDS 2006/Inst-54

As stated earlier UOP now specifies type E for most process applications because of
the chromel-constantan material is not subject to oxidation as is the type J. Another
advantage of type E is the increased sensitivity as shown above.
UOP will typically limit type E to temperatures up to 1200 °F (650 °C). For higher
temperatures type K will typically be specified not to exceed 2000 °F (1100 °C).
Although it is difficult to see with this graph, the thermocouple EMF non-linear
with respect to the temperature difference. However this tends not to be a set-back
for their use in most applications. This can be compensated for by signal
characterization in the overall circuitry or by providing a temperature transmitter
with a smaller temperature span to improve the measurement reliability.

Other materials of construction for thermocouples are available, but most of UOP’s
processes are covered by types E, J, and K.

UOP Confidential - Do Not Copy 54

 Limits of Error – Thermocouples

Limits of Error
Calibration Thermocouple Type Temperature Range Standard Special
J Iron/Constantan 32 °F to 530 °F ±4 °F ±2 °F
530 °F to 1400 °F ±0.75% ±0.4%
K Chromel/Alumel 32 °F to 530 °F ±4 °F ±2 °F
530 °F to 2300 °F ±0.75% ±0.4%
T Cppper/Constantan [-328 °F to 32 °F ±2 °F or ±1.5% [----]
32 °F to 260 °F ±4 °F ±2 °F
260 °F to 700 °F ±0.75% ±0.4%
E Chromel/Constantan [-328 °F to 32 °F ±3 °F or ±1.0% [----]
32 °F to 600 °F ±3 °F ±2 °F
600 °F to 1600 °F ±0.5% ±0.4%
R Platinum 13% Rhodium/ 32 °F to 1100 °F ±2.5 °F ±1 °F
Platinum 1100 °F to 2700 °F ±0.25% ±0.1%
S Platinum 10% Rhodium/ 32 °F to 1100 °F ±2.5 °F ±1 °F
Platinum 1100 °F to 2700 °F ±0.25% ±0.1%

EDS 2006/Inst-55

One of the disadvantages of the thermocouple is its accuracy of measurement.

Thermocouple “Limits of Error” summarizes the accuracy of the measurement. For
a type E thermocouple that may be used to measure the Platforming Reactor inlet
temperature (say 1000 °F), the limit of error is ±0.5% or in this example ±5 °F.
Therefore the actual temperature lies between 995 and 1005 °F. This is not a major
concern because the measurement is repeatable; and for the Process, the operator is
not concerned about the absolute temperature measurement. His concern will
center around the target Octane number of the Platforming product. If the real
temperature is 996 °F and the Octane number of the product is 99 (with a Target of
100), the operator will raise the inlet temperature approximately 4 °F per difference
in Octane number required. In other words the operator will increase the reactor
inlet temperature set point from 1000 °F to 1004 °F. As long as the thermocouple
measurement is repeatable the absolute temperature measurement is not a necessity
in many applications.

UOP Confidential - Do Not Copy 55

 Limits of Error – Thermocouple Wire

Limits of Error
ANSI Type Thermocouple Type Temperature Range Standard Special
J Iron/Constantan 32 °F to 400 °F ±4 °F ±2 °F
K Chromel/Alumel 32 °F to 400 °F ±4 °F ±2 °F
T Cppper/Constantan 32 °F to 260 °F ±2 °F ±1 °F
E Chromel/Constantan 32 °F to 400 °F ±3 °F ±2 °F
R Platinum 13% Rhodium/ 32 °F to 400 °F ±2.5 °F ±1 °F
Platinum
S Platinum 10% Rhodium/ 32 °F to 400 °F ±2.5 °F ±1 °F
Platinum

EDS 2006/Inst-56

Limits of Error also exist for thermocouple extension wire. For the example of the
Platforming reactor inlet temperature, the limit of error for the type E thermocouple
was ±5 °F. The limit of error for type E thermocouple extension wire is ± 3 °F.
Therefore the temperature measurement lies between 1000 ± 8 °F if thermocouple
extension wire is installed to the control room.
If thermocouple extension wire is not installed (a pair of copy wires can be used),
then the cold junction is moved to the field. The transmitter may also be located in
the field, but the cold junction compensation circuitry is more complex due to the
wider range in ambient temperature. Accessibility also becomes an issue, especially
for high temperature applications.

UOP Confidential - Do Not Copy 56

 RTD Resistance
4.0
Nickel

3.0

Platinum
2.0

1.0
32 250 500 750 1000 1250
Temperature F

EDS 2006/Inst-57

A second type of primary element commonly used for temperature measurement is

the Resistance Temperature Detector (RTD). The RTD senses heat based on the
principle that a change in temperature results in a proportional change in the
electrical resistance of the wire. The most widely used material is Platinum, and is
essentially a long thin wire wound into a small coil.
The Platinum RTD has a stable, accurately known linear relationship between
resistance and temperature as shown in the graph and is an extremely accurate
temperature measurement device. The typical design for a RTD has a known
resistance of 100 ohms at 32 °F (0 °C).
Once the resistance of the RTD is measured, the actual temperature is known,
unlike the T/C which actually measures a temperature difference along with a cold
junction temperature.
Nickel RTD’s are also available, but are not widely used throughout industry.

UOP Confidential - Do Not Copy 57

 Resistance Temperature Detector
Wheatstone Bridge Circuit
When Bridge is balanced
(current is zero): R R1
REF

RREF/RMEAS = R1/R2

EMF

R R2
MEAS

Lead wires add additional resistance

to the process measurement leg of the
Wheatstone Bridge. Error increases IST-R00-(152)
with distance and is significant.

EDS 2006/Inst-58

A Wheatstone bridge circuit with a variable reference resistance (Rref) is employed

to accurately measure the resistance. As the temperature being measured by the
RTD changes, the resistance (Rmeas) changes. If the bridge is unbalanced, then
current will flow across the bridge. The Wheatstone bridge circuitry will adjust Rref
in the proper direction to drive the current to zero balancing the bridge.
The principle behind the Wheatstone bridge is such that the ratio of the two known
resistances is equal to the ratio of Rref / Rmeas when the bridge is in balance. Once the
bridge is balanced and Rref is determined, Rmeas can be determined.
However typical installations with lead wires ranging from tens of feet to hundreds
of feet introduce a significant variable error in the temperature measurement.
Because the circuit is quite sensitive to resistance, the resistance of the copper lead
wires must be taken into account. Both the length of the lead wires and the
temperature of the lead wires will introduce error in the temperature measurement.

UOP Confidential - Do Not Copy 58

 Resistance Temperature Detector
Three-Wire Circuit

R R1
REF

EMF
R
MEAS
R
LEAD R2

Adding lead wires for both the bridge

and reference legs balances the bridge
equally, eliminating the error. IST-R00-(153)

EDS 2006/Inst-59

Three-wire RTD’s counterbalance the effects on the measurement and reference

legs of the Wheatstone bride. In essence the bridge on both the measurement leg
and the reference leg is extended to the field. The length, resistance, and
temperature of the leads are essentially equal and make the error negligible.
The three-wire copper wire system offsets the cost of the two-wire thermocouple
extension wire system.

UOP Confidential - Do Not Copy 59

 Resistance Temperature Detector
Differential Temperature Circuit

R R1
REF

EMF

R R2
MEAS

IST-R00-(154)

EDS 2006/Inst-60

Temperature devices can also be modified to meet special or specific requirements.

In some distillation control applications a differential temperature across a fixed
number of column trays is employed to compensate for changes in column pressure.
Binary distillation is a primary example. (This control scheme will be reviewed in
greater detail later on in the session.)

Utilizing two-two wire RTD’s arranged in such a manner as shown, the reference
resistance RREF and the measured resistance RMEAS yield the temperature difference
across the trays as measured by the Wheatstone Bridge.

Similar set-up can be employed with Thermocouples. Cold junction compensation

is not required in this application. The “reference” junction becomes in essence the
cold junction and the end result is temperature difference.

UOP Confidential - Do Not Copy 60

 Thermocouple and RTD Comparison

Theromcouple RTD
Accuracy Limits of Error wider Limits of Error narrower
Ruggedness Excellent, will not affect life of probe Sensitive to strain, vibration, shock
Temperature Range [-328 °F to 4200 °F [-50 °F to 1100 °F
Size >0.010" sheath diameter >0.062" sheath diameter
Drift Periodic calibration required 0.02 to 0.2 °F per year
Resolution mv/degree, lower signal to noise ohms/degree, higher signal to noise
Cold Junction Required not Required
Lead Wire Mandatory match 3-lead copper wire (typical)
Response millisecond response time > 1 second response time
Sensitivity Tip sensitive Thermal mass prevents tip sensitivity
Linearity Non-linear Linear
Cost Lower Higher

EDS 2006/Inst-61

As a comparison between thermocouples and RTD’s, each has its pluses and
minuses.
The thermocouple is much more rugged, faster responding to temperature transients,
tip sensitive, and covers a much broader temperature range. Cost wise the
thermocouple is less expensive, but cost are offsetting somewhat with the overall
installation costs associated with the thermocouple extension wire.
On the other hand the thermocouple limits of error are wider, requires periodic
maintenance (calibration check), requires cold junction temperature compensation,
and has a non-linear signal (characterization of the signal is required).
The RTD is much more accurate in absolute temperature measurement, has a linear
signal with respect to temperature, and is a direct measurement of the temperature
(cold junction compensation not required).
However the RTD is sensitive to shock, vibration, strain, etc, is limited to
temperature less than 1100 °F, has a slower response to temperature transients, and
requires a three-wire (minimum) copper system offsetting its overall installation
cost advantage.
UOP typically specs Thermocouples and will spec RTD’s only where an accurate
temperature measurement is required.

UOP Confidential - Do Not Copy 61

 Orifice Plates
Flow Devices
Design
Information
Upstream Bore
Side Beta = Pipe ID
0.1 < Beta < 0.7

Bore

Bore

Bore and Bevel Eccentric

IST-R00-(156A)

EDS 2006/Inst-62

One of the more common process measurements encountered is the “flow rate”.
The industrial work horse for flow measurement, accounting for over 50% of the
total flow devices, is the orifice plate and pressure differential transmitter.
The square edge, beveled, concentric orifice plate is used to measure either liquids
or gases. Variations of this type of plate are known in industry to allow for certain
fluid constraints. Weep holes are often drilled above or below the bore to allow for
entrained liquids or gases to escape.
As shown above the eccentric plate is often used by UOP to measure %
vaporization at a fired heater outlet. The eccentricity of the plate prevents a build-
up of liquid upstream of the plate. Typical heater design has 40 to 60%
vaporization at the heater outlet, the plate effectively is a measure of heat input in
terms of vaporization (This application will be reviewed in more detail later on in
the session.)
The orifice plate handle is often stamped with pertinent design information on the
upstream side of the plate and is characterized by the “BETA RATIO”. By
definition: Bore Diameter/Pipe Inside Diameter is equal to BETA. An abundance
of correlated data sets the beta limits between 0.1 and 0.7.

UOP Confidential - Do Not Copy 62

 Orifice Pressure Profile
HP LP
Bernoulli’s Theorem applies
law of conservation of energy
to the flow of fluids in a pipe.
Flow D
Total energy is equal to the
sum of elevation head,
velocity head, and pressure
1” 1” head:
2-1/2D 8D Total Head = H
Perm anent Loss Velocity Head = V2/2g
hw Pressure Head = 144P/Rho
Elevation Head = Z
H = V2/2g + 144P/Rho + Z
PT 1 FT 1 FT 2 PT 2
Vena C ontracta
O rifice
IST-R00-(156)
Plate

EDS 2006/Inst-63

Flow through an orifice plate is based on Bernoulli’s principle as it applies the law
of conservation of energy to the flow of fluid in a pipe. The total energy at any
particular point in the system is equal to the sum of the velocity head plus the
pressure head plus the elevation head:
H = V2/2g + 144P/Rho + Z
The Total Head of the fluid up-steam of the orifice plate is equal to the Total Head
of the fluid as the fluid passes through the orifice plate; and for a horizontal meter
run, the elevation head (Z) cancel. The cross-sectional area of the orifice plate is
smaller than the cross-sectional area of the pipe, resulting in an increase in fluid
velocity through the orifice plate (velocity = volumetric flow divided by cross-
sectional area). The law of conservation of energy therefore indicates that the
pressure head decreases as the fluid passes through the orifice plate. Down stream
of the orifice plate the cross-sectional area expands once again and the fluid velocity
decreases while the pressure head increases. Approximately 8 pipe diameters down
steam of the orifice plate, all possible pressure recovery has taken place. The
difference between the pressure upstream at this point and the pressure up-stream of
the orifice plate is the permanent pressure loss or the energy absorbed by the
system.
Flange taps (FT) or pressure taps (PT) can be used to measure a dP and correlated to
flow. The industrial application has standardized on orifice flange taps 1 inch up-
stream and 1 inch down-steam of the orifice plate.

UOP Confidential - Do Not Copy 63

 Transmitter Static Head Effects

HP LP

Flow P1 P2

H
RHO1 RHO2

Differential P3 P4 dP = (P3-P4)
Pressure = (P1+H x RHO1)-(P2+H x RHO2)
Transmitter = (P1-P2)+H(RHO1-RHO2)
(where RHO1 and RHO2 are fluid densities)

EDS 2006/Inst-64

For the typical liquid flow measurement installation the transmitter not only
measures the orifice plate differential, but also measures the static head in the
transmitter impulse lines. If the fluid densities are different (different composition
or different temperature), then the differential pressure at the orifice plate will not
be equal to the differential pressure at the transmitter. The resultant error increases
with an increasing difference in densities and/or an increasing difference in
elevation between the orifice plate and transmitter.

As a percent of flow this error decreases as the orifice differential increases.

Therefore the higher the differential the less error in flow measurement. However
an upper limit on orifice differential is governed by the economics of pumping the
fluid. The industrial standard for most applications designs the orifice differential at
100 inches water column at the meter maximum. The meter maximum generally is
chosen so the normal flow is between 70 and 80% of the meter maximum.
A square root relationship exists between flow and pressure differential. If the flow
was at 70% of meter maximum, the measured differential pressure across the orifice
plate would be 0.7 squared or 0.49 times 100 inches, which is 49 inches. This
equates to about 50% of the transmitter calibrated span. Keep this fact in mind as
we will see most instrumentation specified at 50% of its calibrated span or 50% of
its capacity as in the case of control valves.

UOP Confidential - Do Not Copy 64

 Permanent Pressure Loss
100

90

80

O rific e w ith
F la n g e T a p s

70

P re s sure Los s in % of Actua l Diffe re ntia l

60

50

40

30

S h o rt C o n e Ve n tu ri
20

Lo n g C o n e Ve n tu ri

10

0
0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 0 .8
D ia m e te r R a tio , ß

EDS 2006/Inst-65

As we discussed earlier the pressure differential at the orifice plate taps and the
pressure differential at the pipe taps (2-1/2 pipe diameters up-stream and 8 pipe
diameters down stream) can be correlated as a function of beta ratio and also in
terms of permanent pressure loss.
For the typical orifice flange installation, the above graph illustrates permanent
pressure loss in % of actual pressure differential measured at the orifice flanges (y-
axis) as a function of orifice plate Beta Ratio (X-axis).
As for our typical meter installation, if the beta ratio is 0.70, the meter maximum is
designed for a 100 inches water column differential at maximum flow, and the flow
is at meter maximum; then the permanent pressure loss is 52% of the measured
pressure differential. For this case at a flow rate equal to the meter maximum the
permanent pressure differential is 0.52 times 100 inches water column or 52 inches
water column. In terms of pressure units:

(Inches H20)(fluid density)(ft2/144 inches2)(ft/12 inches) = pounds per inch2

(52 inches)(62.4 pounds/ft3)(ft2/144 inches2)(ft/12 inches) = 1.88 psi

With Beta ratios ranging from 0.1 to 0.7, the graph illustrates that permanent
pressure loss will range from about 50% of the measured differential up to about
97%. The graph also illustrates that the smaller the beta (higher resistance for a
given flow rate) the greater the energy consumption will be.

UOP Confidential - Do Not Copy 65

 Orifice Minimum Straight
Run Requirements
Orifice Plate
39D
31D Valve
19D
14D 5D

Flow
Globe, Two Less
Control, Plane Than 10 D or
or Check Flow* 10 D Greater
Valve

Gate or Concentric Concentric

Cock Reducer Reducer
Tee,
Cross, or
Lateral
*Where Distance Between Elbows is 10 D or Less
Note:
Dimensions are for Flange Tap Orifice Surfaces Only, Based on AGA-ASME Report No. 3

IST-R00-(159)
EDS 2006/Inst-66

As stated previously, an orifice plate is characterized by its beta ratio and this ration
will range from 0.1 to 0.7. Insufficient experimental data exist outside of this range,
and the accuracy of the correlation can not be confirmed. Following UOP’s
engineering practice for line sizing, most installations will not require Beta ratios
greater than 0.7. However in some cases where the line sizing is sized “tight”
(approaches the upper limit of pipe pressure drop per unit length), the meter run
may require swaging to a larger pipe size in order to keep the beta ratio below 0.7.
In addition to the beta ratio requirements, a certain amount of straight run pipe up-
stream and down stream of the the orifice plate is required to insure that a fully
developed turbulent flow profile exists at the orifice. This allows for application of
the flow correlation developed from the experimental installations.
Also the amount of straight run pipe is a function of the beta ratio. The larger the
bore for a given pipe size (increasing beta ratio), the greater the upstream pipe
requirement.
The information provided above is UOP’s recommended practice and is based on a
beta ratio of 0.7. This is a conservative approach.

UOP Confidential - Do Not Copy 66

 Basis for Orifice Minimum Straight
Run Requirements

„ Upstream pipe diameter varies with Beta

„ Minimum requirements based on Beta = 0.7
„ Installations with shorter upstream pipe runs with
smaller betas are not incorrect
„ Being conservative allows for replacement of existing
orifice plates with larger betas without the need to
modify piping meter runs in the future

EDS 2006/Inst-67

UOP selected the upstream pipe requirements based on a beta ratio of 0.7. If the
beta ratio is say 0.45, the correlated experimental data would require less straight
run requirements than recommended in the previous slide.

UOP has chosen the data for 0.7 to be on the conservative side. For future
debottlenecking and expansion projects, where the orifice bore may need to be
increased to meet the process requirements of the expansion, changes to meter run
piping will be held to a minimum; thus minimizing the cost associated with the
expansion project.

UOP Confidential - Do Not Copy 67

 Venturi Tube

„ Alternative to the orifice plate

„ Venturi’s taper allows for stream lining flow into/out
of throat area
„ Eliminates turbulence and boundary layer separation
as in an orifice plate
„ Pressure/Velocity heads nearly reversible (permanent
pressure loss is minimized)
„ Capital costs are high, but energy savings may be
significant (centrifugal compressor)

EDS 2006/Inst-68

A second type of primary flow element is the venturi meter. The venturi is tapered
both at the inlet and outlet and is designed to guide the flow streamlines in toward
the throat area (minimum cross-sectional area), and to expand them back out to a
fully developed velocity profile without developing the turbulence and boundary
layer separation that occurs with an orifice plate. Like the orifice plate the venturi
is also characterized by a beta ratio, with the limits of 0.4 to 0.7. (Beta ratio is
defined as the Throat Diameter/Pipe ID.)

Due to the tapered design the static and kinetic energy exchange (pressure and
velocity heads) is more reversible and the overall permanent pressure loss is held to
a minimum. On the other hand the cost of a venturi meter is much higher relative to
the orifice plate.

Venturi meters are often designed in centrifugal compressor circuits where line
sizes are on the order of 10 inches or larger. For compressor circuits less pressure
drop across the primary flow element translates into an overall reduction in
horsepower requirements to operate the compressor. Payouts for the increase in
capital cost between the venturi and orifice through a reduction in operating costs
are exceptional fast.

UOP Confidential - Do Not Copy 68

 Permanent Pressure Loss
100

90

80

O rific e w ith
F la n g e T a p s

70

P re s s ure Los s in % of Actua l Diffe re ntia l

60

50

40

30

S h o rt C o n e Ve n tu ri
20

Lo n g C o n e Ve n tu ri

10

0
0 .2 0 .3 0 .4 0 .5 0 .6 0 .7 0 .8
D ia m e te r R a tio , ß

EDS 2006/Inst-69

Reviewing the plot of Permanent Pressure Loss in % of Actual Differential vs. Beta
Ratio for a venturi, the permanent pressure loss is less than 20% over the
recommended beta ratio range of 0.4 to 0.7 and less than 30% with beta ratios down
as low as 0.2.

Herein lies the advantage of the venturi in terms of energy savings in the operation
of centrifugal compressors.

The disadvantage of the venturi is its overall capital cost. Also on expansion
projects, an orifice plate can be re-bored to a larger diameter or replaced
inexpensively with another orifice plate. However the throat diameter of the venturi
can not be altered without also making changes to the inlet and outlet cones and
overall length.

UOP Confidential - Do Not Copy 69

 Dimensions of Classical
Venturi Tube

HP
d/2
LP
Flow
19-23° d 5-15° D
See Note 1
d
See Note 2

Notes:
1. Round to 0 to 1.375 D Radius
2. Round to 3.5 to 3.75 d Radius
3. LP Tap Radius <=d/6
4. For Herscel Venturi, Exit Cone Angle is 7-8°

EDS 2006/Inst-70

For the classical venturi tube, the above depicts the typical dimensions. Atypical
design could put the beta close to 0.7 in a 20 inch pipe. This implies that the throat
diameter is approximately 14 inches. Base on these requirements and the above
dimensional requirements for the classical venturi tube, the over length is on the
order of 6.5 ft (approximately 2 meters in length).

UOP Confidential - Do Not Copy 70

 Square Root Relationship
Head vs. Flow
Flow is proportional to the 100 10 100=10
square root of the differential 90
pressure measured across the
primary element. 80 9

Accuracy of the overall 70

measurement usually expressed 8
60
in terms of ‘% of full scale’. Design
50 7 50=7.07
Therefore a much larger error
exists in flow at reduced rates 40
6
than at the design rate.
30
5 25=5
Turn down usual limited to 4:1 20
4
10 3 Minimum
2 Turndown
0 1

EDS 2006/Inst-71

Flow is proportional to the square root of the differential pressure measured across
the primary element. The above square root chart illustrates the measured
differential in % of maximum differential with respect to flow rate from 1 to 10
units of flow (10 units of flow representing meter maximum).

Meter accuracies are generally ± 2% of full scale. If the meter maximum was 100
gpm and the normal flow was 75 gpm, the actual flow rate is 75 ± 2 gpm (which is
75 gpm ± 2.67%). If turndown is at 50% of normal, the actual flow rate is 37.5 gpm
± 2 gpm (which is 37.5 ± 5.33%). The bracketed accuracy of the measurement is
getting much larger at reduced flows. UOP will typically limit the design of
differential head type devices to a 4 to 1 turndown. Modifications to the design will
be incorporated if the range of flows exceed 4 to 1 turndown. Various
modifications can be employed, depending upon the design criteria. These
modifications include re-spanning transmitters, installation of two transmitters (1
high span and 1 at a low span), and even the installation of parallel flow meter runs
for extreme cases.

UOP Confidential - Do Not Copy 71

 Additional Flow Meters

„ Turbine flow meter

– Consists of straightening vanes, rotor with small
imbedded magnet, and bearings
– speed of rotor proportional to fluid velocity
„ Electromagnetic flow meter
– Fluid flows through a magnetic field and fluid must
be conductive (hydrocarbons are not)
– Voltage produced proportional to fluid velocity

EDS 2006/Inst-72

A multitude of other technologies are available for flow measurement and each has
its unique place in the industry. Various laws of physics and electrical laws are
used to correlate flow to some intrinsic property.

The turbine meter is an in-line flow measuring device that has reported accuracies
of ± 0.25%. These are often specified and used on raw material and product steams
for inventory control. These meters are costly compared to orifice plate
installations, have a significant increase in pressure drop (potential increase in
operating costs for pumped systems), and are potentially high maintenance items
because of the rotating parts.

The “mag” meter on the other hand has no moving parts and essentially “zero”
pressure drop because of no internal obstruction within the body. However for this
meter to be able to measure flow the fluid must be conductive. Gases, steam, and
most hydrocarbon liquids are non-conductive and do not produce any induced
voltage when flowing through the meter’s magnetic field. Like the turbine meter,
the “mag” meter is costly when compared to the orifice plate installation.

UOP Confidential - Do Not Copy 72

 Additional Flow Meters
(continued)

„ Vortex flow meter

– Consists of a triangular ‘bluff body’ spanning the inside
diameter of the meter
– Vortices are shed in a regular oscillating pattern
– Frequency of oscillation proportional to volume
„ Ultrasonic flow meter
– based on the principle that the velocity of sound in a
fluid in motion is the resultant of the velocity of sound in
the fluid at rest plus/minus the velocity of the fluid itself

EDS 2006/Inst-73

Another primary flow element that has gained much popularity over the past 10
years or so is the Vortex meter. A “bluff” body is installed in the cavity of the
meter, which generates vortices in the flowing stream (down steam of the bluff
body) that are shed at a rate proportional to the flow rate. Early designs were
limited to temperature limits not to exceed 400 °F and could not tolerate any
systems with minor amounts of pipe vibration. Current technology has addressed
these issues, which account for the vortex meter gaining wide-spread acceptance.

The last flow meter that we will briefly discuss is the ultrasonic flow meter. Based
on either the Doppler effect or transit time of a sound wave to measure the velocity
of a fluid. The transit time meter performs well in clean services whereas the
Doppler effect meter performs well in dirty services.

For these last four devices, they either have internal moving parts or internal
electrical parts that are subject to failure. When compared to the orifice plate which
has neither, bypassing the flowmeter installation is required in order for
maintenance to be conducted. Meter failure typically requires that the primary
element be removed from the process piping. This generally is not the case for an
orifice plate.

UOP Confidential - Do Not Copy 73

 Diaphragm Type Differential
Pressure Instrument
Lead
Wires Capacitor
Plates
Sensing
Diaphragm
Rigid Insulation
Silicone
Oil Fill

Isolating Welded Seals

Diaphragm Rosemount Electronic
The Cell DP Transmitter

EDS 2006/Inst-74

The orifice meter utilizes a differential pressure transmitter to convert the

differential pressure across the orifice plate into a standard instrument signal (can be
a digital, electrical, or pneumatic). Typical transmitter designs today use a capsule
made from two very thin and flexible metal diaphragms on either side of a heavy
support member that is designed to withstand a differential pressure equal to the full
design pressure of the transmitter.

Some designs use strain gauges to measure the stress on the diaphragm generated
from the differential pressure; some measure the resonant frequency of a wire that
varies with the differential pressure; and some (as shown above) measure changes in
capacitance as a function of differential pressure.

The δ−cell uses the latter technology to measure differential pressure. Pressure up
stream of the orifice plate and pressure downstream of the orifice plate exert
pressure on their respective isolating diaphragms. In turn the incompressible
silicone fluid distorts the sensing diaphragm, which makes up part of the
“capacitor”. This distortion, on the order of 0.0004 inch maximum, changes the
capacitance of the cell with respect to differential pressure. The two lead wires are
connected to the electronic capsule on the top works.

UOP Confidential - Do Not Copy 74

 Level Instruments
(Displacers)

„ Based on principle of buoyancy (effective weight is

less when suspended in a fluid)
„ Consists of cylindrical displacer suspended by a
hanger in a chamber housing, torque arm, knife edge
bearing, and torque tube (i.e. essentially a weigh
scale)
„ Zero point: displacer suspended in vapor space
(weight0% = weight of displacer)
„ 100% point: displacer immersed in liquid

EDS 2006/Inst-75

Based on Archimedes’ principle (a body immersed in a fluid is acted upon by a net

force that is vertically upward and equal in magnitude to the weight of the fluid
displaced by the body) UOP specifies external displacers for measurement lengths
of 60 inches or less.

The displacer, which is not a float, is in a chamber typically external to the vessel.
As the liquid level in the vessel changes, the fluid level in the external chamber
rises/falls and the displacer is more/less immersed in the fluid. The total movement
of the displacer from 0 to 100% immersion is less than 1/4 of an inch.

One manufacturer’s design consists of a cylindrical displacer, hanger, torque arm,

knife edge bearing, and torque tube. This assembly is essentially a weigh scale that
is used to measure the effective weight of the displacer over the range of the
displacer (0 to 100% immersion) as illustrated with the bullet points.

Typical displacer lengths available from most manufacturers are 14, 32, 48, 60, 72,
84, 96, and 120 inches.

UOP Confidential - Do Not Copy 75

 Level Instruments
(Displacers, cont’d)

„ At 100% : weight100% = weight of displacer - weight

of hydrocarbon displaced)
„ Effective weight over the range of displacer
– EW = weight0% - weight100%
– EW = weight of hydrocarbon displaced
„ For a given EW at a specified density (D):
– Volume of HCBN = VHCBN = EW/D
„ At constant cross-sectional area (A) :
– Height of HCBN = VHCBN/A

EDS 2006/Inst-76

As shown over the range of the displacer the effective weight is equal to the weight
of hydrocarbon displaced. Knowing the density of the fluid and the cross-sectional
area of the cylindrical displacer, the effective weight measurement is proportional to
the height of the fluid in the chamber.

UOP Confidential - Do Not Copy 76

 Level Instruments
(Displacers used for interface measurement)

„ Zero point: displacer suspended in HCBN (weight0%=

weightdispacer- weightHCBN disp’d)
„ 100% point: displacer suspended in water
(weight100%= weightdisplacer- weightH O disp’d)
2
„ Effective weight over the range of displacer
– EW = weight0% - weight100%
– EW = weightH 0 disp’d - weightHCBN disp’d
2
„ Interface is contingent upon differences in gravity
between the two fluids, > 0.1

EDS 2006/Inst-77

The displacer can also be utilized for interface measurement. The most common
application is an interface between a lighter hydrocarbon phase and a heavier water
phase. Most designs for the common displacer will handle gravities (and gravity
differences) between 0.1 up to 1.5. If the gravities of the two fluids are nearly
equal, the interface level will not be distinguishable.

UOP Confidential - Do Not Copy 77

 Displacer Type Level Equations

Po

g3 g4 W

l
X Y

g1 g2
Z
A

EDS 2006/Inst-78

The complete chamber assembly comes in 4 orientations. The above diagram

represents the Top/Bottom orientation. The Top/Side, Side/Bottom, and Side/Side
orientations are the three remaining orientations, and define the location of the
flanged connections. Depending upon the vessel style (horizontal vs. vertical),
room availability, and actual measurement length and location, 1 of the 4
orientations will be selected to meet the design criteria and installation
requirements.
The standard transmitter has the advantage of a simple zero check. Any time the
level is below the displacer, the transmitter should show 0% output. The zero
setting is not affected by temperature or composition changes. The actual level
measurement are affect by these changes. If the vessel temperature is relatively
higher than the temperature of the fluid in the external chamber, the level in the
vessel will be different than the level in the displacer due to difference in gravities.

Y = [X(g1-g3) + Z(g1-g2) + (w-l)(g3-g4)]/(g2-g4)

If well insulated, then g1 and g2 are equal and g3 and g4 are equal and essentially
negligible relative to the liquid density. Therefore Y = X.

UOP Confidential - Do Not Copy 78

 Differential Pressure Level Instrument
(Installation Detail)

Reference
To Level Leg (Note3) Note 2
Instrument
Nozzles on
Vessel Low
Pressure

Level Transmitter
Notes: High Pressure
1. All Piping is 3/4"
2. Contractor Shall Advise Length of Reference Leg
3. Fill Reference Leg With Sealing Liquid Specified

EDS 2006/Inst-79

For level measurement lengths greater than 60 inches, a differential pressure

instrument is employed to measure the static head of the vessel fluid. The high
pressure tap on the transmitter is connected at the lower vessel connection. The
static pressure, Phigh = LIQ HEAD + Po, is the pressure exerted on the high pressure
side of the transmitter. The low pressure tap is connected at the upper vessel
connection. The static pressure, Plow = Po, is the pressure exerted on the low
pressure side of the transmitter. The differential pressure is:
ΔP = Phigh - Plow = LIQ HEAD + Po - Po = LIQ HEAD

Vapors at or near the dew point of the vapor pose a unique situation with this type
of application. As vapors condense in the impulse line on the low pressure side
(reference leg) of the transmitter, liquid build-up in the impulse line provides a
negative static head effect. In the worse case a true 100% liquid level would
indicate a 0% level (static head on both sides of the cell would be the same provided
the densities were equal). Filling of the reference leg could also occur, if during an
upset, the vessel level is becomes higher than the upper vessel connection.
To compensate for this type of application or mis-operation, the reference leg is
filled with the same fluid as that inside the vessel prior to or during unit
commissioning. The transmitter must have zero elevation capabilities.

UOP Confidential - Do Not Copy 79

 Differential Pressure Level Instrument

„ Measure static head as fluid level changes

– Static head = level span x fluid specific gravity
„ Differential pressure instrument
– High side = static head + vessel pressure
– Low side = vessel pressure
– ΔP = static head
„ Fill fluid required on low side of transmitter whenever
fluid is at or near bubble point
„ Transmitter capable of zero elevation

EDS 2006/Inst-80

The differential pressure is:

ΔP = Phigh - Plow = LIQ HEAD + Po - Po = LIQ HEAD
The calibrated span for the transmitter, similar to the flow transmitter, is in terms of
“inches water column” or its metric equivalent. The calibrated span is determined
by multiplying the measurement length (level span) times the fluid operating
specific gravity.
The fill fluid can be something other than the actual liquid in the vessel. However
this material must be compatible with the process. Contamination of the process
could lead to catalyst poisoning or other ill effects on the process. In colder
climates a 50-50% mixture of glycol and water can be used.
Zero elevation is a feature with most transmitters that compensates for the negative
differential pressure induced by filling the reference leg with a sealing fluid. As
long as the correct transmitter has been specified, the negative differential pressure
present in the reference leg is “zeroed” out during instrument calibration.

UOP Confidential - Do Not Copy 80

 Analytical Type Devices

„ Open Loop Devices

– Relies on operator to take corrective action
– Quicker analysis than supplied by Lab
„ Commonly specified analyzers
– Moisture monitors
– Oxygen Analyzers
– Gas/Liquid Chromatographs
„ Sampling system, if required, will add complexity
to the operation of the analyzer

EDS 2006/Inst-81

Process analyzers are usually specified by UOP as open loop devices and requires
the operator to take corrective action on the process. The advantage of on-line
analysis is a much quicker update of the process measurement in comparison to
laboratory analysis. Their main justification comes in during upset conditions.
Being able to supply the operator with a much quicker analysis during upsets,
allows the operator to make adjustments in a more timely manner minimizing off-
spec product.
Moisture, oxygen, hydrogen, and chromatograph analyzers are among the more
common analyzers specified by UOP. Some analyzers are in-line, that is the probe
is installed in the process piping. Others have additional piping to route a slip-
stream to and from the process and some even incorporate an elaborate sampling
system to “cleans” the process material of any contaminants which may interfere
with the operation of the analyzer. Lag time is minimized by designing the slip-
stream with a relatively high flow rate.

UOP Confidential - Do Not Copy 81

 Zirconium Oxide Oxygen Sensor
(Thermox Oxygen Analyzer)
Sample Gas 120
100 Temperature=775°C
Ceramic
80
Controlled Heat O= 60
for Constant 40
High Temperature 20
0
Electrodes
-20
100 50 20 10 5 2 1.0 0.5 0.2 0.1
Oxygen Concentration,%
Air Reference Sensor Output vs. Concentration
4.40
Volt
- + Meter 4.20
4.00
Ceramic Sensor 3.80
3.60
Flue Gas Mixture: O2 = 4.2% 3.40 Operating
CO2 = 12.0% Region of
3.20
Sensor
N2 = 83.8% 0
0 100 200 300 400 500 600 700 800
(Partial pressure of Oxygen is constant)
Temperature, °C
Temperature Response
IST-R00-(166A)

EDS 2006/Inst-82

Oxygen analyzers are used most commonly in the flue gas from fired heaters. The
sensor is made of zirconium oxide ceramic coated platinum. Instrument air is used
as a reference on one side of the ceramic, and the flue gas on the opposite side.

At the operating temperature of about 1285 degF, a voltage is developed between

the two sides of the ceramic in accordance with the Nernst Equation.

UOP Confidential - Do Not Copy 82

 Thermox Oxygen Analyzer

Nernst Equation: E = AT Log (20.9/02)

Where ‘A’ is a constant, ‘T’ is the cell temperature, and

‘O2’ is the oxygen content in the sample gas.

EDS 2006/Inst-83

The zirconium oxide analyzer comes in an in-situ version and an extraction type
arrangement. The latter version requires an eductor to withdraw the sample from
the heater stack.

UOP also specifies portable oxygen analyzers for checking oxygen breakthrough
during catalyst regeneration steps. The portable oxygen analyzer, a Teledyne
Micro-Fuel Cell, produces an electric current proportional to the O2 concentration
in the sample by means of the above chemical reactions.

UOP Confidential - Do Not Copy 83

 Gas-Liquid Chromatography

Sample Injection
(Light, Medium, & Heavy Molecules)

Carrier To
Gas Detector

EDS 2006/Inst-84

Gas-Liquid Chromatography is the most widely used form of Chromatography in

the Process Industry. With similarities to Paper Chromatography, the typical
chromatograph is based on the differences in rats of adsorption/desorption of the
various molecular components.
The chromatograph is composed of a long, thin, packed column and accessories to
vaporize liquid samples. The packing is coated with a liquid which absorbs and
desorbs the sample components. A sample, either liquid or gas, is injected and
flushed through the packed column by a carrier gas to the detector. Because of the
adsorption/desorption mechanism, the lighter components travel through the packed
column at a faster rate than the heavier molecules, thus separation of the molecules
is achieved.
Specific chromatograph designs can distinguish between some components in
ranges on the order of 100’s ppm levels. In some UOP advance process control
applications, the chromatograph specifications require some 30 to 35 component
analyses. In more basic applications, one or two key component analysis is required
in the operation of fractionating columns. As with most instrumentation, the more
complex the system the more costly that system will be.

UOP Confidential - Do Not Copy 84

 Feedback Control Loop

„ Loop Components
– Process
– Measuring Means
• Temperature, Flow, Pressure, Level, Analysis
– Final Control Element
• Control Valve Body Types
• Control Valve Performance and Accessories
• Trim Characteristics
• Hydraulics and Sizing Equations

EDS 2006/Inst-85

We have now completed the review of the most common types of instruments to
measure such variables as temperature, flow, pressure, etc. These devices are used
to measure the process variable. Next we will investigate the most common final
control element - THE CONTROL VALVE. There are several body types,
performance issues, trim characteristics and sizing issues.

UOP Confidential - Do Not Copy 85

 Feedback Control Loop
(Block Diagram)
Controller

Set
+ Error Signal Function
Point Generator
-

Process
Output
Variable
Measuring Final Control
Means Element
Measured Manipulated
Variable Variable

Process

Load
Variable

EDS 2006/Inst-86

Thus far we have investigated the Process and the Process Variable side of the
controller. This is the means of being able to measure the process variable, which is
an input to the process controller.
Now we will look at the Final Control Element, which is the mechanism that varies
the Manipulated Variable in order to move the Process Variable in the direction of
the Set Point.

UOP Confidential - Do Not Copy 86

 Final Control Element

„ Control valve is the most often used final control

element in the Process Industry
„ Control valve dissipates energy in order to control
the process (adds operating cost)
„ Alternatives to the control valve
– Variable speed/variable stroke, pumps
– Variable speed turbines, centrifugal compressor
– Variable pitch fans, air condensers

EDS 2006/Inst-87

The most often used final control element in the Process Industry is the control
valve. The control valve is essentially a variable resistance, located somewhere in
the process line between Point A (the source) and Point B (the destination), and
dissipates energy of the system in order to “throttle” the manipulated variable. For
a pumped system the dissipated energy adds operating costs, by increasing
incrementally the horsepower requirements of the pump. Good engineering
practices can optimize the amount of wasted energy, yet maintain some guarantee of
additional throughput above the normal design flow rate.

Alternatives to the control valve are variable speed drivers and controllers. In the
case of centrifugal compressors, variable speed turbines are often part of the
compressor drive package. As we say earlier with the selection of a venturi flow
meter over the conventional orifice plate, the venturi flow meter reduces the overall
pressure drop of the system (by a couple of psi at the most).

With a variable speed compressor, the work required to operate the compressor can
be minimized without the energy dissipation across a control valve. This energy
savings can be a significant reduction in overall operating costs.

UOP Confidential - Do Not Copy 87

 Control Valves

„ Single-seated globe style valve for services requiring 2

inch and smaller body size
„ Quarter turn rotary globe style valve for sizes > 2 inch
(limited in flange rating)
„ Cage-guided globe style valves where rotary globes are
not suitable
„ Butterfly (wafer) valve for low delta P
„ Multi-stage angle valve for high delta P
„ Globe valve for high noise in vapor service

EDS 2006/Inst-88

Control valves come in a variety of styles. Globe, rotary globe, ball, and butterfly
are among the most common styles. Other styles and modifications to the most
common styles are available for special applications.
UOP will specify the single-seated, sliding stem, globe valve with a pneumatic
spring/diaphragm actuator for 2 inch and smaller body sizes. This essentially will
be 1, 1-1/2, and 2 inch control valves. The single-seated valve is available from
some manufacturers up to and including 6 inch body sizes. However actuator
sizing, dependent upon the maximum valve differential pressure, becomes an issue
for 3 inch and larger valves.
For 3 inch and larger valves, UOP will specify the quarter turn rotary globe style
(eccentric globe) control valves. These valves often have limitations for maximum
flange rating (600 RF) and differential pressure shutoff. Also the rotary globe is not
capable of reducing noise via a noise reducing trim package.
When the rotary globe control valve can not meet the mechanical requirements of
the system, then UOP will specify the cage-guided globe style control valve. This
style of valve is a “balanced” design and actuator sizing is minimized due to this
‘balanced’ design.
Control valve performance is also an issue and is greatly affect by the various
components and accessories associated with the valve design.

UOP Confidential - Do Not Copy 88

 Control Valve Performance Overview
Valve Dynamic Performance Variables
Actuator design
effects speed of response

Packing friction
effects dead band
and speed of response Mechanical linkages
effects dead band

Positioner design
effects dead band and
speed of response
(optional in some designs)

EDS 2006/Inst-89

There are a number of control valve components that effect the control valves
ability to provide good position control.
Hysteresis is the effect of a valve following a non-linear oval curve when
first fully opening and then fully closing.
Dead band is the effect that occurs when the control valve signal is
changing, but the valve is not moving. Usually valve dead band dominates
over hysteresis.
Dynamic response is how fast a valve moves.
The positioner, actuator, mechanical linkages and packing all contribute to
the overall hysteresis, deadband and dynamic response of a valve.

UOP Confidential - Do Not Copy 89

 Effect of Control Valve Performance

„ Poor control loop performance, for many loops, can be

directly linked to the poor dynamic performance of the
control valve
„ Because of poor control valve dynamic performance,
most control loops increase rather than decrease
process disturbances
„ Typically users size a control valve based on process
capacity requirements without regards to any dynamic
performance criteria

EDS 2006/Inst-90

In the past several years more attention has been given to control loop performance
and the dynamic performance of the control valve itself. The control valve is a
critical part of the control loop and not all control valve and control valve
accessories are created equal. If the valve is in need of maintenance or an inferior
valve has been installed in the process, poor control valve performance can increase
process disturbances.

Along with defining the hydraulics and valve sizing, UOP has a standard
specification in which we try to address dynamic performance in terms of valve
accessories and specific response capabilities.

We can think of a control loop as an instrument chain. Like any other chain, “the
whole chain is is only as good as its weakest link”. We must insure that the control
valve is not this weak link.

UOP Confidential - Do Not Copy 90

 Control Valve Performance
(Control Valve Step Resolution)

„ The minimum step change in input signal to which the

control valve system will respond while moving in the
same direction
– This phenomenon is caused by the tendency for a control
valve to stick after coming to rest
– This is also known as stiction

EDS 2006/Inst-91

We will cover some of the definitions in our latest standard specification, so that we
can have a better understanding of control valve performance. How much does the
input signal have to increase in order for the valve to move? Control Valve Step
Resolution defines the minimum step change in input signal before the valve does
actually move.

UOP Confidential - Do Not Copy 91

 Control Valve Performance
(Control Valve Static Dead Band)

„ The range through which the input signal to the control

valve can be changed, without the control valve
moving position
– Dead Band results from various phenomena such as
backlash and stiction (friction) and causes the valve
system to require extra input change after a reversal of
direction before actual movement is resumed

EDS 2006/Inst-92

In a typical control loop the control valve is modulating i.e. constantly moving its
position to satisfy the process setpoint. Excessive static dead band is seen as limit
cycling. Often it is impossible to tune out the limit cycling without making the
control loop very unresponsive.

UOP Confidential - Do Not Copy 92

 Control Valve Performance Criteria

„ Control Valve Dead Time (Td)

– Time it takes after a change in input signal for the valve
to start moving
„ Control Valve Step Response Time (T63)
– Time it takes after an input step change, for the valve to
have moved to 63% of the step change
– This includes the valve’s dead time (Td)
„ Control Valve Hysteresis
– Combined effect of step response and dead band that
prevents changes in valve travel

EDS 2006/Inst-93

As was the case for transmitter performance once again performance is a measure of
dead time and the time constant. For the control valve the time constant is defined
as the Control Valve Step Response Time.

UOP Confidential - Do Not Copy 93

 Control Valve Performance Overview
Valve Dynamic Performance Variables
Actuator design
effects speed of response

Packing friction
effects dead band
and speed of response Mechanical linkages
effects dead band

Positioner design
effects dead band, and
speed of response

EDS 2006/Inst-94

Several key components of the valve affect Valve Dynamic Performance:

Actuator design and valve packing friction affect speed of response.

Mechanical linkage and valve packing friction affect dead band.

UOP Confidential - Do Not Copy 94

 Control Valve Performance
(Control Valve Hysteresis)

100
Ideal control valve
80 moves up and down on
Valve Position (%)

the same linear line

60

40
Practical valves move up
and down along a
20 hysteresis curve

0
0 10 20 30 40 50 60 70 80 90
Valve Control Signal (%)

EDS 2006/Inst-95

Hysteresis is the effect that a control valve moves up and down on an oval curve as
shown above. The main difference with hysteresis compared to deadband is that the
valve always moves (although on a curved line rather than the ideal linear line)
when the control signal changes.
With deadband the control signal changes but the valve doesn’t move.

UOP Confidential - Do Not Copy 95

 Control Valve Performance
(Control Valve Dead Band)

100
Control valve does not
move in this deadband
80
zone even though the
Valve Position (%)

control signal is
60 At 70%, the
control valve
signal is reduced
40
to close the valve,
but the valve does
20 not move until
the signal is at
62%
0
0 10 20 30 40 50 60 70 80 90 100
Valve Control Signal (%)

EDS 2006/Inst-96

With dead band we have an effect that worsens control loop performance.
This is due to the deadband zone (the shaded zone) where the control valve does not
move although the control valve signal is changing.
As an example, the control valve signal is at 70%. Due to the control loop error
signal, the controller decides to start closing the valve. Because of the dead band
zone, the control valve does not start to close until the control signal drops to 62%.
This is very detrimental to good control and valve performance.
Control valves should be selected with the smallest dead band possible (<1%).

UOP Confidential - Do Not Copy 96

 Control Valve Performance
(Control Valve Actuator)

„ Control Valve Actuator

– Moves valve stem and plug relative to the controller
signal
– Provide fail-safe position for the valve
– Typically operates on a pneumatic signal
„ Actuator Types
– Pneumatic spring diaphragm
– Pneumatic spring return piston
– Double-acting piston

EDS 2006/Inst-97

In closed loop control, the output from the controller is directed to the final control
element and thus “throttles” the manipulated variable. An actuator by definition is
“a pneumatic, hydraulic or electrically powered device that supplies force and
motion to open or close a valve”. It is this valve accessory that positions the valve
plug relative to the control signal.
The pneumatic spring diaphragm actuator is the most popular actuator for sliding
stem control valves, such as the single-seated and cage-guided globe valves
discussed earlier. Various styles include direct acting (increasing air pressure pushes
down diaphragm and extends actuator stem) and reverse acting (increasing air
pressure pushes up diaphragm and retracts actuator stem). The actuator spring
opposes the air pressure pushing on the diaphragm and it is this opposing spring
force that will move the valve to its fail mode on loss of control signal.
The pneumatic spring return piston actuator provides a much higher steam force
output than the spring diaphragm actuator. For services with high differential
shutoff pressures, the piston style actuator must be used. The piston style actuators
take advantage of higher instrument air supply pressures, thus providing the higher
steam force.

UOP Confidential - Do Not Copy 97

 Control Valve Performance
(Control Valve Actuator, cont’d)

„ Actuator sizing criteria (spring diaphragm)

– Required force (thrust = Tm) must seat valve at
maximum specified shutoff pressure(dPmax)
– Tm = (valve seat cross-sectional area) (dPmax)
– A = Tm/(maximum control signal air pressure)
(A = Actuator diaphragm cross-sectional area)
– Once A is known, pick next largest actuator
„ Dynamic forces, process dP, actuator spring,
stem packing, hysterisis, etc. prevent exact
positioning

EDS 2006/Inst-98

Actuators are sized by comparing the required force to stroke a control valve with
an actuator that can supply the force. The major force required to operate a globe
valve include the sum of the static unbalance of the valve plug, seat load, and
packing friction.

The unbalance force results from the process fluid when the valve is in the closed
position with maximum differential pressure imposed on the control valve. The
unbalanced area for the selected trim must be provided in order to determine the
required force. As the control valves get larger for a given differential pressure, the
required actuator size gets larger.

Seat load is also determined by shutoff requirements. Leak classes have been
developed, and UOP typically specifies Class IV shutoff. Class IV shutoff by
definition sets the Maximum Leakage Allowable at 0.1% of rated capacity.

Packing friction is determined by stem size, type of packing, and the quantity of
compressive load placed on the packing either by the bolting or by the process.

UOP Confidential - Do Not Copy 98

 Control Valve Performance Overview
Valve Dynamic Performance Variables
Actuator design
effects speed of response

Packing friction
effects dead band
and speed of response Mechanical linkages
effects dead band

Positioner design
effects dead band and
speed of response

EDS 2006/Inst-99

The Positioner is a position controller that is mechanically connected to a moving

part of a final control element and that automatically adjusts its output to the
actuator to maintain a desired position in proportion to the input signal.

Recent advancements in positioner design has improved on control valve dead band
and control valve speed of response. Choosing the right positioner can eliminate
some of the effects of dead band and improve the speed of response of the control
valve.

UOP Confidential - Do Not Copy 99

 Control Valve Performance
(Control Valve Positioner)

„ Purpose of Control Valve Positioner

– To maintain and control valve position
– To compensate for control valve dead band
„ Dynamic performance of control valve is influenced
by the type of positioner
– Single stage pneumatic spool valve positioner
– 2-stage high performance pneumatic positioner
– Digital high performance positioner

EDS 2006/Inst-100

The mechanical feedback from the actual valve position allows the positioner to
adjust its output to the actuator to maintain a desired position in proportion to the
process controller input signal within the limitations of the system. Most spring
diaphragm type actuators can withstand air pressures up to 60 psig without damage
to the diaphragm. This is the limiting factor in determining thrust requirements for
a given actuator size.
The most important feature of a good positioner for increased dynamic performance
is that it be a high gain device. Unless the positioner is sensitive to small input
signal changes, the valve assembly will not be able to respond to minor disturbances
in the process variable. Therefore the positioner must be designed such that it
responds quickly to these small changes in input signal.
A second feature of a good positioner is that it must supply the power (in the form
of air supply) needed to move the valve quickly. This power comes in the form of
rapid air flow as needed to move the valve.
The simplest positioner is the single state pneumatic spool valve positioner. In
order to meet the second feature, the spool valve must be modified in order to
supply the rapid air flow to the actuator. However this modification increases
overall air consumption for the control valve assembly.
Two-stage positioners, although more complicated, give excellent dynamic
performance with minimal steady-state air consumption.

UOP Confidential - Do Not Copy 100

 Control Valve Performance
(Positioner/Cascade Loop)
Positioner is the secondary controller in a cascade loop
Output

0% 100%
Flow loop Controller
The control valve position PV SP (Primary)
controller must have a PV
speed of response
(dynamic response)
SP 4-20mA

superior to that of the

4-20mA
loop (primary) controller.

This is very important in

fast control loops such as
flow.
Valve Position
Controller
(secondary)

EDS 2006/Inst-101

For any control loop, the response time of the control and measuring equipment
must be less than the response time of the process.
For the control valve positioner, it’s speed of response must be superior to that of
the loop controller.
For flow loops or other fast loops, only high performance positioners will help
provide good valve performance.

UOP Confidential - Do Not Copy 101

 Control Valve Performance
(Positioner Comparison)

Position Speed of Design Micro-

Control Response Complexity Processor
Single ±1.5% Slow Simplest NO
Stage
Two ±1.0% Fast Complex NO
Stage
Digital ±1.0% Fast Complex YES
HP

EDS 2006/Inst-102

Because of its simplicity, the single state positioner is the most common positioner
provided with control valve assemblies. A single stage positioner may have the
simplest design but provides the worst position control. With the increasing
emphasis upon economic performance of process control, better performing
positioners should be considered. In the past, maintenance personnel often liked the
single stage positioner for it’s simplicity and ease of maintenance. The performance
of the single stage spool valve, internal to the positioner, in controlling the air
pressure in the control valve actuator, limits the ability of the control valve to
provide position control to no better than +/- 1.5%.
The two stage design provides better position control and if set up correctly can
provide position control to within +/- 1 %. It is more difficult to set up than a single
stage. Good quality dry and oil free air is required for good performance otherwise
the air quality effects the performance of the first stage.
A digital valve controller (positioner) should provide the same level of control as a
two stage positioner with the added benefits of the digital interface for set-up and
diagnostics. Digital high performance positioners are invariably easy to set up,
provide very good valve position control and often have the capabilities to perform
control valve diagnostic tests.

UOP Confidential - Do Not Copy 102

 Control Valve Performance
(Positioner Testing)
Open Loop Step Test
70 Digital Positioner Valve A
65
0.5% Steps 1% Steps 2% Steps 5% Steps 10% Steps
60

55
(%)
50

45

40

35 0.5% Steps 1% Steps 2% Steps 5% Steps 10% Steps

70
Single Stage Positioner Valve B
65

60

55
(%)
50

45 I/P Input Signal

Actuator Travel
40 Filtered Flow Rate
35
0 50 100 150 200 250 300 350 400 450
Time (seconds)

The above flow loop test response shows that valve “A” responds to 0.5% step
changes in control signal. Valve “B”, typical of many installed control valves, starts
responding correctly to step changes of magnitude 5% and greater in control signal.

Unless users start specifying dynamic performance criteria when purchasing

control valves , then they will always have to tolerate the type of control valve
response shown by Valve “B”. Dead band is a major contributor to excess process
variability, and the control valve assembly is often a primary source of dead band in
the control loop.

In both valves, the actuator stem motion (green) changes in unison with the input
signal (black) changes. However for Valve B, 2% step changes resulted in the valve
faithfully moving in conjunction with changes in input signal. This can also be
seen in Valve A with changes in flow rate (red). Step changes on the order of 0.5%
in input signal resulted in changes in flow rate; where as with Valve B, step changes
in the order of 2% were required to obtain a corresponding change in flow rate
response.

UOP Confidential - Do Not Copy 103

 Control Valve Performance
(Dynamic Response)

„ Dynamic Performance (dynamic response)

– A measure of how well a control valve performs at
constantly changing valve position to meet the demands of
modulating, multi-frequency disturbances to a closed loop
process
– One way of measuring control valve dynamic performance
is to see how well it tracks the controller set point
– Deviations away from the set point is known as variability
(high variability,measured in %, shows poor control)

EDS 2006/Inst-104

The real test of a control valve is how well it handles disturbances (i.e, load
changes) and how well it tracks the controller setpoint.
The following is a rough measure of variability, deviation away from set point:
Very good variability is 0.5 to 2 %
Good variability is 2 to 4 %
Poor variability is 5 to 10 %
Extremely poor variability is >10%

Therefore during normal operation with the process in automatic, a control audit
could be used to measure variability. The various control loops could be evaluated
based on criticality and variability. Once identified as a control loop with high
(poor) variability and deemed as a critical loop, an investigation could be launched
to identify and improve the dynamic response of the control loop.

UOP Confidential - Do Not Copy 104

 Control Valve Performance
(Requirements)

„ Control valve performance specification

– Primary objective of the specification is to control a process
variability to within the required variability limits
„ Unfortunately there is no one performance specification
for all applications
„ Throughout Industry a large variation in the performance
of control valves exist
„ Users must define the specification and then find valves
that can meet the specification

EDS 2006/Inst-105

Control valves need to do more than just work (i.e move eventually with a large
change in control signal). Hysteresis, dead band and speed of response are all
parameters that effect the control valves dynamic performance.

For Control Valve Dynamic Performance:

Need to define limits on hysteresis and dead band.
Need to specify speed of response.

It’s a control valves ability to have good dynamic performance that reduces
variability (i.e. control a variable, flow, pressure etc. to specific limits of deviation)
Is there a standard? Unfortunately not yet, although the Instrumentation, Systems,
and Automation Society (ISA) is looking at such a standard.

UOP Confidential - Do Not Copy 105

 Control Valve Performance Specification
(Definitions)

„ Step Resolution
– The minimum step change in input signal to which the
control valve system will respond while moving in the
same direction
„ Dead Band
– The range through which the input signal to the control
valve can be changed, without the control valve moving
position
„ Overshoot
– Maximum amount in excess of step change

EDS 2006/Inst-106

The only current dynamic performance standard is the one developed by Entech.
The Entech standard was developed for the paper mill industries and was developed
by Entech engineering consultants to allow them to develop loop performance
criteria for control loops in the paper industry. They found that loop controllers
were very difficult to tune for optimum control because of poor performance in the
control valves.
The performance criteria are geared for control valves in the paper industry.
The Entech standard may not be the ultimate standard for all control valves in the
process industry, however this type of standard should be developed by each user
of control valves to produce a specification that meets their requirements. As such
the Entech standard is a good model to use.
It should be noted that this Entech standard cause many control valve vendors to
redesign their valves to meet this standard.
UOP has formulated a standard for control valve performance and has used the
Entech standard as a model and applied it to the Process Industry. These are some
of the definitions and criteria that appear in UOP’s standard.

UOP Confidential - Do Not Copy 106

 Control Valve Performance Specification
(Definitions cont’d)

„ Dead Time (Td)

– Time it takes after a change in input signal for the valve
to start moving
„ Step Response Time (T63)
– Time it takes after an input step change, for the valve to
move 63% of the step change
„ Step Response Time (T86)
– Time it takes after an input step change, for the valve to
move 86.5% of the step change

EDS 2006/Inst-107

The only current dynamic performance standard is the one developed by Entech.
The Entech standard was developed for the paper mill industries and was developed
by Entech engineering consultants to allow them to develop loop performance
criteria for control loops in the paper industry. They found that loop controllers
were very difficult to tune for optimum control because of poor performance in the
control valves.
The performance criteria are geared for control valves in the paper industry.
The Entech standard may not be the ultimate standard for all control valves in the
process industry, however this type of standard should be developed by each user
of control valves to produce a specification that meets their requirements. As such
the Entech standard is a good model to use.
It should be noted that this Entech standard cause many control valve vendors to
redesign their valves to meet this standard.
UOP has formulated a standard for control valve performance and has used the
Entech standard as a model and applied it to the Process Industry. These are some
of the definitions and criteria that appear in UOP’s standard.

UOP Confidential - Do Not Copy 107

 Control Valve Performance Specification
(Graphical Representation)

100
86.5%
80
63% 60

40
Percentage Change (%)
20

0
0 Time
Td
T63
T86

EDS 2006/Inst-108

The graphical representation depicts a first order response with dead time and
graphically represents Td, T63, and T86.
T63 includes Td (deadtime).
T86 also includes Td.

UOP Confidential - Do Not Copy 108

 Control Valve Performance Specification
(Speed of Response – Fast Loops)

Speed of response for control valves

in flow, differential pressure, and
Speed of Response: pressure control loops
Any step change in the range of
2 to 10% of full valve travel Valve Size Td T63 T86
(inches) (sec) (sec) (sec)
Dead Band: 0 to 2 0.25 0.5 0.75
Less than 0.5% of full valve travel
3 to 6 0.5 1.0 1.5
Step Resolution:
Less than 0.25% of full valve travel 8 to12 0.75 1.5 2.25
Overshoot:
14 to 20 1.0 2.0 3.0
Less than 10%
22 to 24 1.25 2.5 3.75

EDS 2006/Inst-109

Control valve step response and dead time criteria according to UOP control valve
standard.
Using digital valve controllers with diagnostics, control valve step responses can
be measured.

UOP Confidential - Do Not Copy 109

 Control Valve Performance Specification
(Speed of Response – Slow Loops)
Speed of response for control valves
in temperature, level, hand, and
Speed of Response: analytical control loops
Any step change in the range of
2 to 10% of full valve travel Valve Size Td T63 T86
(inches) (sec) (sec) (sec)
Dead Band: 0 to 2 0.5 1.0 1.5
Less than 0.5% of full valve travel
3 to 6 1.0 2.0 3.0
Step Resolution:
Less than 0.25% of full valve travel 8 to12 1.5 3.0 4.5
Overshoot:
14 to 20 2.0 4.0 6.0
Less than 10%
22 to 24 2.5 5.0 7.5

EDS 2006/Inst-110

This is the same information by category for slower loops, such as temperature,
level, etc.

UOP Confidential - Do Not Copy 110

 Measuring Valve Performance

„ How well is the valve performing?

„ Is the valve sized correctly?
„ How well is the valve controlling the process?
„ Does the valve need maintenance?
„ Would a digital positioner with built-in diagnostics
answer these questions?

YES, YES, YES

EDS 2006/Inst-111

It is for the reasons listed above that valve manufacturers have developed digital
valve controllers.
These digital valve controllers overcome many of the problems inherent to
pneumatic positioners.

UOP Confidential - Do Not Copy 111

 Drawbacks With Pneumatic Positioners

„ Single stage positioners do not provide tight position

control
„ 2-stage positioners improve position control but are
more difficult to set-up and maintain
„ Strap-on test equipment is required to perform valve
diagnostic tests
„ Difficult to measure accurately the valve travel

EDS 2006/Inst-112

It is for the reasons listed above that valve manufacturers have developed digital
valve controllers.
These digital valve controllers overcome many of the problems inherent to
pneumatic positioners.

UOP Confidential - Do Not Copy 112

 Advantages of Digital Positioners

„ Improved valve position control

„ Contains a microprocessor to perform valve control
functions, communicate with a host device, and
diagnostics
„ Auto-calibration only takes a few minutes
„ Monitor any process or valve status alarms, as well
as valve position and input control signal while in
service

EDS 2006/Inst-113

Digital valve controllers/positioners (DVC):

DVC’s replace the conventional pneumatic positioner, usually can be retrofitted to
replace existing pneumatic positioners, and contain a microprocessor to perform
valve control functions, diagnostics and communication with a host device (PC,
DCS, Hand held communicator).
The digital positioner provide better valve position control.
Auto calibration of a valve only takes a few minutes. The valve can be monitored in
service to assess any process or valve status alarms, as well as monitor valve
position and input control signal.
The advantages of the digital devices are endless.

UOP Confidential - Do Not Copy 113

 Control Valve Characteristics
100
90
80
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
% Maximum Lift
IST-R00-(171A)

EDS 2006/Inst-114

By definition the inherent characteristic of a control valve is defined as the

relationship between valve capacity and the closure member travel as the valve is
moved from the closed position to rated travel at a constant pressure drop across the
valve.
Bearing in mind valve flow is a function of both the valve travel and the pressure
drop across the valve (as we shall see later), flow characteristic tests are performed
at a constant pressure drop to provide a systematic means of comparing one valve
characteristic design to another. The most common characteristics are linear, equal
percentage, and quick opening.
One needs to evaluate the process hydraulics over its entire anticipated flow range
in order to evaluate which trim would be best suited for the application. The key to
the trim selection is based on the installed characteristic of the valve. Remember
that the above characteristics are based on constant pressure drop across the valve.
The installed characteristic may be distorted contingent upon the real valve pressure
drop variation with respect to flow.

UOP Confidential - Do Not Copy 114

 Installed Characteristic – Linear
100
90 dPmin
80 dPmax

70
60
50
40
30
20 Line resistance distorts a linear characteristic
towad that of a quick-opening valve.
10
0
0 10 20 30 40 50 60 70 80 90 100
% Maximum Lift
IST-R00-(171B)

EDS 2006/Inst-115

If at high flow rates the valve pressure drop is low relative to the valve pressure
drop at low flows, then the ratio of dPmin/dPmax is less than 1. The smaller the ratio
the more the inherent linear characteristic is distorted; and the installed
characteristic will mimic that of a quick-opening style trim. Pipe and equipment
resistance will distort a linear characteristic toward that of a quick-opening
characteristic provided substantial differential pressure is consumed by the system.
Essentially if the differential pressure across the valve is relatively constant, ie
dPmin/dPmax is equal to 1, then a linear trim would have a linear install
characteristic. This implies that the gain of the valve is constant. For a process in
which the ratio is approximately 1, the process would be considered linear and a
linear valve (a valve with constant gain) would be the appropriate choice.
Therefore examining this ratio as provided in the hydraulic analysis is a good basis
for trim selection and we would want to choose a linear trim if this ratio is near 1.

UOP Confidential - Do Not Copy 115

 Installed Characteristic – EQ%
100
90 Line resistance makes an equal-percentage
characteristic appear more linear.
80
70
60
50
40
dPmin
30 dPmax
20
10
0
0 10 20 30 40 50 60 70 80 90 100
% Maximum Lift IST-R00-(171C)

EDS 2006/Inst-116

Again if at high flow rates the valve pressure drop is low relative to the valve
pressure drop at low flows, then the ratio of dPmin/dPmax is less than 1. The smaller
the ratio the more the inherent equal percentage characteristic is also distorted; and
the installed characteristic will mimic that of a linear style trim. Pipe and
equipment resistance will distort an equal percentage characteristic toward that of a
linear characteristic provided substantial differential pressure is consumed by the
system.
Essentially if the dPmin/dPmax is relatively small, ie dPmin/dPmax is less than or equal
to equal to 0.3, then an equal percentage trim would have a linear installed
characteristic. This implies that the installed gain of the valve would be constant.
For a process in which the ratio is small (<0.3), the process would be considered
non-linear and a non-linear valve (a valve with a variable increasing gain) would be
the appropriate choice.
Therefore examining this ratio as provided in the hydraulic analysis is a good basis
for trim selection and we would want to choose an equal percentage trim if this ratio
is much less than 1.

For processes where the ratio is say between 0.3 and 0.7, there is no obvious
selection of one trim over the other. In these cases either trim selection will be
adequate for the service.

UOP Confidential - Do Not Copy 116

 Basic Flow Equation for Liquid Service
Q = Cv (dP /G)0.5 where

Q = Flowrate, gpm at flowing temperature

G = Specific Gravity at flowing temperature
dP = Differential Pressure, psi

Minimum Design Pressure Drop

The valve minimum design pressure drop is the greatest of the following:

1) 50% of system friction drop, exclusive of the control valve, at the normal flow rate;
2) 10% of pump differential (the pump differential is the differential obtained if the
control valve drop is set in accordance with item 1 above);
3) 25 psi.

EDS 2006/Inst-117

The basic equation for liquid flow indicates that flow is proportional to the square
root of the differential pressure drop across the valve. Cv is defined as the valve
flow coefficient and by definition is “ the number of gallons per minute of water
which will pass through a given flow restriction with a pressure drop of 1 psi”.
Basically it is a capacity index with the valve in the wide open position, and allows
the engineer to rapidly and accurately estimate the required size of a restriction in
any fluid system. For a 1 inch valve the Cv = 12. Therefore if the dP is 1 psi, then
12 gpm would be flowing if the fluid was water at 60 °F. If the dP is 4 psi, then 24
gpm would be flowing.
UOP’s normal practice is to calculate the required valve flow coefficient (required
Cv) at the normal conditions and double the calculated Cv to obtain the
“Approximate Valve Cv”. Therefore at the normal flow condition, the valve should
be at approximately 50% of its capacity. The minimum and maximum Cv’s should
also be estimated to ensure that the valve selected is adequate.

Rearranging the basic flow equation and solving for Cv, we can see that for any
given flow rate the differential pressure across the valve is required. For a pumped
system the rules stated above are based on a system where the design flow is 110%
of the normal throughput. If the design flow is say 125 or higher these rules may
require modifications.

UOP Confidential - Do Not Copy 117

 Maximum Flow with 50%
Frictional Drop for Valve

Q = 100 gpm System Friction Drop

G = 0.8
P1 = 150 psig P2 = 100 psig P3 = 0 psig

Basic Equations
0.5 0.5
CV = Q (G/dP) and Q = k (dP)

Valve Sizing and Selection

0.5
CV = Q (G/dP) Valve Cv = 2 x Cv

IST-R00-(172A)

EDS 2006/Inst-118

The following example will help to illustrate the effects of line losses and the
variable pressure drop across the control valve as flow rates change. Two basic
equations can be used to define the liquid filled system. The first is the basic flow
equation, rearranged solving for Cv. The second is the relationship of flow as a
function of the pressure drop of the system.
This example shows 100 psi frictional drop in the system; therefore, allowing for a
valve dP of 50 psi (rule 1) requires that the source pressure be equal to 150 psig. If
the source pressure is constant, what is the maximum flow through the system when
the valve is wide open? We are given two equations with two unknowns. The two
unknowns are the maximum flow through the system (Qmax) and P2 at Qmax.
Once the size of the valve is selected, the two equations can be set up with the two
unknowns and solved simultaneously to determine what the maximum throughput
through the system is with a constant source pressure.

UOP Confidential - Do Not Copy 118

 Maximum Flow with 50% Frictional
Drop for Valve Example

„ Valve Sizing, Selection & Maximum Flow

– See Figure 40 following page 39 in Text Material

EDS 2006/Inst-119

The step by step solution is provided in Figure 40 following page 39 of the text
material. The first step is to determine the calculated Cv at the normal flow rate of
100 gpm. The calculated Cv is 12.6. The Approximate Valve Cv is determined by
doubling the calculated Cv. Therefore the Approximate Valve Cv is 25.2. This
would lead to a valve selection of a 1-1/2 inch single seated globe valve with a rated
Valve Cv equal to 25.
The next step is to use the basic flow equation with the Cv equal to 25. This first
equation has two unknowns. With the Cv at 25 (valve wide-open) the flow then
becomes Qmax (first unknown) and the downstream pressure P2 is the second
unknown. Likewise the second equation, the system pressure drop relationship, can
be set up in terms of Qmax and P2.
The two equations can be solved simultaneously. The solution yields Qmax equal to
115.3 gpm or a 15% increase in throughput. Even with the valve wide open the
available pressure drop across the valve at normal conditions is taken up as
frictional drop in the system, allowing for only an additional 15% increase in flow
rate.

UOP Confidential - Do Not Copy 119

 Reactor System Hydraulics

PT=200 psig
Δ P=10
SEP

ΔP=5 ΔP=5 ΔP=5

ΔP=5 ΔP=5 ΔP=5

ΔP=25
Ps=?

Piping: ΔPfrict = 35 psi

Equipment: ΔPequip = 65 psi
@ flow = 100% (normal), ΔPtotal = 100 psi

EDS 2006/Inst-120

Heat & Weight balance data provides the design basis for the typical reactor system.
Equipment and Piping Specialists will design the equipment and pipe sizes along
with providing pressure drop data for the system.
In the example above the separator pressure Pt (terminal pressure) is given as 200
psig. Based on the H&WB data the piping and equipment hydraulic pressure drop
is estimated to be 100 psig at 100% (normal) flow. What is the pressure drop
allotment across the control valve and what is the source pressure Ps?
Using the previously defined rules the pressure drop across the valve at normal flow
is 50 psig. Therefore the discharge pressure at the pump will be 350 psig. This
value will be used in selecting the appropriate pump for this service. But what
happens to the system at different flows other than normal flow? UOP will
typically design the system for a maximum flow of 110% of normal and a minimum
flow of 60% of normal.

UOP Confidential - Do Not Copy 120

 Low Pressure Pump, Motor Drive
500
S. G. = 0.80 (100 °F)
S. G. = 0.78 (150 °F)
Pump Discharge
400

Valve dP

300 100

200 0
Separator Pressure
Pump dP Friction Drop = 100 psi
50% of Friction Drop = 50 psi
100 Pump dP = 300 psi
10% of Pump dP = 30psi
Pump Suction Pressure
Therefore Friction Drop Governs,
50 Use 50 psi for Control Valve
0
0 20 40 60 80 100 120
Flow, GPM
(or % Design Flow) IST-R00-(173A)

EDS 2006/Inst-121

A system pressure curve as a function of flow rate can be developed to illustrate the
overall hydraulics of the system. The system friction drop curve can be generated
using the basic friction drop relationship. We know that at 100% flow the system
pressure drop is 100 psi. At 60% of normal flow the system pressure drop will be
36 psi and at 110% the system pressure drop will be 121 psi. Therefore the system
friction drop curve can be generated on the curve.
We have also determined that at 100 % normal flow that the pump discharge will be
350 psig. Knowing the flowing gravity of the fluid, the pump head requirement can
be calculated and a pump selected to provide this head. Once the pump head curve
has been determined, then the pump discharge pressure can be generated and plotted
on the system pressure curve at various flow rates.
The differential pressure between the pump discharge pressure curve and the system
friction drop curve is the pressure differential across the control valve. Therefore
the Cv requirement for design (110%) and turndown (60%) can be determined.
This data then allows for determining the rangeability of the control valve along
with valve sizing.

UOP Confidential - Do Not Copy 121

 High Pressure Pump, Motor Drive
(Pump Head & Friction Drop Considered)
S. G. = 0.80 (100 °F)
Pump Discharge S. G. = 0.78 (150 °F)

1400

1300

1200

Valve dP
1100 100
Pump dP

Friction Drop = 100 psi

1000 50% of Friction Drop = 50 psi 0
Pump dP=(1100+50)-50=1100 psi
Pump Suction Pressure 10% of Pump dP=110 psi
50 Pump dP governs,
Use 110 psi for Control Valve
0
0 20 40 60 80 100 120
Flow, GPM (or % Design Flow)
IST-R00-(175A)

EDS 2006/Inst-122

The last example is similar to the second example with the exception of the
inclusion of rule 2 in determining the pump discharge curve. This positions the
pinch point to the right of the 110% of design. Therefore the system is a workable
solution. However notice that the pressure drop at the normal flow rate is higher
(more energy waste is required at the normal flow) but this a consequence for
meeting the design flow rate with a higher head pump.

UOP Confidential - Do Not Copy 122

 Control Valve Hydraulics
4.0
Max Cv/Design Cv=100

3.5 Max Cv/Design Cv=10

Max Cv/Design Cv=5.0
Max Cv/Design Cv=4.0
Max Cv/Design Cv=3.0
3.0 Max Cv/Design Cv=2.0
Max Cv/Design Cv=1.5
Max Cv/Design Cv=1.25
2.5

2.0
UOP
Minimum
Max Cv/Design Cv

1.5
1.15
1.0 0.5
0.1 1.0 10 100 1000
Control Valve Design dP/System Friction Design dP

EDS 2006/Inst-123

The graphical representation illustrates how the relationship between design valve
differential and design friction drop affects the relationship between flow and valve
Cv. This family of curves assumes constant supply pressure upstream and constant
termination pressure downstream of the system.

As the ratio of design valve differential to design system drop increases (x-axis), the
ratio of maximum to design flow (y- axis) approaches the ratio of maximum to
design Cv. In other words as the majority of the pressure differential is available for
the control valve with minimal pressure drop due to piping and equipment, Q is
proportional to valve Cv. Therefore if the valve is 10 times larger then the flow is
10 times larger.

This also illustrates that over-sizing the control valve, will not necessary over-
compensate for bad hydraulics. As an example if the design valve differential is
only 20% of the frictional drop (instead of 50%), even a valve Cv 100 times larger
than required will have a difficult time trying to meet a 110% design throughput.

UOP Confidential - Do Not Copy 123

 Hydraulics – Stabilizer Bottoms
(Non-pumped circuit)

Stabilizer
Column

P = 250 psi Bottoms

Feed-Bottoms Bottoms
Exchanger Cooler Tank
dP = 10 psi dP = 10 psi
Static
Head
Feed
P1 dPF = 20 psi
Cooling Water
Battery Limits
Pressure = 40 psi

1. If Pipe dP Inside Battery Limits = 5 psi, P1 = (250 - 10 - 10 - 5) = 225 psi

2. Valve dP = (225 - 40) = 185 psi
3. If Tank Static Head = 20 psi, dPF = (40 - 20) = 20 psi
4. Total Friction Drop = (10 + 10 + 5 + 20) = 45 psi
5. Valve dP / Total Friction Drop = 185 / 45 = 4.1

EDS 2006/Inst-124

For a non-pumped circuit, the system must be analyzed for pressure drop
considerations. UOP will typically use the doubling of the calculated Cv to
determine the required valve Cv, but other methods could be applied as well.
In the above system, the Valve dP/Total Friction Drop was calculated to be 4.1.
The previous graph can be used to determine what size valve is required to obtain a
115% increase in throughput based on this ratio equal to 4.1

UOP Confidential - Do Not Copy 124

 Control Valve Hydraulics
4.0
Max Cv/Design Cv=100
Max Cv/Design C
v=10
3.5 Max Cv/Design Cv=5.0
Max Cv/Design Cv=4.0
3.0 Max Cv/Design Cv=3.0
Max Cv/Design Cv=2.0
Max Cv/Design Cv=1.5
Max Cv/Design Cv=1.25
2.5
2.0
Max Cv/Design Cv

1.5
1.15
1.0 0.5
0.1 1.0 10 100
1000
Control Valve Design dP/System Friction Design dP

EDS 2006/Inst-125

In order to be able to obtain a 115% increase in throughput with the ratio of 4.1 (x-
axis) the Max Cv/Design Cv of the valve is approximately 1.25. Therefore if the
Design Cv (calculated Cv at normal flow) is 8, a 1 inch valve with a valve Cv equal
to 12 would suffice for this system.

UOP Confidential - Do Not Copy 125

 Control Valve Sizing Equations

„ Control Valve Sizing Handbook

– Liquid Flow Equations
• Non-vaporizing (sub-critical flow)
• Vaporizing (critical flow)
• Non-turbulent
– Gas and Vapor Flow Equations

EDS 2006/Inst-126

Masoneilan’s control valve sizing Handbook has been provided. The various
equations and terminology is provided for Liquid/Gas applications along with the
sub-critical/critical/choked equations.
The equation that we investigated earlier was the sub-critical liquid equation. The
critical liquid equation introduces an FL factor, the liquid pressure recovery term
and a modified pressure differential limit based on fluid vapor pressure and critical
pressure. The factor is a function of valve geometry. Like the orifice plate, the
fluid accelerates past the seat and forms a vena contracta (area of highest velocity
and lowest pressure). As the fluid decelerates back to normal velocity some of the
pressure drop is recovered. A control valve with a smooth shaped outlet (rotary and
butterfly valves) show a higher percentage of pressure recovery than a valve which
has more changes in flow direction (globe valves).
While pressure recovery in a flow measurement device is beneficial, pressure
recovery in a control valve is not. A high pressure recovery corresponds to a low
pressure recovery factor, which increases the possibility of critical flow. If the
pressure anywhere in the system falls below the vapor pressure of the fluid, then
vaporization will occur. If during pressure recovery the pressure rises above the
vapor pressure of the fluid, then the vapors collapse back into the liquid phase. This
phenomenon is known as cavitation. Flashing occurs if after pressure recovery, the
static pressure is still below the fluid vapor pressure.

UOP Confidential - Do Not Copy 126

 Feedback Control Loop

„ Loop Components
– Final Control Element
• Control Valve Body Types
• Control Valve Performance and Accessories
• Trim Characteristics
• Hydraulics and Sizing Equations
– Controller
• P, PI, and PID Modes
• Applications
• Tuning Methods

EDS 2006/Inst-127

This concludes the section on the final control element. The next topic and last part
of the feedback control loop is the controller itself.

UOP Confidential - Do Not Copy 127

 Feedback Control Loop
Controller

Set
+ Error Signal Function
Point Generator
-

Process
Output
Variable
Measuring Final Control
Means Element
Measured Manipulated
Variable Variable

Process

Load
Variable

EDS 2006/Inst-128

In terms of the block diagram the controller has two inputs and 1 output. The two
inputs are the Process Variable and Set Point. The output of the controller is
connected to the final control element and modulates the Manipulated Variable.
The Feedback Controller generates an error signal (difference between Set Point
and Process Variable) and adjust the controller output as defined by the PID
algorithm.
We have discussed dynamic performance of both the transmitter and final control
element, but have not discussed the “transmission means” of the Process Variable or
Controller Output. The means of transmission would not matter if it did not affect
the loop performance. Historically the means of transmission was initially a 3 to 15
psig pneumatic signal representing 0 to 100% of the signal range. The pneumatic
world, as we shall see, was slow and limited process control to the field for most
control loops. We the advent of electronic controllers and transmitters, the
transmission means was an instantaneous 4 - 20 ma electrical signal representing 0
to 100% of the signal range. This lead to the concept of centralized control rooms.
In today’s digital environment, the transmission means is still fast, but is not
continuous as we say with the update rates in the transmitters. This will have an
overall affect on loop performance if the process variability is as fast or faster than
the transmitter update rate.

UOP Confidential - Do Not Copy 128

 Pneumatic Tubing Time Response
First Order Lag plus Dead Time
(1000 feet of 1/4 inch tubing)
100
90
80
- t / TC
% Change = 100(1 - e )
70
63.2%
60
Td
50
TC TC TC TC TC TC TC
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45 50
Time t, sec

Time Change, Controller

t, Time % Full Input, % Full Scale
sec Constant Scale % Full Scale from Final

0 0 0 0 100.0
1 0 0 0 100.0
8 1 63.2 63.2 36.8
15 2 23.3 86.5 13.5
22 3 8.6 95.0 5.0
29 4 3.1 98.2 1.8
36 5 1.2 99.3 0.7
43 6 0.4 99.8 0.2
50 7 0.1 99.9 0.1

EDS 2006/Inst-129

Pneumatic systems are not commonplace in today’s environment, but a brief

discussion on this system is useful to demonstrate the principles behind control
theory. In a compressible medium (instrument air), the speed of a pressure change
(transmitter or controller output) is limited to the speed of sound (≈1100 feet per
second). If the pneumatic transmitter is located 1000 feet from the controller,
nearly 1 second would elapse before the pressure at the controller would even start
to change. In addition each unit length of air tubing has a capacitance (a given
volume) and resistance to flow. The net result is that once the pressure at the
controller does start to change, it changes in an exponential manner (first order lag
with dead time).
The above data illustrates a step change in a transmitter’s output and the measured
pressure at the inlet to the controller as a function of time. The plot of the data
illustrates the transient response of a single time constant process plus dead time.
The dead time is the time from the initial step change in transmitter output until the
controller input signal first start to change. The time constant of the system is the
time required from the initial change in controller input signal until it reaches 63.2%
of the total change. These two times, dead time and time constant, are important in
the determination of system response and stability.
For the 1000 feet of tubing the dead time is shown to be 1 second and the time
constant for the system is 7 seconds.

UOP Confidential - Do Not Copy 129

 Pneumatic Tubing With
Sinusoidal Input

0
Input Output
Signal Signal

A B

Gain = B / A
Phase Lag = 0

0 90 180 270 360 450 540 630 720

Degrees

EDS 2006/Inst-130

Response can also be measured in terms of frequency. To determine the frequency

response of any component, a sinusoidal signal of varying frequency is applied to
the input side of the component. On the output side of the component, the
magnitude and phase shift of the output signal are measured for each input
frequency.
At low frequency input signals the resulting magnitude of the output signal is
typically equal to the magnitude of the input signal and the phase lag is 0 degrees,
i.e. the output signal faithfully reproduces the input signal. The component or
system being tested can be pneumatic, electrical, mechanical, or even the “Process”.
However as the input frequency increases, it becomes more difficult for the system
or component to faithfully reproduce the input in terms of the output signal.
Depending upon the system or component in question, eventually the magnitude of
the output signal begins to diminish and a phase lag less than 0 degrees develops.
Graphically the ratio of the magnitudes is plotted as a function of the input
frequency on a log-log scale, and the phase lag is plotted as a function of the input
frequency on a semi-log scale. These characteristic curves are commonly known as
a Bode Plot.

UOP Confidential - Do Not Copy 130

 Pneumatic Tubing Frequency Response
(1000 feet of 1/4 inch tubing)
1.0
0.707

0.1

0.01

0.001
0.0001 0.001 0.01 0.1 1.0 10 f, Hz
0

-45°
7 Sec Tc Lag
-90
(Lag + 1 Sec Td
Dead Dead Time
Time)
-180
fc = 0.023 Hz
fc = 1 /( 2 pi Tc)
IST-R00-181B

EDS 2006/Inst-131

The logarithmic plot of gain vs. frequency is advantageous from the standpoint that
the product of individual component gains in a loop is equal to the overall loop gain,
i.e, GL=G1xG2x… and can be quickly multiplied to obtain the overall gain at each
frequency.
If two straight lines are drawn asymptotic to the gain curve, the intersection of the
two segments occurs at the “corner frequency”. The frequency at this point is
directly related to the time constant found in the step response previously discussed.
The corner frequency in radians per unit time is equal to the reciprocal of the time
constant:
1/Tc = 2πf
where:
Tc = time constant
f = corner frequency, Hz
For the system in question the corner frequency was determined to be 0.023 hertz.
Solving for Tc yields a time constant for the system equal to 7 seconds.

UOP Confidential - Do Not Copy 131

 Pneumatic Tubing Frequency Response
(Variable Tubing Length)

1.0
0.707

0.1 Tube Length Tc

1000 ft 7 sec
350 ft 1 sec
0.01
100 ft 0.13 sec

0.001
0.0001 0.001 0.01 0.1 1.0 10 f, Hz Tc = 1/[2(Pi)fc]
fc = 0.023 Hz fc = 1.2 Hz
fc = 0.16 Hz

EDS 2006/Inst-132

For the 1000 feet of pneumatic tubing the corner frequency is found and the
resultant Tc is determined to be approximately 7 seconds. In terms of the corner
frequency (0.023 Hertz) this is much to slow a response for good control. UOP
requires a frequency response of at least 1.0 Hertz.
Shortening the pneumatic tubing to 350 feet results in a Tc = 1 second and a
frequency response equal to 0.16 Hertz. In order to meet the 1 Hertz requirement
for good control the pneumatic tubing run has to be limited to something less than
150 feet. The frequency response for 100 feet is shown to be 1.2 Hertz.
This is the main reason why many pneumatic systems for flow and pressure control
loops are mounted in the field as close to the control valve and transmitter as
possible
In terms of electrical devices, electronic analog transmitters and controllers and
temperature devices the frequency response is essentially instantaneous.
The advent of the electrical devices gave way to centralized control rooms.

UOP Confidential - Do Not Copy 132

 Digital Sampling
(Assumed 6 updates per second)
1 1

0.5 0.5
2 sec 1/2 sec
0 0

-0.5 1/2 Hz -0.5 2 Hz

-1 -1
1 1

0.5 0.5
1 sec 1/3 sec
0 0

-0.5 1 Hz -0.5 3 Hz

-1 -1
0 90 180 270 360 450 540 630 720 0 90 180 270 360 450 540 630 720
Degrees Degrees

EDS 2006/Inst-133

The use of electronic analog transmitters have given way to microprocessor base
instruments. Commonly known as smart transmitters, the 4 - 20 ma signal is
continuously updated at discrete intervals.
If the process variable is oscillating in a sine wave fashion and the update rate is
high compared to the frequency of oscillation, then the updated signal yields a fairly
representative picture of the process variable (1/2 and 1 Hz).
Again as the frequency of the process variable increases and approaches the the
update rate of the transmitter, an increasing distorted view of the process variable
results (2 and 3 Hz).
For most fast responding processes, transmitters with slow update rates degrade the
control system and in some instances are completely unacceptable (compressor anti-
surge control).

UOP Confidential - Do Not Copy 133

 Control Algorithm Performance

Control Algorithm Performance

Control Service Update Rate Execution Rate
Flow 4 updates per second 250 milliseconds
Liquid Pressure 4 updates per second 250 milliseconds
Gas Pressure 2 updates per second 500 milliseconds
Differential Pressure 4 updates per second 250 milliseconds
Temperature 2 updates per second 500 milliseconds
Level 1 update per second 1 second
All Others (Analysis, pH, etc) 2 updates per second 500 milliseconds

t1, transmitter updates output

t2, controller updates input

EDS 2006/Inst-134

In terms of the execution rate for the distributed control system, a similar
phenomenon exits with the input to the controller. If the transmitter updates at t1
and the controller updates at t2 before the transmitter updates again, then the
difference in time (dead time) determines how “old” the process variable is, and in
the worse case if completely out of sink, the controller is making adjustments on the
oldest data. If the update rate of the transmitter is doubled, then the time difference
(dead time), is cut in half. Therefore the update rate of the transmitter needs to be
somewhat faster than the controller execution rate. UOP recommends that the
update rate of the transmitter be twice as fast as the update rate of the distributed
control system. In general for control systems which meet these guidelines, the
signal transmission and control will not be the limiting factor in control
performance of the process.

The above table is based on experience with various distributed control systems
available in industry today. Control algorithm execution time is used as a measure
of performance. Flow and liquid pressure loops are fast acting processes and
require the faster execution times. Lower transmitter update rates and higher
controller execution times increase the control loop dead time, the delay before the
output reflects a change in the input.

UOP Confidential - Do Not Copy 134

 Proportional Only Controller
Oc = G e + B
where:
Oc = Controller Output
G = Gain
e = Error
B = Bias (Manual Reset)

e = (PV-SP)
where:
PV = Process Variable
SP = Set Point

Gain = change in Output/change in Input

conversely, G may be represented in terms of Proportional Band (PB%)
PB% = 100% / Gain

EDS 2006/Inst-135

The basic feedback controller has three modes of control: PID. Proportional (P),
Integral (I), and Derivative (D) action operates on the error generated in the
controller and adjusts the output to the Final Control Element accordingly. Error is
defined as the difference between the Process Variable and Set Point. Any
combination of PID can be implemented in most controllers.
We shall investigate the response of each individually. The proportional only
controller is the simplest mode and is defined above:
controller output is equal to the controller gain times error plus bias
The equation which represents the proportional only controller is an equation for a
straight line. The gain of the controller is the slope of the straight line and the bias
term is the y-axis intercept. Gain then is defined as change in output per change in
input. With the input being the error signal, if the error is 1% and the gain is 1, then
the output will change 1%. If the gain is 2, then the output will change 2% for a 1%
error. Essentially the output will change proportionally to the error being generated
by the process.

At UOP, the proportional only control mode is used most often for level
measurement. Surge drums are primary examples where exact level control is not
mandatory in the operation of the unit.

UOP Confidential - Do Not Copy 135

 The Level Process

Qi
100
Oc = G e + B
Level % LN

LT LIC
0

Q0
FC

EDS 2006/Inst-136

In examining the proportional only controller per UOP’s typical design criteria, lets
say the normal flow rate to the vessel is 50 gpm. If a valve is selected with exactly
twice the capacity as the normal rate, the controller output can be viewed as a flow
rate in gpm.
During operation of the P only controller the operator will typically line-out the
process in manual. If the inflow is 50 gpm, then the operator would manually adjust
the output to the valve until the outflow is equal to the inflow. Let’s say this occurs
at a level measurement of 50%. This matches the engineering design of the system
with the level at 50% flow at design and the control valve at half capacity.
The operator switches to automatic control. PV is equal to SP, and no error exists.
The output of the controller remains at 50% until something disturbs the process.
For our equation of controller output, this scenario sets the bias of the controller to
50%, independent of the gain of the controller.

UOP Confidential - Do Not Copy 136

 The Level Controller
(Proportional Only)

100
80 Oc = G e + B

60
Set Point = 50%
Level % 40
B = 50%
20
0
0 20 40 60 80 100
Controller output, % full scale
(or control valve GPM)

EDS 2006/Inst-137

In terms of the controller at the point of switching to automatic control, the PV and
SP are equal and e = zero. The Bias of the controller is set equal to the output of the
controller. This is commonly called “bumpless” transfer.
If the gain of the controller is 1 (100% PB), then the equation is represented as
straight line with a slope equal to 1 passing through the 50% points.
If the inflow decreases such that the level moves to 30%, then an error would exist:
e = 30 - 50 = -20. The output of the controller would be equal to 30% or in terms of
flow 30 gpm. The system for p-only control eventually matches input flow to
output flow, but will always have an offset between PV and SP.

UOP Confidential - Do Not Copy 137

 The Level Controller
(Proportional Only)

Qi = 30 OC = (100/PB)e + B
100
100
80
60
L1 Level % 40
LT LIC L1
0 20
0
0 20 40 60 80 100
FC Q1
Q0 = 30
Controller output, % full scale
(or control valve GPM)

EDS 2006/Inst-138

Assuming that the controller was placed in automatic with a 50% set point when the
level was at 50% sets the B to 50%, fixes the controller equation. If the inflow
changes from 50 to 30 gpm, then the error generated is -20% and the controller
output is 30% or 30 gpm and outflow matches inflow at steady state. What happens
to the system now if we have a step change in inflow from say 30 gpm to 70 gpm?

UOP Confidential - Do Not Copy 138

 The Level Controller
(Variable Gain & Bias)

100
80
L2 OC = (100/PB)e + B
60
Set Point = 50%
Level %
40
L1 B = 50%
20
Offset = PV -SP = (PB/100)(Oc-B)
0
0 20 40 60 80 100 Two ways to reduce offset
Q1 Q2 1. Increase Gain
or
Controller output, % full scale 2. Increase Bias
(or control valve GPM)

EDS 2006/Inst-139

Steady state error , the deviation between the lined-out value of the process variable
and the set point, is also known as offset. This offset can be reduced by increasing
the gain of the controller (smaller PB). Theoretically the gain could be increased to
infinity and thereby decrease the offset to zero. However increasing the gain will
eventually lead to oscillation and control loop instability.
Rearranging the controller algorithm in terms of the steady state error or offset,
indicates that the offset could be eliminated by adjusting the manual bias B equal to
the output of the controller. As indicated this would shift the control line to the
right for the system where the inflow had increased to 70 gpm. The problem with
this solution is at low inflows the tank level will go to zero and the valve will
remain at 20% open.
Proportional control is often selected for level control when inventory control is not
a major concern for process control. In other words the vessel will be used as a
surge drum minimizing upsets to downstream operations.

UOP Confidential - Do Not Copy 139

 The Level Controller
(In-Flow Step Change)

80
Q2
Flow, 60 In terms of surge, “proportional
gpm Qin only” may be a good partner in
not disturbing downstream
40 Qout operating units
Q1
20
t 0 T 2T 3T 4T 5T 6T 7T
Time, Time Constants

EDS 2006/Inst-140

For the example of changing from an inflow of 30 to 70 gpm, the inflow is

represented by a step change at time “zero”. Depending upon the gain of the
controller as the level increases, the output of the controller begins to increase and
as a result the outflow from the vessel increases. The process follows a first order
lag response and is represented by the Qout curve. The area between the two
curves, Qin and Qout, represents the accumulated level in the tank.
If the gain could be adjusted to have the offset equal to zero then the area under the
curve would be zero and as soon as Qin changes Qout would change
instantaneously.
In terms of downstream effects on operation, the outflow is gradually increased and
does not “slug” the changes in inflow to the downstream equipment.

UOP Confidential - Do Not Copy 140

 Proportional Plus Integral Controller
Oc = G[e + (1/Ti) f0t e dt]
where:
Oc = Controller Output
G = Gain
e = Error
Ti = Integral Time, minutes per repeat

e = (PV-SP)
where:
PV = Process Variable
SP = Set Point

Integral Time, (Ti) = minutes per repeat

conversely, Ti may be represented in terms of Reset (I), repeats per minute
I = 1 / Ti

EDS 2006/Inst-141

With the proportional only controller, we discovered earlier that we could reduce
the offset by manually readjusting the bias. By replacing this manual bias with the
integral term (I in PID) the form of the equation is similar except the manual reset is
replaced by the integral action and automatically moves the controller output in the
direction that minimizes the offset. The integral time is defined as minutes per
repeat. One minute per repeat means that in each one minute interval, the Integral
action of the controller will cause a change in controller output that is equal to the
change in output that results from the Proportional action alone.
As the integral time gets larger and larger the reset action diminishes. If in fact we
set the integral time to infinity, the reset action of the control is essentially zero, i.
e.. proportional only control.

UOP Confidential - Do Not Copy 141

 Open Loop Integral Response
(Gain = 2.0)

Proportional Only
Gain = 2
(PB=50%)

PV

PV SP
0 1 2 3 4 5
Time, minutes

EDS 2006/Inst-142

What is the effective output of a PI controller? An open loop response will give an
indication of the integral action contribution. With the process lined-out and the
controller in manual (PV, SP, and OP at steady state), a load change occurs at t=0.
Assuming that the load change causes a step change in PV equivalent to 1 unit and
the GAIN of the controller is 2 (proportional action), the proportional action of the
controller will change the output of the controller by 2 units at t=0. With the
integral time set to infinity, the output of the controller would remain constant for t
greater than 0. Notice that as time progresses the model shows both the PV and SP
are constant; therefore the error signal in the controller is constant.
If the integral time is set to 1 minute/repeat, then by the definition of integral action
the controller output will change an additional 2 units every minute. If the integral
time is set to 4 minutes/repeat, the controller output will change an additional 2
units every 4 minutes.
The Integral contribution as shown in the equation is the time integral of the error
function of the controller. Therefore when the controller is in automatic and the
controller output is throttling the manipulated variable, the error signal should be
diminishing. How fast this occurs depends upon the controller settings for the PI
controller, that is what is the gain setting of the controller and what is the Integral
setting of the controller. There is an infinite number of combinations: some good
and some bad.

UOP Confidential - Do Not Copy 142

 Open Loop Integral Response
(Gain = 0.5)

PV
Integral Proportional Only
Gain Gain = 0.5
PV (PB = 200%) SP
0 1 2 3 4 5
Time, minutes

EDS 2006/Inst-143

This is another example of the open loop response of a PI controller. The gain of
the controller is set at 0.5. Again for a 1 unit load disturbance in the PV, the output
of the controller is immediately change by 0.5 units due to the proportional action of
the controller. Depending upon the Integral time, the controller output integrates
the error signal with respect to time.

UOP Confidential - Do Not Copy 143

 Proportional + Integral
(Frequency Response Curves)
100
fc corresponds
Ti = (2(pi)(f)) - 1 = 1.6 sec.
to Ti (Integral)
10

Controller
Process Gain = 2
1.0 For system to be
stable, the sum of
the gains must be
0.1 less than 1.0 at a
90 phase shift equal
45 to -180 degrees
Process Controller
0
-45
-90
-135
-180
0.0001 0.001 0.01 0.1 1.0 10
Frequency, Hz

EDS 2006/Inst-144

Similar to the pneumatic tubing examples shown previously, we can investigate the
frequency response for the PI controller. Individual components of the loop can be
shown on the frequency response curve. The unique feature of the frequency
response curve is that individual components of the loop are additive on the log
scale. The process is shown as a single time constant process similar to the
pneumatic tubing case investigated previously. (Flow loops and pressure loops are
good examples.) The open loop response for the controller is shown with the
Proportional action set at GAIN=2 and Integral action set at Ti=1.6 seconds/repeat.
For high frequencies the proportional action of the controller is the dominant action.
For low frequencies the Integral action of the controller will have significant affect
on the controller output. In terms of the “corner frequency”, the Integral time is
equal to the corner frequency. Therefore by adjusting either the Integral time, the
Gain, or both, the controller and sum of the controller plus process will be altered.
However experimentally this is difficult to collect this type of data for any real
process.
When the combined open loop response is plotted, Process plus Controller, the
stability of the loop can be checked. For the system to be stable, the sum of the
gains must be less than 1 at a phase shift equal to -180 degrees. There are many
settings that will provide stable control and many that are unstable.

UOP Confidential - Do Not Copy 144

 P + I + Derivative Controller
Oc = G[e + (1/Ti) f0t e dt + Td(de/dt)]
where:
Oc = Controller Output
G = Gain e = Error
Ti = Integral Time, minutes per repeat
Td = Derivative Time, minutes

e = (PV-SP)
where:
PV = Process Variable
SP = Set Point

Derivative Time, (Td) = minutes

The derivative contribution is proportional to the
Rate of Change of the error signal

EDS 2006/Inst-145

PI control is typical for flow and pressure loops. These are fast-acting loops which
have little, if any, lag in the process and small time constants. The amount of
energy stored in the process is small. For a liquid flow loop as soon as the valve
moves, the flow measurement changes almost instantaneously. For vapor flow or
pressure loops, a small amount of energy is stored in the process due to the
compressibility of the vapor. However the process is still first order with a slightly
larger time constant.
For systems with a large amount of stored energy, such as temperature loops, lag in
the process is significant. This type of process can not be modeled with the
simplified first order process with a single time constant. Temperature loops also
inherently possess a significant amount of dead time. A decrease in heater outlet
temperature, causes an increase in burner pressure increasing the heat release. This
must first heat up the tubes, then the process, and eventually get to the temperature
measuring device on the heater outlet. This process has significant lag and dead
time.
To be effective and catch up with the process change, the controller needs a high
gain at the higher frequencies where the changes occur the fastest. This is
accomplished with Derivative action. Derivative action takes into account how fast
the error is changing and is proportional to the rate of change of the error signal
with respect to time. Setting Td to zero, defaults to PI control.

UOP Confidential - Do Not Copy 145

 Open Loop Derivative Response
(Gain = 2.0)

PV

SP
PV 0 1 2 3 4 5
Time, minutes

EDS 2006/Inst-146

What is the effective output of a PD controller? An open loop response will give an
indication of the derivative action contribution. With the process lined-out and the
controller in manual (PV, SP, and OP at steady state), a load change occurs at t=1.
Unlike the open loop response for the integral action (a step change in PV was
initiated), the load change causes a constant rate change in PV equivalent to 0.3333
units per minute. Assuming the GAIN of the controller is 2 (proportional action),
the proportional action of the controller will change the output of the controller
proportional to the rate of change of the PV. If Td is set to zero, that is proportional
only control, the controller output would ramp up twice as fast at the change in PV.
With the rate of change at 0.3333 units per minute, Td set at 3 minutes, and GAIN
set at 2, the Derivative action of the controller is equal to the product of these three
settings (0.3333 x 3 x 2 = 2.0 units). Therefore as long as the rate of change is
constant, the Derivative action is constant at 2 units. With the proportional action
set with a gain of 2 this must be added to the derivative response, which is
accumulative because of the change in error signal. At t=1 the load change appears
and at t=4 the load change disappears with the error being constant at one unit. At
t=4 the Derivative contribution is zero because the rate of change of error with time
is zero. However there is a constant error of 1 unit and the proportional action
remains with a change in controller output equivalent to 2 units.

UOP Confidential - Do Not Copy 146

 Proportional + Integral + Derivative
(Frequency Response Curves)
100
Ti = (2(pi)(f)) - 1 = 1.6 sec. Td = (2(pi)(f))- 1 = 0.16 sec.

10 Ti Td

Process
1.0
For system to be
0.1 stable, the sum of
90 the gains must be
45
less than 1.0 at a
Process phase shift equal
0 to -180 degrees
-45
-90
-135
-180
0.0001 0.001 0.01 0.1 1.0 10 100
Frequency, Hz
IST-R00-186A

EDS 2006/Inst-147

Similar to the PI frequency response shown previously, we can investigate the

frequency response for the PID controller. Again the process is shown as a single
time constant process. The open loop response for the controller is shown with the
Proportional action set at GAIN=2, Integral action set at Ti=1.6 seconds/repeat, and
now Td set at 0.16 seconds. For high frequencies the derivative action of the
controller is the dominant action, and as before for low frequencies the Integral
action of the controller will have significant affect on the controller output. In terms
of the “corner frequency”, the Integral time is equal to the corner frequency for that
portion of the controller response and likewise the Derivative time is equal to the
corner frequency as determined for the other portion of the controller response.
Therefore by adjusting either the Integral time, the Gain, Derivative time, or any
combination of the three modes, the controller and sum of the controller plus
process will be altered. Again however experimentally this is difficult to collect this
type of data for any real process.
When the combined open loop response is plotted, Process plus Controller, the
stability of the loop can also be checked. For the system to be stable, the sum of the
gains must be less than 1 at a phase shift equal to -180 degrees. There are many
settings that will provide stable control and many that are unstable.

UOP Confidential - Do Not Copy 147

 Gas Concentration Unit Stripper
Primary controller

Overhead vapor rate is used

FT FIC OP
to control stripping rate by
adjusting the flow of heat
medium to the reboiler.

Thermal lags in reboiler Secondary controller

makes D action useful for
the primary flow
LIC LT
controller

SP
FIC
FT

Heavy LCO to and

from FCC unit
IST-R00-189A

EDS 2006/Inst-148

PID control is typically reserved for temperature control where thermal lags in the
reboiler (either exchangers or fired heaters) and transport lags internally in the
column makes control more difficult. Because it takes longer for a change in
control valve position to have an effect on temperature, the derivative action is a
means of offsetting that delay. When the temperature begins to change, a controller
with derivative action responds immediately anticipating or predicting the
temperature in advance based on the rate of change.
For the gas concentration unit stripper flow is cascaded with flow to the reboiler
heat input. The system deals with again similar thermal and transport lags in the
reboiler circuit and derivative action is useful for the secondary flow controller.
However because the secondary controller involves a flow measurement,
dampening of the flow transmitter is required to eliminate any high frequency noise.
Remember derivative action is based on rate of change and a noisy process variable
will have a negative impact on good control.

UOP Confidential - Do Not Copy 148

 Controller Tuning

„ Frequency Response Plots

– Difficult to obtain data
– Excessive amount of time gathering data
– Unit continuously upset
„ Real Time Methods
– Closed Loop Method
• Ziegler-Nichols Method
– Open Loop Method
• Process Reaction Curve

EDS 2006/Inst-149

As we reviewed the Bode Plots (frequency response curves), one feature was quite
evident. Tuning parameters, Gain, Ti, and Td, were randomly selected to illustrate
the slope and magnitude of their contribution to the controller frequency response.
To determine the frequency response of the controller, the process variable would
need to be varied in a sinusoidal fashion and the respective output of the controller
would be recorded for various combinations of the tuning parameters. On real
processes this is very impractical. The bode plots lend themselves to a good
understanding of control theory, but in the real world are not practical for loop
tuning.
There are various methods available for loop tuning. Two of the more common
methods are the Ziegler-Nichols closed loop method and the Process Reaction
Curve open loop method. Both have their advantages and disadvantages. However
both will provide adequate tuning parameters that will result in stable control, and
in both cases, additional fine tuning of the control loop will provide adequate closed
loop control.

UOP Confidential - Do Not Copy 149

 Ziegler-Nichols Closed Loop Method

„ On-Line Closed Loop

„ Determine Period of Oscillation for P-only
„ Procedure
– In manual line-out P-only controller
– Set Gain; switch to automatic mode
– Change Set Point; 2-3% Step
– Observe controller output oscillation
– adjust Gain accordingly, repeating procedure
until continuous oscillation (GC) is found

EDS 2006/Inst-150

The Ziegler-Nichols closed loop method determines the gain (proportional action)
and period of Oscillation for continuous cycling with a Proportional only controller.
At this gain the total gain of the loop is 1.0. For higher controller gains the total
gain of the loop is greater than 1.0, and the loop will be unstable (the magnitude of
each additional period is growing larger). For lower controller gains the total gain
of the loop is less than 1.0; however, the loop will be stable (the magnitude of each
additional period is growing smaller)and will eventually line-out at some steady
state value.

UOP Confidential - Do Not Copy 150

 Ziegler-Nichols Closed Loop Method
(GC and P Determination)

Unstable
Continuous Cycling
Stable

P = Period of Oscillation
GC = Gain at Continuous Cycling

Time IST-R00-190A

EDS 2006/Inst-151

By following the procedure, a trial and error approach is conducted until the gain of
the controller produces continuous oscillation. Continuous oscillation occurs when
the magnitude of each cycle is the same. If the gain of the controller is too large
then the process is unstable and the resultant oscillation increases in magnitude.
Once continuous oscillation is found, the gain of the controller, Gc, and period of
oscillation, P, are noted.

UOP Confidential - Do Not Copy 151

 Ziegler-Nichols Closed Loop Method
(Calculating P, I, and D Tuning Constants)

„ Proportional only
– G = 0.50 x GC
„ Proportional plus Integral
– G = 0.45 x GC
– Ti = 0.85 x P
„ Proportional plus Integral plus Derivative
– G = 0.60 x GC
– Ti = 0.50 x P
– Td = 0.12 x P

EDS 2006/Inst-152

For level control UOP will typically recommend proportional only control or PI
control with a large Integral time to eventually bring the level back to its set point
(surge drum level control). For most flow and pressure control loops, PI control is
sufficient. PID control is usually limited to temperature control loops; and in some
instances, PID control can be used on analyzer loops.

Once the controller modes are selected, whether it be P, PI, or PID control, the
tuning parameters are estimated as shown above based on the Gain and Period
determined from continuous oscillation.

UOP Confidential - Do Not Copy 152

 Ziegler-Nichols Closed Loop Method
(Determining P, I, and D constants for a PID Controller)

„ For a Temperature Controller

– it was determined that GC = 10 and P = 5 min
– select PID controller
„ Calculate Tuning Parameters
– G = 0.60 x GC = 0.60 x 10 = 6.0
– I = 0.50 x P = 0.50 x 5 = 2.5 minutes
– D = 0.12 x P = 0.12 x 5 = 0.6 minutes

EDS 2006/Inst-153

As an example the Ziegler-Nichols closed loop method was applied to a temperature

control loop. The closed loop method yielded continuous oscillation with the gain
of the controller set at 10. The Period of Oscillation was measured to be 5 minutes.
PID controller is selected for the temperature loops and the collected data yields the
G equal to 6, Ti equal to 2.5 minutes, and Td equal to 0.6 minutes.

UOP Confidential - Do Not Copy 153

 Open Loop Method

„ On-Line Open Loop

„ Determine Dead Time and Reaction Rate
„ Procedure for Process Reaction Curve
– Line out at steady state with manual control
– Induce step change (dM) in controller output
– Observe the Process Reaction Curve
– Obtain Dead Time (L) and Reaction Rate (R)
– Calculate Parameter (X) = (L x R)/(dM)

EDS 2006/Inst-154

The Process Reaction Curve open loop method determines the dead time and
reaction rate of the process. Once the process dead time and reaction rate are
determined, then the tuning parameters can be determined. The procedure is
straight forward and easy to implement.

UOP Confidential - Do Not Copy 154

 Open Loop Method
(Process Reaction Curve)

Dead Time = L, min

Rate = R = A/B, % Change/min
X = (R x L)/dM Point of
Inflection

A = % Change in
Controlled Variable

L B

dM = % Change in
Controller Output

t0 Time IST-R00-190B

EDS 2006/Inst-155

The Procedure for Process Reaction Curve is as follows:

1) With the controller in manual, line out at steady state
2) Induce a step change (dM) in controller output (2 - 5% change)
3) Observe the Process Reaction Curve
4) Draw a straight line through the inflection point on the curve
5) Estimate Dead Time (L) and Reaction Rate (R)
6) Calculate Parameter (X) = (L x R)/(dM)

Step 4 is the most difficult step with this graphical method. L and R are based on
the inflection point and slope of the line drawn.

UOP Confidential - Do Not Copy 155

 Open Loop Method
(Calculating P, I, and D Tuning Constants)

„ Process Reaction Curve Yields

– Step (dM) % Change in Controller Output
– Dead Time (L)
– Rate (R) % Change in Controlled Variable/min
– Parameter (X)
„ Controller Tuning Constants
– G = 0.375 x (X) (Range 0.25 to 0.5)
– Ti = 2.5 x (L) (Range 2.0 to 3.0)
– Td = 0.5 x (L) (when Derivative is used)

EDS 2006/Inst-156

Once the parameters are extracted from the Process Reaction Curve the controller
tuning parameters can be determined. Due to the uncertainty in the Process
Reaction Curve the tuning parameters have bracketed ranges and can be adjusted
with in these ranges for overall improvement in the control loop response.

UOP Confidential - Do Not Copy 156

 Open Loop vs. Closed Loop Method

„ Advantages
– Tuning Constants obtained in minimal time
– Unpredictable amplitude oscillations avoided
– Small Step Change avoids Process Upsets
„ Disadvantages
– Controller and Valve Dynamics not included
– Noise interferes with Graphical Parameters
– Tuning Constants not as accurate

EDS 2006/Inst-157

These are some of the advantages and disadvantages of the open loop method
compared to the closed loop method. The open loop method minimizes the time to
obtain the process reaction curve, avoid unpredictable amplitude oscillations, and
can be implemented with little likelihood of a process upset.
However as pointed out previously, this graphical method interferes with a noisy
process and the tuning constants are not as accurate partly due to the fact that the
straight line drawn through the inflection point will alter the results if not drawn
correctly. Another disadvantage of this method, is the valve dynamics and
controller dynamics are not included because it is an open loop test method.

UOP Confidential - Do Not Copy 157

 Proportional Only Controller Response
Recovery after Load Change

Controller Gain Set Point

1.0
2.5
4.0

Time IST-R00-191A

EDS 2006/Inst-158

For the example provided the proportional only controller is at steady state prior to a
load change. The gain of the controller was increased as indicated from an initial
setting of 1.0 to 4.0. As the gain is increased the offset is smaller, but the period of
oscillation is decreasing, the frequency of oscillation is increasing, and the control
loop is becoming less stable. For a proportional only controller with level control,
setting the gain at around 2.5 would be adequate.

UOP Confidential - Do Not Copy 158

 P plus I Controller Integral Response
(Load Disturbance)
Recovery after Load Change

Set Point

Controller Gain Integral

2.0 50
2.0 20
2.0 10

Time IST-R00-191B

EDS 2006/Inst-159

For PI control the gain is fixed at 2.0 in this example and the Integral action is
varied from 50 seconds/repeat down to 10 seconds/repeat. With the Integral time at
50, the controller appears to respond as a proportional only controller. Eventually
with the small amount of Integral action the process variable will line-out at it’s set
point. As the Integral action is increase the amplitude of oscillation tends to
increase, the period of oscillation increases, and as a result overshoot also increases.
As can be seen above too much integral action causes excessive gain in the
controller and a significant amount of oscillation around the set point. For this
example Integral time somewhere around 20 second/repeat provides adequate
control for a load change on the process.

UOP Confidential - Do Not Copy 159

 P plus I Controller Integral Response
(Change in Set Point)
Recovery after Set Point Change

Set Point

Controller Gain Integral

2.0 50
2.0 20
2.0 10

Time IST-R00-191C

EDS 2006/Inst-160

This example explores the same process as previously described, but instead of a
load variable disturbance we have a change in set point that disturbs the system.
With the controller gain remaining at 2.0 and the Integral time varying between 50
and 10, a similar response is seen for the variation in Integral time. Again with very
little Integral time (50 seconds/repeat) the controller responds slowly to the change
in set point but eventually will line out at the new set point. As the Integral time is
decreased (more Integral action) the magnitude of the oscillations increase and more
overshoot is experienced. Again an Integral time at 20 provides adequate control.

UOP Confidential - Do Not Copy 160

 PID Controller Derivative Response
(Load Disturbance)
Recovery after Load Change

Set Point

Controller Gain Integral Derivative

3.0 20 2
3.0 20 5
3.0 20 10

Time IST-R00-191D

EDS 2006/Inst-161

Addition of the Derivative mode has a tendency of reducing the period of

oscillation. Unlike Integral action, which accounts for a lag on the system,
Derivative action anticipates how fast the error is changing and accounts for a lead
on the system. This actually counterbalances the lag generated by the Integral
action. As can be seen in this example with gain and Integral time held constant, an
increase in Derivative time increases the controller gain. As a result the magnitude
of the oscillations is increased and will lead to instability. For this example a
Derivative time of 2 minutes provides adequate control minimizing overshoot and
drives the process variable back to it’s set point.

UOP Confidential - Do Not Copy 161

 Initial Tuning Constants
GAIN INTEGRAL DERIVATIVE
(PB%) (min/repeat) (minutes)
LEVEL 1.1 (90%) > 60 0
FLOW 0.4 (250%) 0.05 – 0.10 0
PRESSURE (gas) 2.0 (50%) 0.5 0
TEMPERATURE 2.5 (40%) 0.5 0.5 – 1.0

• Liquid pressure or liquid differential pressure loops are set similar to

flow.
• Column tray temperature loops use higher integral and lower derivative
times.
• Analyzer (composition) loops are set similar to tray temperature.
• UOP philosophy on level control allows vessel to act as surge volume
allowing offsets to occur in the level for set point or load changes. The
addition of a long integral time, will slowly return the process variable
to its set point.

EDS 2006/Inst-162

In general UOP recommends the above settings for initial tuning parameters at start-
up of a new unit. These values will typically provide stable but sluggish control.
The intent here is to insure that no one controller is unstable. Once the unit is up
and running, loop tuning can be address on an individual basis.

UOP Confidential - Do Not Copy 162

 Quarter Amplitude-Decay Ratio

P = Period
P a:b = 1:4

b
a

There can be many controller settings

which will yield a decay ratio of 1:4

Time IST R00 191

EDS 2006/Inst-163

What should one shoot for in terms of controller response? One method looks at
quarter amplitude decay ratio. In this method the ratio of Magnitude on successive
cycles is 4 to 1. Another method may be to minimize overshoot. By reducing either
the gain or Integral time in the above example would reduce the controller gain;
thus the overshoot would be reduced and the magnitude of the oscillations would be
less (this would result in a higher ratio as compared to the Quarter Amplitude-
Decay Ratio).
Care must be taken not to reduce the gain significantly otherwise the controller will
become too sluggish and will not provide sufficient gain in the controller to drive
the PV back to its set point in a reasonable amount of time.

UOP Confidential - Do Not Copy 163

 Feedback Control Loop

– Loop Components
• Controller
– P, PI, and PID Modes
– Applications
– Tuning Methods
„ DCS System Requirements
– Control System Performance
– Loop Integrity
– Data Storage and Retrieval
– Analog Outputs

EDS 2006/Inst-164

This concludes our discussion on the feedback control loop and the individual
components of the feedback control loop. Now we will look at some of the UOP
requirements for distributive control systems (DCS). Performance, integrity,
historical data storage, etc will be briefly discussed.

UOP Confidential - Do Not Copy 164

 System Requirements
(Control System Performance)

„ Digital DCS Systems

– Control algorithm execution rate shall be used as a
measure of performance
– Execution rate is defined as the total time the control
system takes to read the process data from the input
processor, perform the control algorithm, and change
the controller output to the output processor

EDS 2006/Inst-165

As discussed previously UOP uses as a measure of performance the control

algorithm execution rates. This is the time it takes for the input processor to read
the PV, update the controller, perform the control algorithm, update the controller
output, and for the output processor to update the output to the field.
This execution rate is important for fast acting loops such as flow and liquid
pressure loops.

UOP Confidential - Do Not Copy 165

 Control Algorithm Performance

Control Service Update Rate Execution Rate

Flow 4 updates per second 250 milliseconds
Liquid Pressure 4 updates per second 250 milliseconds
Gas Pressure 2 updates per second 500 milliseconds
Differential Pressure 4 updates per second 250 milliseconds
Temperature 2 updates per second 500 milliseconds
Level 1 update per second 1 second
All Others (Analysis, pH, etc) 2 updates per second 500 milliseconds

EDS 2006/Inst-166

Again for processes that are instantaneous (flow, liquid pressure, differential
pressure) the update rate is 4 times per second as a minimum. For process that have
some capacity in the system (gas pressure, temperature, etc), the update rate is 2
times per second. For level measurement, as long as the vessel has been adequately
sized, the update rate is 1 time per second as a minimum. If the particular DCS
system has Update Rates that exceed these requirements, process control will be
enhanced provided that the transmitters have a comparable update rate.

UOP Confidential - Do Not Copy 166

 System Requirements
(Control System Performance, cont’d)

„ Analog Electronic Systems

– Control loop frequency response shall be used as
a measure of performance
– Control loop frequency response is defined as the
frequency response from the output connection at
the transmitter to the input connection at the
actuator of the final control element
– This frequency response shall be at least 1 Hertz
(1 cycle per second)

EDS 2006/Inst-167

When we looked at the pneumatic tubing case, it was mentioned that the frequency
response of the pneumatic system should be at least 1 cycle per second. This
requirement lead to most pneumatic controllers to be field mounted because of the
distances involved with centralized control rooms. For distances in excess of 100
feet the time constant was too large to provide adequate control for most processes.
With the advent of electronic controllers, moving now from the speed of sound to
the speed of light, transmission of the PV and OP signals are essentially
instantaneous and do not inhibit the control system as compared to the pneumatic
transmission of these signals. However control loop frequency response shall also
be used to evaluate electronic control systems. For most systems this requirement is
generally met without limitations.

UOP Confidential - Do Not Copy 167

 System Requirements

„ Loop Integrity
– A single component failure will not cause the loss of
more than one control loop
– Shared equipment shall be furnished with redundant
or “backup” systems
• Uninterruptible Power Supply
• Redundant Equipment
– Standby Manual Stations provide access to the final
control elements

EDS 2006/Inst-168

Analog electronic controllers and single loop digital controllers meet the Loop
integrity requirement. However design of DCS systems have optional equipment
that can be added to meet the loop integrity criteria. Most DCS systems can be
purchased with redundant power supplies, redundant controllers, redundant I/O
processors, etc. Also UPS systems can be provided and sized to ensure that the
system is operational for a given amount of time in the event of a main power
failure to the DCS equipment.
An optional piece of equipment is the Standby Manual Stations. These devices
allow the operator or maintenance personnel to manually drive the final control
elements. This equipment is external to the DCS operator console and allows the
operator the interface to the final control element when access to the operator
consoles is loss.

UOP Confidential - Do Not Copy 168

 System Requirements

„ Data Storage and Retrieval

– Plant material balance determination
– Equipment performance trend analysis
– Malfunction troubleshooting analysis

Minimum Requirement
Specified process variables shall be sampled and the instantaneous
value shall be stored at intervals of five seconds or less. Storage
capacity shall be sufficient to store process variable data for at
least a 7 day period.

EDS 2006/Inst-169

UOP will typically specify the minimum amount of data to be stored long term for
each process. This data is can be used for Plant material balance considerations,
equipment performance trends, troubleshooting, etc. The UOP P&ID’s will
graphically represent the data to be recorded long term. If the client wishes to store
more data, UOP will not object. UOP’s requirements, as shown on the P&ID’s, are
the minimum requirement and should be stored for at least a 3 day period.

UOP Confidential - Do Not Copy 169

 System Requirements

„ Analog Outputs
– Controller and manual station outputs should be
capable of operation from 25% below to 25% above
the standard 0 to 100% signal range
– The (+) or (-) 25% margin is used to provide
additional force at the control valve actuator in
order to insure proper control valve closure during
normal operation

EDS 2006/Inst-170

For analog systems, controllers and manual stations should be capable of operating
at ±25% above/below the standard 0 to 100% signal. This is to ensure proper
closure during normal operation of the control valves. However in DCS systems
this number is more on the order of ±7 to ±10%. Actuator sizing and bench setting
may be more important and more precise today than in years past to ensure that the
valves seat properly when closed to prevent inadvertent leakage.

UOP Confidential - Do Not Copy 170

 DCS System Requirements

– Control System Performance

– Loop Integrity
– Data Storage and Retrieval
– Analog Outputs
„ Advanced Process Control Applications
– Feedback Control
– Summers, Multipliers, and Scaling
– Signal Selectors
– Split Range Control

EDS 2006/Inst-171

This was a brief review of the DCS system requirements. Now we will start looking
into applications of the Feedback Control loop along with some of the APC
applications.

UOP Confidential - Do Not Copy 171

 Advanced Process Control Applications
(continued)

– Cascade Control
– Ratio Control
– Feedforward Control

EDS 2006/Inst-172

We will examine some simple APC along with some of the more difficult APC
approaches. Feedforward and dynamic compensation are complex functions and
rely heavily on process modeling. However if the process model is simulated well
then their contribution greatly enhances the overall control system.

UOP Confidential - Do Not Copy 172

 Feedback Control Loop
(Controller Action)

„ Definition: Controller Action is the change in

controller output in response to an increase in process
variable above its set point
– Direct Acting Controller: For an increase in process
variable above its set point, output of the controller
increases
– Reverse Acting Controller: For an increase in process
variable above its set point, output of the controller
decreases

EDS 2006/Inst-173

Earlier when we reviewed the various equations for P, PI, & PID control, error was
defined as PV-SP. With this as the definition of error, if the PV goes above the SP
then by definition OP of the controller increases. If we apply this to a simple flow
control loop and the control valve is air to open (fail closed actuator), then for the
case of an increase in flow, the output of the controller would increase. What
happens to the control valve in this situation? The valve opens? What happens then
to the flow rate when the valve opens? The flow rate goes up. Is this what we
want? NO!!!!!!! What if the error was defined as SP- PV? Then for an increase in
PV above the SP, the error would be negative and the OP of the controller would
decrease. And if the valve is air to open, then the valve would start to close, and the
flow rate would be reduce moving towards the SP.
As a result any controller has the option of being either a direct acting or a reverse
acting controller. The evaluation of whether the controller should be direct acting
or reverse acting is done on a loop by loop basis. During configuration and
commissioning the instrument personnel will configure/set the controller action
appropriately.

UOP Confidential - Do Not Copy 173

 Feedback Control Loop
(Controller Action)

SP FIC FY FIC
F SP FY
F

FT FV FT FV
FE FE

FC FO

Element Action Element Action

FE (Process) Direct FE (Process) Direct
FT Direct FT Direct
FIC unknown FIC unknown
FY Direct FY Direct
FV Direct FV Reverse
Controller FIC needs to Controller FIC needs to
be Reverse Action be Direct Action

EDS 2006/Inst-174

In order for any feedback control loop to function properly, the loop must have
negative feedback (in terms of frequency response this negative feedback
corresponds to a 180 degree phase shift somewhere in the loop).
One can logically step through the process to determine the appropriate controller
action required for any feedback loop. For the example on the left with an air to
open valve, if something happens that causes the flow to increase above the set
point, the flow transmitter output to the controller also increases. In order for the
flow control loop to function properly, the control valve must close to reduce the
flow and bring the flow back to the controller’s set point. Therefore the output of
the controller must provide the 180 degree phase shift in the loop; when the input
signal to the controller is increasing, then the controller output is decreasing (a
reverse acting controller). This phase shift takes place in the controller as a result of
specifying a reverse acting controller.
Another approach in determining the appropriate controller action is to evaluate the
action of each individual components in the loop. Each reverse acting component
will provide a 180 degree phase shift. Therefore an odd number of reverse acting
components are required in the loop in order to provide a net 180 degree phase shift.
The example on the right has an air to close control valve (fail open valve), which is
a reverse acting component. If the controller action was set to reverse acting, the
net phase shift is not 180 degrees and the loop will not function properly.

UOP Confidential - Do Not Copy 174

 Summers, Multipliers, and Scaling

„ Develop Engineering Equations

– relevant terms must be included
– mathematics precisely defined
„ Hardware Considerations
– analog and low level digital devices
– digital control systems
„ Normalize the Variables
„ Substitute and Simplify

EDS 2006/Inst-175

Summers and Multipliers are a couple of functions frequently employed to

mathematically manipulate either input or output signals. Depending upon the task
at hand, relevant mathematical equations need to be determined, hardware selected,
and instrument signals scaled.
These type of applications are as simple as split ranging a control signal to two or
more control valves to the more complex feed forward mathematical modeling
applications and beyond.

UOP Confidential - Do Not Copy 175

 Summers, Multipliers, and Scaling
(Develop Engineering Equations)

„ Basic Flow Equations

– Liquid: F = Cop[(hw)/(Grf)]0.5
– Vapor: Q = Cop[(hw)(Pf)/(Tf)]0.5
– Summing: FT = F1 + F2
„ Exchanger Duty
– Heat Transfer: Q = (M)(Cp)(DT)
„ Process Conditions
– Pressure: Pa = Pg + 14.7
– Temperature: TR = TF + 460

EDS 2006/Inst-176

The first step is to develop the engineering equations to represent the task at hand.
As an example the basic flow equation for vapor service is shown above. If during
normal operation the pressure and/or temperature differs from the design basis, the
measured flow is in error. Obviously the greater the deviation the greater the error.
One technique is to measure both pressure and temperature and mathematically
compensate for deviation from the design conditions.
For any orifice plate in vapor service, the meter flow range, differential pressure
range, a design temperature (Tf), and a design pressure Pf was chosen from the
process data as a design basis for the manufacturing of the orifice plate. The
engineering equation is shown above and the coefficient of the orifice plate, Cop,
can be obtained from the design data supplied with the orifice plate from the
manufacturer.
Other types of equations are represented above. In some instances signal
conditioning may be as simple as converting a pressure or temperature measurement
to an absolute scale.

UOP Confidential - Do Not Copy 176

 Summers, Multipliers, and Scaling
(Hardware Considerations)

„ Analog and low level digital devices

– Scaling typically required
– Limited to simply math functions
– May require more than 1 device
– Changes require recalibration
„ Digital control systems
– Based on engineering units
– Complex math functions
– Application reconfigurable

EDS 2006/Inst-177

With the electronic analog systems, the instrument signals are typically 4 - 20 ma
signals representing 0 to 100% of the span of the instrument in Engineering Units.
Scaling of the instrument signal is required in order to manipulate these signals
mathematically. Because of the complexity of these type of systems, more than 1
device may be required; and in most applications, respanning any transmitter input
signal will require recalibration of the device(s).
With the advent of the digital control systems, a major portion of the complexity has
been removed. Typically the DCS systems convert the input signals to engineering
units; thus the scaling of the instrument signals is eliminated. Also complex
mathematical functions can be utilized in the DCS systems allowing for a wide
variety of applications.

UOP Confidential - Do Not Copy 177

 Summers, Multipliers, and Scaling
(Normalize the Variables)

„ Scaling Instrument Signals

– E = Span(En) + Zero
• E = Engineering Units
• Span = Upper Value - Lower Value
• En = Instrument Signal (0 to 1.0 or 100%)
• Zero = Lower Value
„ Process Variable Examples
– TF = 100(Tn) + 50
– Pg = 50(Pn) + 0 = 50(Pn)
– Gr = 0.5(Grn) + 0.5

EDS 2006/Inst-178

For a given pressure transmitter that has been calibrated for 0 to 50 psig, the
instrument signal (0 to 100%) represents 0 to 50 psig. The zero of the instrument is
the Lower Value; and in this case, Lower Value is equal to 0. The span of the
instrument is equal to the Upper Value minus the Lower Value. In this case the
Upper Value is 50. Therefore the span is 50 minus 0, which is 50. If the pressure in
Engineering Units is 25 psig, then the instrument signal in terms of % would be
50% (0.5). Mathematically then a signal of 50% would represent 25 psig in
Engineering units based on this equation.
Scaling instrument signals is typically the first step in conditioning these signals for
the various applications.

UOP Confidential - Do Not Copy 178

 Summers, Multipliers, and Scaling
(Summing 2 Flows)

Minimum Flow Spillback

Design Considerations:
FIC
A FICA: 0 - 100 gpm
FIB: 0 - 40 gpm
FT Minimum Flow: 30 gpm
Engineering Equation:
FICT = FICA + FIB
FIC SUM
T Scaled Equations:
FICA = 100FICAn + 0
FI n
B FIB = 40FIB + 0
FICT = (100 + 40)FICTn + 0
FT

IST-R00-193A

EDS 2006/Inst-179

Looking at the summer in the above example the engineering equation used is the
sum of the two flow meters. The first step is to scale each flow measurement based
on the design conditions of each meter. Once each flow meter scaled equation is
determined, the next step is to substitute into the engineering equation and simplify.

UOP Confidential - Do Not Copy 179

 Summers, Multipliers, and Scaling
(Summing 2 Flows cont’d)

„ Engineering Equation
• FICT = FICA + FIB
„ Substitute Scaled Equations
• (140)FICTn = (100)FICAn + (40)FIBn
„ Simplify to match hardware
• FICTn = (0.713)FICAn + (0.287)FIBn
„ Analog Summer (2 inputs/1 output)
• Output = k1(Input1) + k2(Input2) + Bias

Output is a normalized signal ranging from 0 to 1.0;

span has been chosen to be 0 - 140 gpm

EDS 2006/Inst-180

After the equations have been scaled and simplified, specific hardware is selected to
accomplish the task. In the above example an analog summer with w in puts and 1
output was selected to sum the two flow signals. If both signals are at 50%, then the
total flow would be 50 plus 20 or 70 gpm. Analyzing the analog summer equation
results in 0.713 x 0.5 + 0.287 x 0.5 + 0. This simplifies to 0.3565 + 0.1435 + 0 =
0.5. The output of the summer is 50%. The span is 140; therefore the analog
summer also represents 70 gpm.

UOP Confidential - Do Not Copy 180

 Summers, Multipliers, and Scaling
(Summing 2 Flows cont’d)

„ Digital Control Systems (no scaling)

• FICT = FICA + FIB
„ Simplify to match function block
„ Summer math function
• Output = k1(Input1) + k2(Input2) + Bias

If the digital control system performs scaling and inputs are converted
to engineering units, then k1 = k2 = 1.0 and Bias = 0 with the output
being in the appropriate engineering units.

EDS 2006/Inst-181

In a DCS system the input signals are converted to Engineering units. The function
block equation is provided above and the constants are set to 1. Therefore the
summer of the two is quite simple compared to the analog device previously
illustrated.
If one or both of these transmitters are respanned to different engineering units, the
analog device equation will change and recalibration of the device is mandatory.
For the DCS system the equation stays the same. A simple change in the flow
measurement span is required, but can be accomplished in a matter of seconds by
the control engineer. Since the DCS is using Engineering Units a minimal amount
of work is required to update the configuration of the loop in question.

UOP Confidential - Do Not Copy 181

 Summers, Multipliers, and Scaling
(Gas Flow Compensation)
Temperature and Pressure Compensation
Gas Flow Computation

Flow

Orifice Plate Design

Design Considerations: Q = 278 hw P f
d/p Cell Pressure Temp Tf
Pg: 0 - 50 psig Transmitter Transmitter Transmitter
Tf: 40 - 140 deg F hw Pf Tf
hw: 0 - 100 inch h2o INPUT3

Engineering Equation: INPUT1 Q

n

2 h P 2
Q = (77284) w f (Qn )
Tf Analog Square Root
Scaled Equations: Device Extractor
n INPUT2
Pa = 50P + 14.7
n
TR = 100T + 500
n
hw = 100 hw + 0
n
Q = (Span)Q + 0 IST-R00-193B

EDS 2006/Inst-182

Gas flow compensation is common within many process streams coming into or
leaving a particular process unit as a means of material balance closure. Based on
the engineering equation for vapor flow through an orifice plate, the relationship
exists as shown previously. Assuming that the molecular weight of the gas is
unchanged for this example, the engineering equation has three inputs: the
differential pressure across the orifice plate, the flowing pressure and the flow
temperature. Again the first step is to develop the scaled equations for the
instrument signals, followed with substitution and simplification into the
engineering equation. Due to the complexity of this task, two analog devices are
required to complete the flow compensation for changes in operating temperature
and/or pressure.

UOP Confidential - Do Not Copy 182

 Summers, Multipliers, and Scaling
(Gas Flow Compensation cont’d)

„ Engineering Equation
• Q = Cop[(hw)(Pf)/(Tf)]0.5 where Cop = 278
„ Determine flow meter span
• Qmax = 278[ (100)(64.7)/(500)]0.5 = 1000 scfh
• Q = 1000Qn + 0
„ Substitute Scaled Equations
• 1000Qn = 278[(100hwn)(50Pn + 14.7)/(100Tn + 500)]0.5
• (Qn)2 = 0.077284[(100hwn)(50Pn + 14.7)/(100Tn + 500)]
• (Qn)2 = 7.7284[(hwn)(50Pn + 14.7)/(100Tn + 500)]

EDS 2006/Inst-183

The maximum flow within the range of pressure and temperature measurements
will occur when the pressure is at its 100% point (64.7 psia) and when the
temperature is at is 0% point (500 °R). This value is then used to set up the scaled
equation for the gas flow rate. Also note that the coefficient of the orifice plate was
given from the manufacturer as a value of 278.

UOP Confidential - Do Not Copy 183

 Summers, Multipliers, and Scaling
(Gas Flow Compensation cont’d)

„ Simplify to match hardware

• (Qn)2 = 7.7284[(hw)(50Pn + 14.7)/(100Tn + 500)]
– Input signals must be factored not to exceed 1.0
• (50Pn + 14.7) = 64.7[(50/64.7)(Pn) + (14.7/64.7)]
• (50Pn + 14.7) = 64.7[0.77(Pn) + 0.23]
• (100Tn + 500) = 600[0.17(Tn)+ 0.83]
• (Qn)2 = 0.83(hwn)[(0.77Pn + 0.23)/(0.17Tn + 0.83)]
„ Analog Device (3 inputs/1 output)
• Output =[k1Input1+b1][k2Input2+b2]/ [k3Input3+b3]
• Square Root Extractor yields flow signal
Output from analog device is proportional to flow2; Output from Square Root
Extractor is a normalized flow signal (0 - 1.0) with a span of 0 - 1000 SCFH.

EDS 2006/Inst-184

As shown in the previous diagram the first analog device has 3 inputs and 1 output
with the mathematical function as indicated above. The differential pressure
measurement becomes Input 1, the pressure measurement becomes input 2, and the
temperature measurement becomes input 3. The output of the device is proportional
to the flow squared. The second device is a square root extractor, which effectively
has one input and 1 output. The output of the square root extractor is a normalized
flow signal representing 0 to 100% of the flow signal. As we determined earlier in
terms of Engineering Units, 0 to 100% represents 0 to 1000SCFH.
Again in the analog devices if any of the instruments spans are changed, then the
first analog device would have to be recalibrated also. However in a DCS system,
where the input signals are converted to Engineering Units, the changes would be
minimal.

UOP Confidential - Do Not Copy 184

 Signal Selectors

„ Compares 2 or more inputs signals, preferentially

selecting one of the signals as the actual output
signal
„ Types of Signal Selection
– Low Signal Selector
– High Signal Selector
„ Applications
– Process Control Constraints
– Min/Max Limit Stops

EDS 2006/Inst-185

Signal selectors are a means of overriding control signals. These selectors compare
2 or more inputs and preferentially selects one of the control signals for use in the
control loop while ignoring the other(s) control signal.
A couple of the more common applications include process constraint and limit
stop. We shall look at how the signal selectors are applied.

UOP Confidential - Do Not Copy 185

 Signal Selectors
(Process Constraint)
Pressure Override Control
Design Considerations:
T1, P1
Clay Treater operates at:
Inlet - T1 and P1
Outlet - T1 and P2
Clay Treater dPSOR = P1 - P2
(Liquid Phase)
EOR Considerations:
SP dPEOR = P1 - P2 > dPSOR
P
T1, P2
(P2 is decreasing with time)
PT PIC
Controllers:
Pressure Flow
SP
F
FIC FY
<
SPP > PV SPF = Normal (100%)
Direct Reverse
FT

IST-R00-195A
FC

EDS 2006/Inst-186

In the example above the liquid has been heated to T1 and has a normal operating
pressure of P1 at the discharge of the pump for a flow at 100%. The clay treater is
used to extract water from the incoming feed. The clay treater has been designed
with a maximum differential pressure at end of run conditions. At this point
Operations can plan a turn-a-round to replace or regenerate the clay material as
needed. If P2 drops below the vapor pressure of the fluid, the clay material in the
clay treater will be damaged if vaporization occurs. We want to prevent this from
happening at all cost.
Under normal operation P2 is well above the vapor pressure of the fluid. As time
progresses the delta P across the bed rises and P2 decreases approaching the vapor
pressure of the fluid. Feed to the downstream unit is on flow control. Given that
the control valve is an air to open control valve, we see that the controller action
needs to be reverse acting. In the event that the pressure at the outlet of the clay
treater approaches the anticipated vapor pressure of the fluid (plus some margin),
the operator could place the flow controller in manual and manually decrease the
flow keeping P2 arbitrarily above the vapor pressure.
Alternatively an override process constraint control can be added with the use of a
low signal selector. A pressure transmitter is required on the outlet of the clay
treater and a pressure controller is required. The low signal selector compares the
output of each controller and passes the lower of the two signals to the control
valve. The PIC controller action has to be direct acting.

UOP Confidential - Do Not Copy 186

 Signal Selectors
(Minimum Limit Stop)
Minimum Burner Pressure Limit

Fired Heater Design Considerations:

Process Temperature Controller
is cascaded with Burner Pressure
Controller.
Process Outlet TIC output is PIC set point.
Temperature TT
HIC output is minimum burner
Minimum pressure limit for safe operation.
Burner HIC HY TIC Process
>
Pressure Controllers:
PIC PT Temperature Pressure
Burner Reverse Reverse
Pressure

FC
Fuel Gas
Supply IST-R00-195B

EDS 2006/Inst-187

The above control scheme has the primary temperature controller adjusting the fuel
gas burner pressure to vary the heat release in the fired heater. If the process
requires more heat release, the output of the temperature controller increases the
burner pressure controller’s set point. If the process requires less heat release, the
temperature controller decrease the set point.
During turndown operation the possibility exists of lowering the burner pressure to
a point where there is flame-out at the burner tips. Needless to say this is a
dangerous scenario and must be avoided. A high signal selector can be introduced
into the system to limit just how low the burner pressure set point can be.
Once this constraint is reached and the master temperature controller still request a
reduction in burner pressure, the operator can block in burners on a multi-burner
heater. By doing so the burner pressure required for less burners in service will
require a higher burner pressure and should eventually increase the pressure above
the minimum.

UOP Confidential - Do Not Copy 187

 Split Range Control

„ Output from Controller manipulates 2 or more

control valves independently to control the process
„ Control Element Diagram depicts operation of
control valves over the full range of controller
output
„ Multipliers are incorporated to scale the output
signal from 0 - 100% of the control valve stroke

EDS 2006/Inst-188

In some applications, the process controller can be used to control 2 or more control
valves. This application is termed “split-ranging” the control valves. A control
element diagram (CED) can be constructed to depict the operation of the controller
and manipulation of the control valves (final control elements).
Mathematical multipliers are utilized to scale the output signal over the full range of
the control valves.

UOP Confidential - Do Not Copy 188

 Split Range Control
(Control Element Diagrams)
O
Valve Position O

Valve Position
System II
One valve open;
one valve throttling. A
III B
Example:
A B
Exchanger 2-Valve
C Diverting
0 50 100 C
0 50 100
Controller Output O
Controller Output
meas

Valve Position
A B meas

System I System III

One valve closed; II Two valves in
one valve throttling. series.
Example: Example:
Push-Pull Gas C Vent Gas to
0 50 100
Blanketing Fuel/Relief Header
Controller Output
meas IST-R00-195D

EDS 2006/Inst-189

For system I, such as a push-pull gas blanketing system on a feed surge drum, both
valves are closed when the controller output signal is 50%. When the controller
output signal is greater than 50%, one valve (valve A) remains closed, and the
second valve (valve B) is used to throttle one process stream. Likewise when the
controller output signal is less than 50% just the opposite scenario exists, valve
(valve B) remains closed, and the other valve (valve A) is used to throttle a second
process stream.
For system II, such as an exchanger 2-valve diverting system, both valves are open
when the controller output signal is 50%. As the controller output increases above
50%, valve B closes while valve A remains open. Just the opposite occurs when the
controller decreases below 50%.
For system III, such as vent gas relief system, the valve are operated in series. As
shown in the CED diagram, when the controller output is 0%, both valves are in the
closed position. As the controller output increases from 0 to 50%, valve A moves
from the closed position to the wide-open position while valve B remains closed.
From 50 to 100% valve A remains wide-open and valve B is being manipulated
from the closed position to its wide-open position.
Other systems exist, some with three valves, but these three examples cover the
majority of the cases.

UOP Confidential - Do Not Copy 189

 Split Range Control
(Push-Pull Gas Blanketing)
Design Considerations: Split Range
X X

1. High Press - Vent PY PY

2. Low Press - Make-up
0 to 50% in 50 to 100% in
3. PIC - Direct Action
100 to 0% out 0 to 100% out
3. A& B Valves - FC PIC
4. CED - Valves closed
at 50 % Output Direct

Valve A PT Valve B
Control Element Diagram
Gas Supply Vent
O
FC FC
Valve Position

Gas Blanketed Vessel

A B

C
0 50 100 Liquid In Liquid Out
Controller Output
IST-R00-195E
meas

EDS 2006/Inst-190

This is an example of the Push-Pull Gas Blanketing System. Notice in this case as
shown both valves are air to open (fail closed valves). Utilizing the CED for system
I, both valves are closed at a controller output equal to 50%. If the pressure in the
drum increases above the set point, then the vent valve must be manipulated in
order to bring the pressure in the drum back down to the set point. We can
accomplish this by making the controller action direct and establishing the fact that
valve B will be the vent valve on the CED. Therefore the multiplier has an input of
50% to 100% with a corresponding output of 0 to 100%. Notice that this multiplier
is direct acting, i.e. if the input increase then the output increases also, but at a ratio
(slope) of 2 to 1.
With the controller action set as direct acting, if the pressure decreases then the
output of the controller also decreases. Therefore we need to open the gas supply
valve(while the vent valve is closed) to bring the pressure back up to its set point.
However the gas supply valve is also an air to open valve. When the controller
output is 50% this valve should also be in the closed position. In order to
accomplish this task, the multiplier must be reverse acting. In other words when the
controller output is 50%, the output of the multiplier needs to be 0% in order for the
valve to be in the closed position. Then as the output of the controller decreases
below 50%, the signal to the gas supply valve increases towards 100%. The ratio
(slope discussed earlier) is -2 to 1.
What would happen if the controller action was set as reverse acting?

UOP Confidential - Do Not Copy 190

 Split Range Control
(Exchanger 2-Valve Diverting)

Design Considerations:

1. Fail - Extract Heat

2. High Temp - Bypass
50 to 100% in
3. Low Temp - Series 0 to 100% out
X

4. TIC - Reverse Action TY FO

5. CED - Valves open A
at 50 % Output
X
0 to 50% in
TY 0 to 100% out
Control Element Diagram
O
B
Valve Position

FIC FC

B A FT
TIC
Hot 150 oF
Reverse
C Cold 100 Fo
0 50 100
Controller Output
IST-R00-195F
meas

EDS 2006/Inst-191

In the above Exchanger 2-Valve Diverting System, one valve is in series and one
valve bypasses the heat exchanger. The failure mode of these two valves is
contingent upon the process. In this case the process is heat extraction; therefore,
the series and bypass valves are designed as air to close and air to open,
respectively. Utilizing the CED for system II, both valves are open at a controller
output equal to 50%. If the temperature of the controller increases above the set
point, more heat must be extracted by forcing additional process fluid through the
exchanger. This is accomplished by throttling the bypass valve with the series valve
wide open.
Looking at the bypass valve (this is a fail close valve) with a controller output at
50%, the valve should be wide-open with a 100% air signal. If the controller is
setup as a reverse acting controller, then the bypass valve becomes the B valve on
the CED. Therefore the multiplier has an input of 0 to 50% with a corresponding
output of 0 to 100% (slope 2 to 1).
If the temperature of the controller decreases below the set point, too much heat has
been removed by the heat exchanger. Therefore less fluid should pass through the
heat exchanger. This is accomplished by throttling the series valve while the bypass
valve is wide-open. At a 50% controller output signal the A valve (series valve)
should be wide open. This is accomplished with the output of the multiplier at 0%
to a fail open valve. Therefore the multiplier has an input of 50 to 100% with an
output of 0 to 100%.

UOP Confidential - Do Not Copy 191

 Cascade Control

„ Consists of a primary controller and a secondary

controller
„ Output of the primary controller is the set point of
the secondary controller
„ Response of the secondary loop must be much more
rapid than the primary loop
„ Minimizes effects of load variables in the secondary
loop before these variables influence the primary
loop

EDS 2006/Inst-192

In cascade control the output of one controller is used to manipulate the set point of
another controller. The two controllers are then said to be cascaded, one upon the
other. Each controller has its own process variable input, but only the primary
controller has an independent set point and only the secondary controller has an
output to the process. The manipulated variable, the secondary controller, an its
process variable constitute a closed loop within the primary loop.
The control valve positioner was sited as a secondary loop with the manipulation of
the final control element. Temperature, pressure , and level are often cascaded with
flow loops to minimize load variables such as line pressure changes. The output of
the primary controller is essentially a manipulation of the flow of mass or energy by
the primary controller. Cascade control is of great value where high performance is
mandatory in the face of random disturbances.

UOP Confidential - Do Not Copy 192

 Cascade Control
(Temperature/Flow)

Basic Cascade
Temperature Temperature to
Loop Flow
TT TIC TT TIC Primary

Process R Process R

Steam Steam, P1
FIC Secondary
R
FT

Condensate P2
FC FC
Condensate
Design Considerations:

1. Basic Temperature Loop will adjust condensate flow for deviation in process temperature
2. Load Variables P1 and P2 influence the pressure drop (flow) across the control valve
IST-ROO-197A
3. Cascading TIC to FIC minimizes effect of variation in the Load Variables
3. Output from TIC is Set Point for FIC
4. FIC configured for Local/Remote Set Point capability

EDS 2006/Inst-193

Steam is often used throughout the Process Industry as a heating medium. As

shown in the Basic Temperature Loop, the process outlet temperature measurement
is the PV to the temperature controller. The manipulated variable is the condensate
flow rate through the exchanger. As the flow of condensate varies, the amount of
‘condensate backup’ in the exchanger varies exposing more/less tube heat transfer
area. The output of the controller is a function of temperature. If the PV is low
more heat is required opening the control valve and visa versa. However what
effect does an increase in the steam header pressure have on the process
temperature?
If the steam or condensate header pressures vary, then the driving force (valve
differential pressure) across the control valve varies; and as a result, the flow of
condensate across the valve will change as the differential pressure fluctuates.
Therefore the heat transfer surface area is varying the amount of heat transfer and
the process outlet temperature may not be controlled very well.
In the cascade loop a flow measurement and flow controller have been added to
control the flow of condensate from the exchanger. The output of the primary
controller, now the set point to the secondary controller, is a function of condensate
flow instead of process outlet temperature. At steady state the temperature
controller requires ‘X’ gpm of condensate. If the condensate rate changes, due to a
header pressure disturbance, the secondary controller will immediately throttle the
valve to maintain the flow of condensate at ‘X’ gpm.

UOP Confidential - Do Not Copy 193

 Ratio Control

„ Manipulates material balance to control the ratio of

one variable to another to satisfy a defined objective
„ Ratio Flow Control - Multiplier Function
– Input Variable: Flow Measurement (X)
– Gain: Ratio of Flows - Y/X
– Output = Input times Gain = (X)(Y/X) = Y
„ Output is required flow of stream Y and is cascaded
as set point to flow controller Y

EDS 2006/Inst-194

Ratio control is a feed forward system wherein one process variable is manipulated
in a predetermined ratio to another process variable to satisfy some higher-level
objective. In blending systems the higher-level objective may be composition,
Octane Number, or whatever; but this higher-level objective cannot always be
measured, in which case the real objective cannot be controlled by a feedback loop.
The real controlled variable in a ratio control system is the ratio of two measured
process variables. The typical system manipulates one valve controlling one of the
process variables, while the other is a “wild” uncontrolled process variable. As the
uncontrolled variable changes, then the system will manipulate the controlled
variable in direct proportion as dictated by the predetermined ratio.

UOP Confidential - Do Not Copy 194

 Ratio Control
(Blending Application)

Design Considerations:
LIC
LT
Flow FIX varies as level changes
Maintain constant blend of Streams X & Y
Ratio Control Station HIC adjustable Y/X
STREAM X Multiplier FY: Input X times Y/X = Y
HIC FT
RATIO
X Set Point FICY is Multiplier FY output
X (FICY is configured for local/remote SP)
FY FI
X
As Stream X flowrate changes the ratio
FIC between streams X and Y is maintained
Y
by adjusting the set point to Stream Y’s
flow controller
FT
Y
STREAM Y IST-R00-197B
To Blending
Tank

EDS 2006/Inst-195

The “wild” process variable is stream X. Stream X’s flow rate is contingent upon
the level in the drum. The objective of the control scheme is to blend stream X and
stream Y in a pre-determined ratio of Y/X. This ratio then is the gain of the FY
multiplier block.
The “wild” flow is measured and the PV is an input to the FY multiplier block.
With a gain of “Y/X” times the PV “X”, the output of the multiplier block is the
required set point of stream Y to maintain the ratio of stream Y to stream X.
As the level varies in the drum the level controller will adjust the outflow from the
drum (stream X), and the control scheme will continuously control stream Y in
direct proportion to the changes induced in stream X by the level controller.

UOP Confidential - Do Not Copy 195

 Ratio Control
(Forced Draft Fired Heater)

„ Forced Draft System – air flow to heater is controlled

variable
„ Transient states are always fuel lean/air rich
„ For an increase in fuel demand, control system will
first increase air flow followed by an increase in fuel
supply
„ For a decrease in fuel demand, control system will
first decrease fuel supply followed by a decrease in
air flow

EDS 2006/Inst-196

One way to improve the overall efficiency of a fired heater is to design the heater
with a forced draft system where the air to the heater is preheated by the hot flue gas
exiting the heater. In order to accomplish this air preheat the air flow to the heater
is a controlled variable.
As a result of this added complexity, the control system for a natural draft fired
heater has been modified, to prevent starving the fire box of oxygen. The control
scheme will essentially insure that excess air is available and that transients states
are always fuel lean and air rich.

UOP Confidential - Do Not Copy 196

 Ratio Control
(Air/Fuel Ratio)

„ Air/Fuel stoichiometric quantities

– Methane - CH4 + 2O2 = 2H2O + CO2
– Butane - C4H10 + 6.5O2 = 5H2O + 4CO2
„ Fired Heater designed for 15% excess air
„ RatioWT = [(1.15MO2/0.21)MWAIR]/[(MFUEL)MWFUEL]
– FUEL AIR/FUEL RatioWT
• Methane 19.9
• Ethane 18.5
• Propane 18.0
• Butane 17.8

EDS 2006/Inst-197

Ratio control is implemented in the control scheme on a weight bases of Air to Fuel.
One common fuel is refinery off-gases from the various process units. These fuel
gases are mixtures of anything from hydrogen to C4’s and C5’s. The above
examples illustrate the stoichiometric equations for complete combustion of a
particular fuel. Based on 15% excess air, the air to fuel ratio varies between 17 and
20 for typical refinery fuel gas. The heavier the fuel the lower the ratio becomes.
For liquid fuels this ratio approaches 17.

UOP Confidential - Do Not Copy 197

 Ratio Control
(Fired Heater Fuel/Air Cross-Limiting Control System)
Heater Outlet

Combustion
m = Mass TIC Heater Outlet
Air
PV = Process Variable Controller
SP = Remote Setpoint

m Gas/hr
< = Low Signal Selector
> = High Signal Selector
X = Multiply FI
FT
÷ = Divide

<

Fuel Gas

m Gas/hr
m Gas/hr

m Gas/hr
Temperature
÷ Compensation
m air/hr

>
FI m Gas/hr
FT

m air/ m Gas
Air/Fuel
TI
Ratio

m air/hr
m Gas/hr HIC

SP PV
>

Damper
PV FIC SP FIC
m Gas/hr m air/hr

PT PIC
IST-ROO-197C

EDS 2006/Inst-198

With experience the operator will be able to observe the burner flame pattern and
for a typical fuel gas adjust the air to fuel ratio to obtain optimum performance in
the fired heater. The above control scheme incorporates a ratio control station with
an adjustable air to fuel ratio between 17 and 20, a multiplier block, a divider block,
and high/low signal selectors. The engineering units for both the fuel gas and air
flow rates are determined on a mass basis.
The primary controller is the TIC on the heater outlet. If the process temperature
decreases below the set point, then the TIC will demand an increase in fuel firing.
However this demand first passes through a low signal selector. This selector
compares the calculated fuel rate at steady state (based on current air flow) with the
transient demand fuel rate (based on the TIC output). Initially the low signal
selector ignores the increase in demand from the primary controller. The demand
signal from the TIC is also an input to a high signal selector. This selector
compares the transient demand fuel rate with the actual fuel consumption at steady
state (based on actual fuel as measured). The high signal selector passes the fuel
demand signal onto the multiplier which in turn increases the air set point.
As illustrated a demand in fuel firing will first increase air and as the air rate
increases the fuel rate will follow. Likewise on a decrease in demand from the the
TIC, fuel will be reduced first followed by a decrease in the air rate.

UOP Confidential - Do Not Copy 198

 Feedforward Control

„ Based on an inferential or empirical process relationships,

the manipulated variable is adjusted directly as load
variables change
„ Basic control loop components
– Feedforward model
• steady state model of the process using material and
energy balance equations and process measurements
• model predicts changes in manipulated variable as a
function of load variable changes
• set point adjustments based on model prediction

EDS 2006/Inst-199

Processes which cannot be controlled well because of their difficult nature are very
susceptible to disturbances from load or set-point changes. A means of solving this
type of control problem directly is called feedforward control. The principal factors
affecting the process are measured and are used to determine the appropriate change
to meet current conditions. One important feature is that the controlled variable is
not used by the system. The controlled variable is used as a feedback trim to
account for inaccuracies in the model.
When a disturbance is initiated, control action starts immediately to compensate for
the disturbance before it changes the controlled variable. Feedforward action
theoretically is capable of exact control, its performance being limited only by the
accuracy of the measurements and engineering calculations.
In general the feedforward control system continuously balances mass and/or
energy delivered to the process against the demands of the load variables. As a
result the engineering calculations made by the control system are mass and energy
balances around the process, and the manipulated variables must be accurately
regulated flows. Therefore the feedforward system will predict transient state
responses to the manipulated variables and make these adjustments (as SP changes)
before ever seeing any changes in the controlled variable. If the calculations are in
error then the predicted state will be in error and a deviation will exist. The better
the model, the better the prediction.

UOP Confidential - Do Not Copy 199

 Feedforward Control
(continued)

„ Basic control loop components

– Feedback (trim) controller
• corrects for errors in the steady state model and for
errors in process measurements
• steady state model usually based on major load
variables and does not account for all load variables
– Controlled variable controller
• based on steady state model prediction and
correction from feedback trim controller,
adjustments to controlled variable set point are
implemented

EDS 2006/Inst-200

The feedback trim controller is an integral part of the feed forward model. It’s
contribution if done properly is minor but will eliminate any offset. The feedback
trim controller will correct for errors in the steady state prediction and for any
inherent error in the process measurements. Also most feedforward systems will
incorporate only the major load variables. Therefore changes in some of the minor
load variables will not be accounted for in the feedforward scheme and the steady
state model will not be exact.
Combining the steady state model prediction with the feedback trim controller,
adjustment are made to the set point of the controlled variable. The feedback trim
controller should be slow responding and makeup only the difference that the model
has not accounted for.

UOP Confidential - Do Not Copy 200

 Feedforward Control
(continued)

TT TIC Primary
Process Feedforward Model
Wp,Tout R

HEATin = HEATout
Steam, P1 W cond(Hvap) = W p(Cp)(Tout-Tin)
FIC Secondary
W cond = W p(Cp/Hvap)(Tout-Tin)
R
Let K = Cp/Hvap
FT
and T*out = desired temp
Wp,Tin P2 then
W cond = K(W p)(T*out-Tin)
FC
Condensate

Design Considerations:

1. Cascade control improved response of the control system for changes in steam conditions.
2. What happens for major Load Variables on process side? Flowrate or inlet temperature?
3. Steady state model represented by:
Heatin = W cond(Hvap) and Heatout = W p(Cp)(Tout-Tin)
where Hvap = Heat of Vaporization for steam
and Cp = Heat Capacity of Process Material
4. Additional Instruments required to measure process flow and inlet temperature. IST-ROO-197D

EDS 2006/Inst-201

We implemented the cascade control system above to compensate for load changes
on the heating medium side of the exchanger. But what about the load variables on
the process side? What happens to the process stream outlet temperature if the
process flow rate changes, or the process inlet temperature changes, or stream
composition changes? In order to implement the feedforward control system, one
needs to identify the major load variables and then implement process measurement
of those load variables. In the above example it was identified that the process flow
rate and temperature were the major load variables. Therefore additional
instrumentation is required to measure both the process flow rate and the process
inlet temperature.
The next step is to develop the engineering equations to represent the heat input at
steady state. On the steam side, the heat of vaporization needs to be an assumed
value as well as the heat capacity of the process fluid. Knowing the heat capacity of
the fluid, the process flow rate and the expected temperature rise across the heat
exchanger, the anticipated duty can be calculated. Likewise knowing the steam heat
of vaporization and this anticipated exchanger duty, the quantity of steam can be
calculated. Therefore an orifice plate and temperature element are required on the
inlet side of the shell in order to measure these process variables.

UOP Confidential - Do Not Copy 201

 Feedforward Control
(continued)

T*out - Tin Wcond = K(Wp)(T*out - Tin)

Feedforward Model
X
+ T*out
UY TY TIC TT HEATin = HEATout
Tout
R
W cond(Hvap) = W p(Cp)(Tout-Tin)
Wp - Tin W cond = W p(Cp/Hvap)(Tout-Tin)
Process Let K = Cp/Hvap
Steam, P1
FIC and T*out = desired temp
then
R
W cond = K(W p)(T*out-Tin)
FT TT FT

P2
FC
Condensate
Design Considerations:

1. FIC represents the controlled variable controller with remote set point.
2. With appropriate Eng Units, TY and UY represent the feedforward model at steady state.
3. TIC represents the feedback (trim) controller, compensating for deviations from set point.
4. The required condensate flow is calculated to compensate for changes in the major load
variables of process flow and inlet temperature.
5. For inefficiencies in the feedforward model, process measurements, etc, the feedback
(trim) controller corrects the desired outlet temperature. IST-ROO-197E

EDS 2006/Inst-202

With the implementation of the process flow rate and temperature measurements,
the exchanger duty can be predicted on the process side and the quantity of steam
required can be determined with the assumed heat of vaporization. This value then
is the steady state prediction and becomes the set point to the controlled variable
controller. However due to temperature or flow measurement errors or errors in the
assumed heat of vaporization or heat capacity, the steady state prediction may be in
error. Thus the feedback trim controller (TIC) is used to tweak the process and
compensate for this error. The output of the feedforward calculation is the set point
to the flow controller (controlled variable).

UOP Confidential - Do Not Copy 202

 Advanced Process Control Applications

– Feedback Control
– Summers, Multipliers, and Scaling
– Signal Selectors
– Split Range Control
– Cascade Control
– Ratio Control
– Feedforward Control
„ Distillation Controls

EDS 2006/Inst-203

This concludes our presentation on advanced process control. This is only a scratch
on the surface, but it should provide some insight into the field of Advanced Process
Control and how the computer can be utilized for difficult to control processes.

Next we will look at some of the more common distillation control schemes.

UOP Confidential - Do Not Copy 203

 Distillation Controls
(continued)

– Strategies for constant Heat Input Control

– Review of Gibbs Phase Rule
– Design basis for Column Pressure Control
– Alternate designs for Composition Control
– Recommended Material Balance Control

EDS 2006/Inst-204

Initially we will look at a couple of control schemes around the bottom of the
column. UOP philosophy centers around constant heat input, but cascaded control
schemes or dual composition control schemes are also a viable alternative. Next we
will look at Gibbs Phase Rule and apply this principle to multi-component
distillation.
After review of the Gibbs Phase Rule we will see that pressure control is one
independent parameter that we will fix in order to achieve the desired split. We will
also review two alternative control schemes for composition control based on
temperature control in the rectifying section of the distillation column. We will also
see some of the selection criteria for picking one control scheme over the other.

UOP Confidential - Do Not Copy 204

 Distillation Controls – Heat Input
(Fired Reboiler TIC Controller)

TIC

Distillation
TT
Column

Fired
Heater

FIC
PT

FT PIC

Column To other Fuel

Bottoms heater passes Supply

EDS 2006/Inst-205

For the most part UOP will design heat input control schemes for constant heat
input. The control scheme listed above is for a multi-pass fired heater. Each
individual heater pass will have its own flow control loop to balance the flows
through each heater pass. Therefore the process load to the fired heater is constant.
The primary controller, TIC control on the combined heater outlet, is cascaded to
the secondary controller, PIC burner pressure controller. The primary controller
will adjust the burner pressure up or down to satisfy the temperature set point of the
controller.
During normal operation if the column bottom material contains too much of the
light key, then the operator will increase the temperatureset point a couple of
degrees. This set point change would increase the pressure set point on the PIC,
thus increasing the heater firing rate.

UOP Confidential - Do Not Copy 205

 Distillation Controls – Heat Input
(Boiling Point Curve)

Temp Change
Slope =
% Vapor Change

Narrow Boiling Pt Curve < 0.1

Wide Boiling Pt Curve > 0.1

0 50 100
Vaporization, %

EDS 2006/Inst-206

What would happen to the temperature profile through the fired heater if the column
bottoms material was a pure component or a narrow boiling point material? For a
binary distillation, such as a benzene-toluene column in an aromatics complex, the
bottom of the column is almost pure toluene and the temperature curve is nearly a
fixed temperature as determined by the operating pressure of the column. For a
narrow boiling point material, such as the xylene isomers (boiling point range of
277 to 291 °F), the Temperature vs. % Vaporization curve would be a very shallow
(flat) curve. If the temperature curve is a relatively flat curve in the design range of
the fired heater, then TIC control can not be implemented
For UOP designs, the typical fired heater will be sized to provide somewhere
between 40 to 60% vaporization at the heater outlet. Therefore UOP will look at the
slope of the boiling point curve in this range. If the slope is greater the 0.1 °C per %
vaporization, then TIC control on the heater outlet can be implemented. However if
the slope is less than 0.1 °C per % vaporization, then TIC control will not work.
Some other alternative must be implemented in order to control the heat input to the
distillation column.

UOP Confidential - Do Not Copy 206

 Distillation Controls – Heat Input
(Fired Reboiler PDIC Controller)

PDIC

Distillation
PDT
Column

Fired
Heater

FIC
PT

FT PIC

Column To other
Fuel
Bottoms heater passes Supply

EDS 2006/Inst-207

In the above control scheme the primary TIC control loop has been replaced with a
PDIC control loop. This is a pressure differential controller with a pressure
differential transmitter (PDT) and pressure differential element (PDE). The PDE is
an eccentric orifice plate that must be installed in a horizontal line in order to allow
for the passage of the two phase fluid through the orifice plate. An eccentric plate is
chosen to ensure that the liquid passes through with the least amount of effort and
minimal hold-up.
The theory behind this application involves the relative pressure differential
between liquid and vapors generated across the orifice plate. For a fixed liquid flow
to the heater (notice that we still have constant heater pass flow rates), the PDIC
(ranged 0 to 100% heat input) is a measure of the vapor rate generation in the fired
heater. In other words the pressure differential is a measure of the heat input to the
distillation column (% vaporization).
During normal operation if the column bottom material contains too much of the
light key, then the operator will increase the set point say from 50 to 55%. Likewise
this set point change would increase the pressure set point on the PIC, thus
increasing the heater firing rate.

UOP Confidential - Do Not Copy 207

 Distillation Controls – Heat Input
(Steam Reboiler Condensate Control)
Design Considerations:
1. Static head required for flow measurement
2. Exchanger designed for minimum steam pressure
3. Changes in flow increases/decreases condensate
level in exchanger exposing less/more surface area

Steam

FIC

5'-0"
Net Min. FT
Bottoms Condensate
Product
Min.
Grade
Locate orifice flanges and control valve
IST-R00-198A
assembly in horizontal run from reboiler

EDS 2006/Inst-208

Steam is often used in conjunction with a thermo-siphon reboiler as the heating

medium. Two potential control schemes exist to control constant heat input from
the reboiler. The first control scheme (shown above) uses a flow control loop on the
condensate from the reboiler.
Inside the reboiler a vapor/liquid equilibrium surface layer is established at steady
state. The liquid at this interface is at its bubble point as well as the vapor is at its
dew point. Care must be taken in the design of the flow element in order to ensure
that flashing does not occur across the orifice plate; otherwise, the flow
measurement is meaningless. UOP will require a minimum static head upstream of
the orifice plate along with using a maximum pressure differential span of 50 inches
water column or less. For low or medium pressure steam headers the minimum
head requirement is 5 feet.
During normal operation if the column bottom material contains too much of the
light key, then the operator will increase the set point to the flow controller. As the
condensate flow rate is increased, the vapor/liquid interface layer is lowered in the
exchanger exposing more surface area of the tubes for condensation (heat transfer)
to occur. Thus the heat input to the reboiler is increased.

UOP Confidential - Do Not Copy 208

 Distillation Controls – Heat Input
(Steam Reboiler Control)
Design Considerations:
1. Heat input is a function of steam temp
2. Q = UA(LMTD) Thermostatic
3. Controlled pressure sets heat transfer rate Trap
Ts - Tp > 30 oF
To Atmosphere
at Safe Location
FIC

FT
Tp Ts

Steam
LT LIC

Free Draining
(No Pockets)
Condensate
IST-R00-(178A)

EDS 2006/Inst-209

The second control system utilizing steam as a heating medium is shown as above.
This control scheme employs steam flow control in place of the condensate flow
control. Additional equipment is used because the reboiler essentially operates dry
utilizing 100% of the exchanger heat transfer surface area and the condensate drains
into a collection drum with level control on the drum. The driving force for heat
transfer is the log mean temperature difference (LMTD), and the steam temperature
is manipulated by controlling the steam pressure in the reboiler.
In order for this type of heat transfer to function properly the steam saturation
temperature (Ts) must be greater than the process outlet temperature plus 30 °F.
During normal operation if the column bottom material contains too much of the
light key, then the operator will increase the set point to the flow controller. As the
steam flow rate is increased, the control valve differential pressure is being reduce;
thus, the downstream pressure increases (and as a consequence the saturation
temperature in the exchanger rises). The LMTD is increased and with a larger
steam flow rate the heat input to the reboiler is increased.

UOP Confidential - Do Not Copy 209

 Distillation Controls
(Gibbs Phase Rule)

„ Provides basis for Composition Control

„ F=C-P+2
– F = Number of Degrees of Freedom
– C = Number of Components
– P = Number of Equilibrium Phases
„ Independent variables are temperature, pressure,
and composition
„ For binary system, F = 2 - 2 + 2 = 2
„ Approximation for multi-components

EDS 2006/Inst-210

The Gibbs Phase Rules states that the total number of Degrees of Freedom is equal
to the total Number of Components in the system minus the Number of Equilibrium
Phases plus 2. For the typical distillation column, the independent variables within
the boundary of the the column are temperature, pressure, and composition. For a
binary system the Number of Components is 2 and the Number of Equilibrium
Phases is 2; therefore the number of Degrees of Freedom is 2.
Gibbs Phase Rule states that if two of these independent variables are fixed
(controlled), the third independent variable will also be fixed. Therefore if pressure
and temperature, pressure and composition, or temperature and composition are
controlled, then the third remaining independent variable is fixed.
For most applications the column pressure and a column tray temperature are the
two independent variables that will be controlled. Fixing these two variables will
fix the composition of the overhead product.

UOP Confidential - Do Not Copy 210

 Distillation Controls
(Pressure Control)

„ Gibbs Phase Rule: F = 2

– Controlling pressure fixes one degree of freedom
– One degree of freedom remains
„ Design Options for OVHD pressure control
– Cooling Medium: Air or Water
– Overhead Composition: Total or Partial Condensation
– Actual magnitude of column pressure

EDS 2006/Inst-211

For multi-component distillation the Gibbs Phase Rule is an approximation and the
number of Degrees of Freedom is 2. Pressure control is relatively simple, but is
contingent upon several design factors. These design factors include the type of
cooling medium, composition of the column overhead material, and the design
operating pressure of the column.
These factors will be reviewed and we will investigate UOP’s philosophy in
reviewing various column designs.

UOP Confidential - Do Not Copy 211

 Distillation Controls – Pressure Control
(Total Condensation, P < 5 psig)

To Flare To Flare
Header Header
Flow
Flow
Orifice
Purge Orifice
Gas Purge
Gas

Cooling Water Condenser Air Fan Condenser

Design Considerations:
1. Receiver designed to float on flare header
2. Condenser located above receiver in either design
3. Purge gas sweeps line to prevent backfow of contaminants
4. Purge gas velocity designed for 0.5 - 1.0 feet/second IST-R00-199A

EDS 2006/Inst-212

The first system, illustrated above, has an operating pressure less than 5 psig
without any non-condensables in the column overhead material. The
receiver/column is designed to float on the flare header; thus operating the column
at or near atmospheric pressure. The size of the vent line to the flare header is
contingent upon the design of the receiver. A nitrogen purge (via pressure regulator
and restriction orifice) is added to the receiver to sweep the vent line and prevent
back-flow contamination of material from the flare header into the receiver during
transient states throughout the refinery. Once the vent line size is determined, the
nitrogen flow rate can be determined based on a design velocity between 0.5 to 1.0
feet/second. The design basis downstream of the restriction orifice is 1 psig.
If a water cooled condenser is in the system design, the exchanger should be
elevated above and free draining to the receiver. The non-condensable purge gas
must be able to get back to the exchanger; otherwise the exchanger will tend to
flood and column pressure will not change until the flooded area changes (causing
an increase in rate of condensation).

UOP Confidential - Do Not Copy 212

 Distillation Controls – Pressure Control
(Total Condensation, 5 psig < P < 25 psig)
PT PIC PT PIC

To Flare To Flare
Header Header

Blanket Blanket
Gas Gas

Cooling Water Condenser Air Fan Condenser

Design Considerations:
1. Condenser located above receiver in either design
2. Split ranged gas blanket/vent system employed to control pressure
3. Maximum blanket gas rate based on rated capacity of overhead pumps
4. Maximum vent gas rate assumed equal to maximum blanket gas rate IST-R00-200A

EDS 2006/Inst-213

The next system illustrated above is also for a totally condensing system, but with
an operating pressure between 5 and 25 psig. A “push-pull” gas blanketing system
is employed for column pressures operating within this range. This system (as
discussed previously) is a split range system which opens a valve to vent gas to the
flare header if the pressure rises above the pressure controller’s set point, and opens
another valve to bring in make-up gas when the pressure falls below the pressure
controller’s set point. For instances when the measured pressure and the pressure
controller’s set point are equal, both the vent valve and the make-up valve will be in
the closed position.
The condensers, whether water or air cooled, are located above the receiver for
optimal pressure control as discussed in the previous example.

UOP Confidential - Do Not Copy 213

 Distillation Controls – Pressure Control
(Total Condensation, P > 25 psig)
PT PIC PT PIC
PDIC
PDT

Syphon
Breaker

Cooling Water Condenser Air Fan Condenser

Design Considerations:
1. Condenser located below receiver for cooling water design
2. Increasing vapor rate raises liquid level in exchanger - less heat transfer
3. Pressure valve control pressure on column for both designs
4. Pressure differential valve maintains drum press/temp for air system IST-R00-201A

EDS 2006/Inst-214

For totally condensing systems with operating pressures above 25 psig, the system
will differ depending upon whether water-cooled or air-cooled condensers are built
into the design.
For a water-cooled system, the design is based on a flooded condenser with the
condenser being located below the receiver. The outlet piping from the condenser is
routed to the bottom of the receiver and therefore the water condenser is “sealed” in
the liquid. The pressure is controlled by bypassing hot vapors around the condenser
directly to the overhead receiver. The pressure differential across the control valve
is equal to the liquid static head from the receiver vapor/liquid interface to the
exchanger interface plus the frictional pressure drop. The receiver and condenser
form a “U” tube arrangement. As the hot vapor bypass valve opens the pressure in
the receiver will increase. The increase in receiver pressure forces liquid out of the
receiver back into the condenser flooding more tubes. With additional tubes
flooded, less surface area is available for condensation; therefore, the pressure in the
column will rise. Alternately with the hot vapor bypass control valve closing, the
column pressure pushes liquid out of the exchanger into the receiver exposing more
surface area for condensation to occur. Therefore the column pressure will drop.
The hot vapor bypass valve is sized for about 6% of the overhead vapor rate (as a
maximum)and specified with an equal percentage characteristic trim.

UOP Confidential - Do Not Copy 214

 Distillation Controls – Pressure Control
(Total Condensation, P > 25 psig continued)
PT PIC PT PIC
PDIC
PDT

Syphon
Breaker

Cooling Water Condenser Air Fan Condenser

Design Considerations:
1. Condenser located below receiver for cooling water design
2. Increasing vapor rate raises liquid level in exchanger - less heat transfer
3. Pressure valve control pressure on column for both designs
4. Pressure differential valve maintains drum press/temp for air system IST-R00-201A

EDS 2006/Inst-215

For an air-cooled system installing the air condenser below the receiver is not
practical and; and therefore the condenser is not operated as explained for the water-
cooled system. Column pressure is maintained by installing a pressure control
valve upstream of the condenser to throttle the overhead vapor and control the
column pressure directly.
However with this in mind the receiver’s pressure/temperature relationship is a
function of the available cooling in the condensers at any given time. Since the
pressure in the receiver is essentially the vapor pressure of the condensing fluid
(total condensation system); and as the amount of heat extraction varies, the reflux
temperature and pump suction pressure will vary. This will lead to very unstable
control of overhead composition as ambient conditions change between
winter/summer, night/day, sunny/stormy conditions, etc. Not only is the reflux
temperature affected, changes in receiver pressure affect the pressure differential
across the control valve. As the temperature drops (and consequently the pressure
drops), the discharge pressure on the pumps for a given flow drops. Thus the
control valve inlet pressure and differential pressure are lower and for a given reflux
rate the valve must respond by opening more and more to compensate for the loss in
driving force.
Installing the hot vapor bypass control valve will maintain the heat balance in in the
system and maintain the receiver at a constant pressure and temperature. The hot
vapor bypass valve is sized for about 6% of the overhead vapor rate (as a
maximum)and specified with an equal percentage characteristic trim.

UOP Confidential - Do Not Copy 215

 Distillation Controls – Pressure Control
(Total Condensation, Steam Generation)
PT PIC

PIC

FE
PT

Steam

Three Element
LT Boiler Feed Water
Control System

FE

BFW

Design Considerations:
1. OVHD product must have sufficiently high boiling point to produce steam
2. Cascade OVHD pressure controller to steam generator pressure controller
3. OVHD pressure controller should be reverse acting controller
4. Increasing the steam generator pressure controller set point, decreases heat transfer rate IST-R00-202A

EDS 2006/Inst-216

For some distillation columns heat integration with the refinery-wide steam system
may be a feasible alternative to generate low to medium pressure steam. The
boiling point of the overhead material must be at a sufficiently high enough
temperature (usually 300 °F or higher) to generate low or medium pressure steam.
A kettle type condenser is used in the overhead of the column and a pressure to
pressure cascade control scheme is utilized to control the pressure in the column.
The column pressure controller is the primary controller in the cascade, while the
steam pressure controller is the secondary controller.
If pressure in the column drops, indicating a reduction in overhead vapor rate, the
set point to the steam pressure controller increases. An increase in the pressure in
the steam drum increases the steam temperature. This in effect reduces the LMDT
across the exchanger reducing the heat transfer rate; and therefore reduces the
condensation rate on the process side leading to an increase in column pressure.

UOP Confidential - Do Not Copy 216

 Distillation Controls – Pressure Control
(Partial Condensation)
PT PIC

Off Gas

Design Considerations:
1. For partial condensation, locate pressure control valve in off-gas line
2. Locate transmitter near column top when using composition control
3. Transmitter may be located on receiver if stripping column
IST-R00-203A
4. Water condensers may be located above or below receiver

EDS 2006/Inst-217

For distillation processes that have non-condensables in the overhead, a vent gas
line is required to route the non-condensables out of the system. The destination
can be the refinery fuel gas header, flare system, storage, etc, depending upon the
quality and value of the product. The control valve is installed in the vent gas line
and the pressure transmitter can be installed either at the top of the column or on the
overhead receiver. If composition control is a requirement, then the pressure
transmitter should be installed near the top of the column for optimum control of the
column products. However if the column design is for stripping then the transmitter
can be installed at or near the receiver for easy accessibility.
The overhead condenser can be either water cooled or air cooled. If water cooled,
the condenser can be located either above or below the receiver.

UOP Confidential - Do Not Copy 217

 Distillation Controls
(Composition Control)

„ With F = 2, pressure control fixes 1 degree of freedom;

1 degree of freedom remains
„ Remaining independent variable are Temperature (T)
and Composition (C)
„ Fixing either T or C fixes the last variable
„ For most applications temperature control is adequate
for control of product composition
„ Alternatively on-line analyzers could be used for
composition control directly

EDS 2006/Inst-218

Pressure control is relatively easy to implement on the overhead of a column. We

have reviewed several applications over a wide range of operating conditions and
included total or partial condensation. In terms of Gibbs Phase Rule (and assuming
the approximation of a binary system for most multi-component applications) fixing
the column pressure defines one of the two degrees of freedom.
Temperature and composition are the remaining degrees of freedom. Therefore if
we can control temperature, then composition will be fixed; or if we can control
composition then the temperature will be fixed.
For most applications temperature control is adequate for control of product
composition. Several variations of temperature control can be implemented
depending upon the difficulty of the separation. However in some instances on-line
gas/liquid chromatographs could be used for composition control directly.

UOP Confidential - Do Not Copy 218

 Distillation Controls
(Composition Control, cont’d)

„ Options for OVHD temperature (T) control

– Simple end point, locate in OVHD piping
– Split between light and heavy key components, locate on
one of the column trays
– For more difficult separations at low operating pressures,
delta T control is used to compensate for variations in
column pressure control
– A variation of delta T control uses the basic temperature
control with pressure compensation as a function of
composition

EDS 2006/Inst-219

These are some of the more common means of temperature control from the
simplest to some more complicated systems. The higher the degree of separation
difficulty, the more sophisticated the system.
For stripping columns a simple end point determination of the overhead product can
be achieved by locating the temperature measurement in the overhead piping.
For the more traditional composition control, the temperature measurement is
located on one of the column trays above the feed tray. The temperature
measurement is then utilized to control the split between the light and heavy key
components in the top of the column.
For the truly binary separation processes, especially when column design pressures
are near atmospheric, a temperature control scheme can be implemented as a means
for pressure compensation.
We can go one step further and model the composition and temperature as a
function of pressure and implement a control system that utilizes temperature
control with pressure compensation as a function of composition.
As the control schemes become more complicated, better VLE data and computer
simulations play an even more important part in developing the control scheme.

UOP Confidential - Do Not Copy 219

 Distillation Controls
(Tray Selection – multi-component)

„ Acceptable to use tray half way between feed tray and

reflux tray
„ Computer tray to tray simulations
– Product Quality vs. Tray Temperature
– Affects of Feed Rate or Feed Composition
– Tray with least variation in product quality at constant
temperature
„ Installation of spare thermowells above and below
control tray selected (optional)

EDS 2006/Inst-220

Tray to Tray simulations will provide a means of predicting product quality versus
tray temperature as a function of various process variables such as column feed rate
and/or feed composition. The tray which shows the least variation in product
quality at a constant temperature is determined. If uncertainty still exists, then spare
thermowells can be located above and/or below the control tray selected.

UOP Confidential - Do Not Copy 220

 Distillation Controls
(Tray Selection – Binary)

„ Low pressure, high purity binary separation

– Tight OVHD product spec
– Designs float on flare header or partial vacuums
– Product purity affected by pressure changes
– Delta T control compensates for pressure affect
„ Design considerations
– Dedicated header connected to flare header
– Reference T located near top of column
– Need to determine location of Composition T

EDS 2006/Inst-221

In an Aromatics Complex with BTX operation (Benzene, Toluene, and Xylene

recovery), benzene/toluene separation is a common example of binary distillation at
low pressure. The benzene spec is typically tight allowing for only PPM levels of
toluene in the overhead product. UOP typically designs the overhead system to
float on a dedicated header to the flare knockout drum.
Slight variations in header pressure affect tray temperatures and ultimately product
purity. A differential temperature control scheme is used to control overhead
composition. One temperature measurement, often referred to as the reference
temperature measurement, is located near the top tray of the column. The second
temperature measurement is located ‘x’ number of trays down the column and is
often referred to as the composition temperature measurement. Since the system is
basically a binary system the top of the column is primarily pure benzene. Further
down the column on tray ‘x’, is a mixture of benzene and toluene. As the
composition changes at constant pressure the temperature at the top of the column is
unaffected, but the temperature at the composition tray will vary as the composition
varies. Therefore if the composition is changing at constant pressure the differential
temperature is changing in proportion to the composition changes. However if the
pressure is not constant, the reference temperature measurement will change with
respect to benzene’s vapor pressure. Likewise on the composition tray, the
temperature will change similarly. Therefore the differential temperature will not
change with respect to changes in pressure at constant composition.

UOP Confidential - Do Not Copy 221

 Distillation Controls
(Tray Selection – Binary, cont’d)
Typical Benzene Column
100

400 ppm Toluene

200 ppm Toluene
100 ppm Toluene
80

Greatest temperature variation with

Differential Temperature, °F
product quality is between trays 19 -22
Select Tray 20 as control tray
60

40

20

Feed - Tray 26

0
0 10 20 30 40 50
ACTUAL TRAYS IST-ROO-204B

EDS 2006/Inst-222

In the above example computer simulations were run for the benzene/toluene
system and the differential temperature between the reference tray and various trays
in the column was plotted as a function of the actual tray. The family of curves
represent simulations at various toluene impurity levels.
From the simulations the greatest temperature variation with product quality is seen
between trays 19 and 22. Therefore for the binary separation, tray 20 was selected
as the composition temperature measurement for this design.

UOP Confidential - Do Not Copy 222

 Distillation Controls
(Tray Selection – Binary, cont’d)
To Flare
Header
1
4 LT

FT
TDT
19 Total
TDIC FIC Tray
20 Loading
FT delta P Composition

Reflux

Design Considerations:
1. Reference point located in top of column (pure component)
2. Control point located where greatest T variation with product quality
3. Changes in T due to pressure swings cancel with delta T control
4. Increasing reflux changes composition decreasing tray temperature;
but increases tray loading/tray pressure drop with increasing tray temperature
5. Low loadings - Composition governs; high loadings tray delta P governs
6. Tray temp is sum of two effects with minimum delta T attainable IST-R00-205A

EDS 2006/Inst-223

The differential temperature control scheme is shown above with the reference tray
at tray 4 and the composition tray at tray 20. The above plot illustrates that at low
tray loadings (small reflux rates) the differential temperature is at its greatest. As
reflux rates are increased less and less toluene will exist in the upper section of the
column and the differential temperature should diminish. However with an increase
in reflux rate, tray loading is increased (an increase in tray pressure drop) and
ultimately affects tray temperature with the increase in pressure on any given tray.
Therefore at low tray loading, composition changes govern the differential
temperature; but at high loading tray, tray differential pressure governs. As can be
seen in the above plot, the differential pressure swings through a minimum with
increasing reflux rate. Normal operation must be to the left of this minimum.
Feedback control will not work to the right of the curve due to the change in slope
of temperature differential vs. reflux flow.

UOP Confidential - Do Not Copy 223

 Distillation Controls
(Traditional Composition Control)

„ Traditional Composition Control

– Temperature control adjusts reflux rate
– Receiver level control adjusts distillate rate
• Preferred if Reflux is smaller than Distillate
• Preferred if Reflux is 10 times greater than Distillate
• Preferred if receiver used as surge for downstream
units
– Good response for changes in feed composition
– Poor response to external disturbances to heat balance
(i. e., rain storm or reboiler upsets)

EDS 2006/Inst-224

UOP uses two distinct control schemes for composition control. The first is called
Traditional Composition Control. For the traditional composition control, the
temperature controller will be aligned with the reflux rate and the receiver level
control will be aligned with the distillate draw-off rate.
This control scheme is preferred if the reflux to distillate rate is less than 1 or
greater than 10. This is also preferred when the receiver is used as a surge drum for
downstream units.
In the event of frequent feed composition changes this control scheme will respond
well. However the control scheme responds poorly to external disturbances to the
heat balance.

UOP Confidential - Do Not Copy 224

 Distillation Controls
(Traditional Composition Control, cont’d)

5 TT

LIC
LT

TIC FIC

FT

Design Considerations:
FRC
1. For disturbances in the OVHD condenser, external FIC
FT
reflux affects column operation by changing internal reflux.
Eventually TIC corrects external reflux.
2. During reboiler upsets LIC may send off-spec product
to storage.
3. System works well also where designs have small
distillate product rates. Level controls mass balance.
IST-R00-207A

EDS 2006/Inst-225

For a disturbance in reflux temperature, such as a sudden rain storm when air-
cooled condensers are used, the decrease in external reflux temperature will
immediately change the column’s internal reflux rate. The external reflux rate will
remain constant until the control tray temperature detects a change in composition,
but by this time the column is out of heat balance.
For a disturbance in the reboiler, such as an increase in steam pressure, the vapor
rate up the column will increase ultimately increasing the level in the receiver. The
traditional control scheme will send this increase in inventory to storage (this
additional material may well be an impurity in the product and the potential for off-
spec product is high).

UOP Confidential - Do Not Copy 225

 Distillation Controls
(Traditional Composition Control, cont’d)

„ Additional instrumentation and calculations required

for internal reflux control
– OVHD Temperature
– Reflux Temperature
„ Calculate internal reflux from heat and mass balance
around top tray
„ Calculated internal reflux becomes the corrected
process variable and external reflux is adjusted
accordingly

EDS 2006/Inst-226

To avoid these types of upsets with the traditional system, the internal reflux must
be calculated from a mass and heat balance around the top tray of the column. The
calculated internal reflux becomes the corrected process variable and the external
reflux is adjusted to manipulated the internal reflux. In order to calculated the
internal reflux, additional instrumentation is required to measure both the
temperature of the overhead vapor and the temperature of the external reflux stream.

UOP Confidential - Do Not Copy 226

 Distillation Controls
(Traditional Composition Control, cont’d)

Internal Reflux Calculations

TVo Vo
Material Balance:
Vi + Re = Vo + Ri + S TRe Re D
Energy Balance:
(Hvi)Vi +(Hre)Re = Vo = Overhead vapor
(Hvo)Vo +(Hri)(Ri + S) Vi = Internal vapor
Re =External reflux
Solve for Ri:
Ri = Internal reflux
Ri = (1 + KdT) -S
S = Sidecut
where:
D = Distillate
K = (Cp/dHvap)
dT = TRe - TVo
S

Ri Vi

EDS 2006/Inst-227

To initiate internal reflux control additional instruments are required. Both the
external reflux temperature and the overhead vapor temperature is required along
with a measurement of the external reflux rate. For the top tray, a material and
energy balance leads to calculating the internal reflux of the column. This is used
as the process variable to an internal reflux controller. The internal reflux controller
manipulates the external reflux flow, thus maintaining the internal reflux flow at its
set point.
As described earlier in our discussion on feed-forward applications, the calculations
above depend upon the assumed value of K. This is the ratio of the heat capacity of
the external reflux and the heat of vaporization of the liquid on the first tray. The
better the approximation the better the modeling will be for the internal reflux
calculation.

UOP Confidential - Do Not Copy 227

 Distillation Controls
(Material Balance Control)

„ Material Balance Control

– Temperature control adjusts distillate rate
– Receiver level control adjusts reflux rate
• Preferred if Reflux is larger than Distillate and less
than 10 times Distillate
• Not preferred if receiver is used as surge for
downstream units
– Level is a measure of Reflux and Distillate rates
– Good response for changes in feed composition and
external disturbances to heat balance

EDS 2006/Inst-228

The second composition control scheme is known as Material Balance Control. In

this control scheme temperature control is aligned with the distillate draw-off rate
and the receiver level is aligned with the reflux rate. This control scheme is
preferred when the Reflux to Distillate rate is greater than 1 and less than 10.
In this control scheme the reflux rate is not measured directly. Instead the level
measurement is a measure of both the reflux and distillate rates. Therefore when
the composition controller makes adjustments to the distillate rate, the incremental
change is immediately compensated for by making a corresponding adjustment in
the reflux flow.
As we saw in the traditional control scheme, the material balance control scheme
responds well to changes in feed composition and responds exceptionally well to
external disturbances in the column heat balance.

UOP Confidential - Do Not Copy 228

 Distillation Controls
(Material Balance Control, cont’d)

Design Considerations:
1
1. Material Balance control (as it is named)
because the temperature controller controls 5 TT
the on-spec distillate in the OVHD and by
material balance out the bottom of the
column on level control.
TIC LIC
LT
2. Better response to both feed composition
changes and heat balance changes. FIC

3. Composition controller changes product

flow: and because the reflux controller is a
function of (R +D), the reflux flow is afftected.
FT R+D
As soon as the distillate flow changes, the
total OVHD liquid flow changes and the
reflux flow is changed simutaneously. This R=R+D-D D
preserves the column heat balance and thus FT
is self-correcting.

FIC

IST-R00-207A

EDS 2006/Inst-229

For the traditional control scheme we saw that for an upset in the reboiler, the
increase in vapor rate in the column translated into an increase in receiver level and
ultimately an increase in product draw-off (good or bad product). However in the
material balance control scheme an increase in level is returned to the column as
reflux. If the upset is severe and the column is severely upset the composition tray
temperature will increase and ultimately the composition controller will adjust the
distillate rate downward. In the extreme the column will automatically be
controlled on total reflux until the composition profile throughout the column is
reestablished.

UOP Confidential - Do Not Copy 229

 Distillation Controls

– Review of Gibbs Phase Rule

– Strategies for constant Heat Input Control
– Design basis for Column Pressure Control
– Alternate designs for Composition Control
– Recommended Material Balance Control

EDS 2006/Inst-230

This concludes the review of distillation control and the various control schemes
utilized by UOP in our typical designs.

UOP Confidential - Do Not Copy 230

 EDS 2006/Inst-231

If there are any questions about my presentation, I have a few minutes to field a
couple of them from the audience.

UOP Confidential - Do Not Copy 231