Vous êtes sur la page 1sur 139

CHAPTER

1
Abstract
Improved performance is the prime directive in todays industrial
processes. Throughputs have increased drastically, decreasing the time
available to detect and control upsets. Equilibrium has become the
exception rather than the rule. Poorly tuned regulatory controllers would
eventually integrate the errors to achieve steady state control, and
variance caused by poorly implemented control strategies and
improperly de-tuned controllers was considered part of the process.
Inadequately designed control systems, malfunctioning control
equipment, and poorly tuned and coupled loops could be tolerated in the
past as plants were designed to operate at constant loads, but not today,
this approach is no longer acceptable.

Introduction
Regulatory control systems employ the application of feedback and
feedforward control loops to control various physical process properties.
A regulatory control system may often contain hundreds, sometimes
thousands, of individual control loops. For a control system to provide
optimum control, it must be designed, installed, and tuned, to minimize
variability, yet adapt efficiently to changing conditions. In a large system,
where the dynamics are unknown, processes interact, measurements are
noisy, valves stick, and load conditions vary, it seems a daunting task to
optimize the control of the system. The purpose of this manual is to
provide information on how to apply the Protunertm WPSA and the
Control System Analysis Procedures to:
Define the individual unit operations that make up the overall
control system
Understand the control strategies employed to control each unit
operation
Inspect the field installation of the control and measurement
equipment
Develop a test sequence for testing each loop in the unit operation
Test individual loops with a procedure which minimizes process
upsets, yet provides the necessary dynamic information for analysis

Application Manual 1
ABSTRACT

Analyze the test data to troubleshoot the system and optimize the
tuning of each loop

One important advantage, that is frequently overlooked in the


application of the Control System Analysis Procedures, is complete
understanding of the process itself. More complete process knowledge
often leads to changes in control strategies that result in significant
improvements in production, product quality, and improved
profitability.

Application Manual 2
CHAPTER

2
Getting Started
Control System Analysis requires a certain amount of preparation before
the loops are actually tested with the Protunertm. First, optimizing a
control system requires an understanding of the control strategy
employed. Secondly, measurements must be accurate and valves must
move in response to changes in the controller output. This section
presents recommended procedures that should be followed before you
start testing a control system.
1. Study the loop drawings to understand the control strategy being
employed. Make a sketch of the control system being tested showing
the location of all the valves, transmitters, and control loops in the
unit operation. REMEMBER, you cannot tune and optimize a control
system you do not understand.
2. Discuss the system unit operation and the loops you plan to work on
with the operators. While some may not always be familiar with in-
depth control strategies or inherent design problems, they have
learned to overcome the systems limitations using manual control,
and kept production running. Operators work with the system on a
continuous basis and often can provide valuable insight concerning
the systems operating characteristics and problems they may have
already identified. Remember, to be successful, you must make the
operators part of the optimization team.
3. Inspect the control equipment and the field installations. Physical
inspection is an important step in the process.

Transmitter Inspection
Calibration
Check the transmitter calibration reports. If the transmitter has not
recently been calibrated it is highly recommended that you get the
transmitter calibrated before you begin testing. Keep in mind: You
cannot control what you cannot measure.

Damping adjustment
The transmitter damping time constant must be set equal to 1.3 times the
controller scan rate to prevent aliasing of high frequency transmitter

Application Manual 3
GETTING STARTED

noise. See the Tech Note on Signal Aliasing for more information on
pre-filtering requirements.

Installation and mounting


It is always wise to inspect the physical installation of the transmitter
mounting to be sure that it conforms with the manufacturers
recommendations. Keep in mind, that all to often, control is an after
thought in the design of a system, resulting in installations that do not
conform to the manufacturers recommendations. It is amazing some of
the installation problems we have found over the years.
For example; insufficient straight runs of pipe for flow meters is
common, signal wires badly corroded due to leaky junction boxes, dp
transmitters located incorrectly above the pipe on flow meter
applications. In one plant, we found a number of newly installed
magnetic flow tubes installed in downlines which resulted in air in the
line and noisy measurements at low flows. In a number of paper mills,
field inspection found magnetic flow meters located just downstream of
the chemical addition pumps, resulting in a continuous changing of the
electochemical nature of the fluid and very noisy flow signals. Recently, a
dp transmitter used for an open tank level measurement, was found with
the plastic shipping plug still in the atmospheric port.

Valve Inspection
Calibration
Calibration of the actuator bench set, I/P transducer, and positioner,
requires stroking the valve through its full travel. Therefore, calibration
cannot be easily done while the process is running. On the whole, control
valves are not calibrated well and are a major source of control problems.
A quick calibration check would be:
1. Place the loop in manual and compare the controller output to the
actual valve position.
2. If there is a significant difference, the positioner zero can be adjusted
to eliminate the variance.
3. Place the loop back in automatic. When the process stabilizes at
setpoint, place the loop in manual and compare the difference
between the controller output and the actual valve position.
4. Adjust the positioner span to eliminate the error.

Application Manual 4
GETTING STARTED

5. Repeat the process until the controller output signal matches the
valve position.

Improper calibration of the control valve travel can result in a number of


control problems. If the loop is tuned with an incorrect calibration, re-
calibration may change the process gain, and thus require changes in the
controller tuning parameters.
Improper calibration will result in controller reset windup problems.

As an example, if the valve is at 100% travel at an 80% controller output,


a large upset may require the controller output to go to 100% to correct
the upset, as the process reaches setpoint, the control action must unwind
from 100% to 80% before the valve starts to move, resulting in a large
overshoot.

Linkage Inspection
Linkages are often used to connect the actuator to the valve stem and the
positioner feedback to the actuator stem. Loose linkages can result in all
kinds of cycling problems and should be eliminated. If you find valves,
where the linkages are worn and can be shown to be the cause of poor
control, a case can be made to replace the valve.

Air Supplies
Check the regulators settings supplying instrument air to the valve
actuator and its accessories.
Look for any undersized tubing, pinched lines, and air leaks. In
inspecting the air supply systems we have identified and fixed a large
number of problems, which if unresolved, would have made optimum
control impossible.
As an example, in a paper mill stock blending system we found a
number of newly installed ball valves with piston actuators controlling
the various stock flows. The air supply to the piston actuators came from
the old wall mounted regulators supplying only 20 psi, and not the 80 psi
required. During installation, the actuators appeared to work, but under
operating conditions they did not have enough force for smooth
operation.
In a chemical plant, the advanced control group installed a fussy logic
controller in an attempt to control what was thought to be a difficult
process. In inspecting the field installation, we found a leaking

Application Manual 5
GETTING STARTED

diaphragm in the I/P which in manual caused the valve position to drift
around. Once the problem was identified and fixed, the control of the
loop was quite simple.

In short, get out of the control room and check the equipment, BEFORE you
start tuning.

Application Manual 6
CHAPTER

3
Loop Analysis Test Procedure
Self Regulating Process Type
This section covers test and analysis procedures for tuning self-
regulating control loops. Self-regulating processes such as flow, small
volume pressure, temperature, consistency, ORP, pH, etc., have one
characteristic in common, in response to a step change in the controller
output, the process variable will stabilize at a new steady state value
after all the process dynamics die out.

NOTE
Loop Analysis testing requires that the Protunertm be connected to record the
PD and PV of the loop being analyzed and the signals of any other loops which
may interact with the loop being tested. When testing a loop to determine tuning
parameters all interactive loops must be in automatic control. When testing a
system to determine relative gain between interactive loops the open loop testing
is done with all competing loops in manual.

TEST 1
Closed Loop As Found Response to Setpoint Changes

The first test in the Loop Analysis Procedure is to record the closed loop
response of the loop being analyzed to changes in its setpoint. The test
objective is to document how well the as found PID parameters control
the loop.

Test Procedure
1. Determine the largest setpoint change that can be made without
causing production problems.
2. Record the response to a setpoint change up and down. Wait after
each change for the process variable to reach the new setpoint.

Application Manual 7
LOOP ANALYSIS - SELF-REGULATING

Figure 3.1 - Typical Closed Loop As Found Test

Figure 3.1 illustrates the Data Plots screen of a typical closed loop as
found response to 5% setpoint changes. Utilize the Autoscale, Zoom,
View Data, and Cursor functions to analyze the closed loop response
data. The following is the recommended analysis procedure:
1. Enter Notes about the test:
a. Loop Tag Number
b. The as found PID tuning parameters
c. Transmitter and controller filter settings
d. Comments on the signal noise
e. Any apparent interaction with other loops
f. The time it took to reach setpoint
g. Any other notes required to document your findings

2. Use the Autoscale, Zoom, and View Data functions to display the PD
and PV graphs in a manner which best illustrates the test results.
Switch the display to the Comments-On mode and add balloon
comments to the graphical displays to document your findings. If in
the analysis of the Closed Loop As Found test data, you discover

Application Manual 8
LOOP ANALYSIS - SELF-REGULATING

evidence of control or equipment problems, you will want to


investigate them further and fix them before proceeding.

TEST 2
Open Loop Step Test Procedure and Loop Analysis

The test objective, is to record a series of open loop step tests which
contain the information necessary to troubleshoot and tune the loop
utilizing the Protunertm WPSA for Windowstm Loop Analysis functions.

Test Procedure:
1. Place the loop in manual.
2. Record equal 3% to 5% step changes in the controller output in the
following sequence, two steps up, three steps down, one up. After
each step, WAIT for the process variable to stabilize before making
the next step.

NOTE
In the real world, valves may have deadband or stiction characteristics and
therefore, may or may not move in response to any single step change. Load
upsets could occur during testing which would distort the PV response data,
and the process gain may not be linear as a function of valve position. The main
objective of the open loop step test procedure is to insure the recording of enough
step tests so that by simply looking at the data you can determine which step
tests contain valid information about the true process variable response. The
recommended up and down step changes recorded can be used to determine the
valve hysteresis along with sufficient steady state data for noise analysis. You
can modify the step size and the sequence, as long as the test data contains
enough information for accurate loop analysis.

Loop Analysis
Preparing the Test Data for the PID Tuning Procedure

The Loop Analysis Tuning screen displays only the PD and PV signals
identified in the Loop Analysis dialog box.

Application Manual 9
LOOP ANALYSIS - SELF-REGULATING

Figure 3.2 - Loop Analysis Screen


The raw step test data displayed on the PD and PV graphs on the Loop
Analysis-Tuning screen, contains the numerous step changes in the
controller process demand and the corresponding responses in the
process variable. The following is the correct step by step procedure to
prepare raw data to calculate optimum tuning parameters.

Loop Analysis Functions:


Noise - Documents the magnitude signal noise present on the PD
and PV signals.
Hysteresis - Documents the normalized hysteresis plus deadband in
the valve.
Tune - Generates the Loop Analysis Report which includes PID
tuning parameter table.

NOTE
Controller Filtering, Editing, Windowing and choosing the correct Controller
are required for Tuning, Noise and Hysteresis are not.

Application Manual 10
LOOP ANALYSIS - SELF-REGULATING

Noise Analysis

Click the Noise tab to call the Noise Analysis screen. Use the Window
and Zoom functions to define a portion of the data where no changes
were made in the controller output that appears to be typical signal
noise. Click Estimate to display the signal noise range in percent. The
magnitude of the signal noise calculated is displayed in the Loop
Analysis Report.

Figure 3.3 - Noise Analysis Screen Windowed for Noise Analysis

Noise is defined as random disturbances occurring at frequencies too


high for the control action to correct. In a feedback control loop, the noise
is seen by the controller as a random error. The controller action, trying
to eliminate the error, causes random meaningless controller output
changes. Thus, excessive signal noise can result in excessive valve
movement at steady state.

Application Manual 11
LOOP ANALYSIS - SELF-REGULATING

Hysteresis Analysis

Click the Hysteresis tab to call the Hysteresis Analysis screen. Window a
section of data where a step up and step down in the PD corresponds to
the response in the PV. Click Self-Regulating, Estimate, to display the
calculated normalized hysteresis in percent.

Figure 3.4 - Hysteresis Analysis Screen Windowed for Hysteresis Analysis

The self-regulating normalized hysteresis is calculated using the


following formula:
Offset in PV
Hysteresis =
( PV / PD)
The estimated hysteresis is displayed on the Loop Analysis Report
screen. In the Loop Analysis Report, you can tab Hysteresis to view and
recalculate the hysteresis estimate.

Application Manual 12
LOOP ANALYSIS - SELF-REGULATING

Final Control Element Maximum


Actuator and Accessories Acceptable
Hysteresis
Spring and Diaphragm with I/P 3%
Spring and Diaphragm with Positioner 1%
Piston with Positioner 1%
Damper with Positioner 2%
Variable Speed Pump 2%
Table 3.1 - Maximum Acceptable Hysteresis

Undersized actuator
Excessive packing friction
Seal friction in rotary valves
Inadequate supply air pressure
Loose or worn linkages
Defective I/P transducer
Defective positioner
Table 3.2 - Sources of Hysteresis plus Deadband

In self-regulating processes the magnitude of the hysteresis plus


deadband does not change the controller tuning parameters or cause
continuous closed loop cycling. However, hysteresis in the control
equipment does affect the closed loop variance and how well the
optimum tuning parameters can control the process.

Tuning
Step 1 - Determine the Controller Filter Time Constant if Required
Inspect the response data to determine if the PV signal contains
excessive noise. Excessive signal noise is noise which cannot be
eliminated at its source, and if left unfiltered will result in excessive
control action and ultimately excessive wear and tear on the valve. You
will need to determine a controller filter constant to smooth the noise
before the controller acts on the signal.

NOTE

Application Manual 13
LOOP ANALYSIS - SELF-REGULATING

A controller filter not only smoothes the noise, it also adds lag to the
measurement. Therefore, you will want to choose the smallest filter constant
possible.

Figure 3.5 - Determining the Correct Controller Filter Constant

1. Click the PV tab to display the PV variable graph.


2. Visually inspect the noise content of the PV graph and decide if
controller filtering is required to smooth the PV signal. That is, is the
noise causing the closed loop control action to excessively cycle the
valve during normal steady state operating conditions.
The decision to use controller filtering is a judgment call.
3. Click the Filter button to call the Filter dialog box, select the filter
type used in your controller.
4. Start with a small filter time constant and increase it. For each filter
time entered, inspect the raw and filtered signals displayed until you
are satisfied with your selection.

NOTE

Application Manual 14
LOOP ANALYSIS - SELF-REGULATING

The Tuning Procedure does not use the filtered PV signal in the calculation of
the PID tuning parameters. The Tuning Procedure uses the unfiltered PV signal
to determine the process transfer function and adds the chosen filter constant in
the frequency domain to determine the tuning parameters.

Step 2 - Window Step Test Response of the True Process


Windowing the step test response data is the most important operation
in the analysis procedure. It allows you to identify which step test best
represents the true response of the PV to a step input in the PD. Choosing
the largest PV response will produce the most conservative tuning
parameters.

Figure 3.6 - Windowing the Step Test Data for Tuning Analysis
1. Visually inspect the test data and determine the step test data that
best represents the true process response.
2. Place the left window marker where the PD step first starts and the
left window marker where the PV step appears to stabilize at steady
state. Click Zoom to view the test data.

Application Manual 15
LOOP ANALYSIS - SELF-REGULATING

3. Refine both the right and left window markers to more precisely
define the response data.

Step 3 - Editing the PD and PV Data If Necessary


If the unfiltered PV or PD data in the tuning window appears to contain
excessive noise which distorts the data, the graphic editing function can
be used to remove the noise without distorting the true response data.
The Tuning Procedure uses the edited test data to determine the process
transfer function and the tuning parameters.

Figure 3.7 - Graphic Editing the Response Data

1. Click Edit on the menu bar and choose the graph you wish to edit.
2. Use the draw features to edit the graph to remove noise and load
disturbance data.

Application Manual 16
LOOP ANALYSIS - SELF-REGULATING

Step 4 - Execute the Tune Function


The Tune function calls the Tuning Conditions dialog box. Identify the
Loop Type as Self-Regulating and be sure that the controller
identification options match the controller being used. See Appendix for
more in-depth information on controllers.

Figure 3.8 - Loop Analysis Report Screen

Application Manual 17
LOOP ANALYSIS - SELF-REGULATING

Application Manual 18
CHAPTER

4
Loop Analysis Test Procedure
Integrating Process Type
This section covers test and analysis procedures for tuning integrating
control loops. Integrating processes such as level, large volume pressure,
batch temperature, etc., have one characteristic in common, they all
control either mass or energy balance processes, therefore, at any given
process load only a single valve position balances the process conditions
and holds the PV steady. Thus, a step change in the controller results in a
process variable response which ramps at a constant rate after all the
process dynamics die out.

NOTE
Loop Analysis testing requires that the Protunertm be connected to record the
PD and PV of the loop being analyzed and the signals of any other loops which
may interact with the loop being tested. When testing a loop to determine tuning
parameters all interactive loops must be in automatic control. When testing a
system to determine relative gain between interactive loops the open loop testing
is done with all competing loops in manual.

TEST 1
Closed Loop As Found Response to Setpoint Changes

The first test in the Loop Analysis Procedure is to record the closed loop
response of the loop being analyzed to changes in its setpoint. The test
objectives is to document how well the as found PID parameters control
the loop. In our experience, a majority of integrating type processes have
proportional gain settings that are too small, and integral settings which
are too fast. Thus, the closed loop response to a setpoint change may
result in a stable response which can take a very long time to stabilize
following a setpoint change.

Test Procedure:

Application Manual 19
LOOP ANALYSIS - INTEGRATING

1. Determine the largest setpoint change that can be made without


causing production problems.
2. Record the response to a setpoint change up and down. Wait after
each change for the process variable to reach the new setpoint.

Figure 4.1 - Typical Closed Loop As Found Test Data


Figure 4.1 illustrates the Data Plots screen of a typical closed loop as
found response to 5% setpoint changes. Utilize the Autoscale, Zoom,
View Data, and Cursor functions to analyze the closed loop response
data. The following is the recommended analysis procedure:
1. Enter Notes about the test:
a. Loop Tag Number
b. The as found PID tuning parameters
c. Transmitter and controller filter settings
d. Comments on the signal noise
e. Any apparent interaction with other loops
f. The time it took to reach setpoint
g. Any other notes required to document your findings
2. Use the Autoscale, Zoom, and View Data functions to display the PD
and PV graphs in a manner which best illustrates the test results.

Application Manual 20
LOOP ANALYSIS - INTEGRATING

Switch the display to the Comments-On mode and add balloon


comments to the graphical displays to document your findings and
Print the report. If in the analysis of the Closed Loop As Found test
data you discover evidence of control or equipment problems, you
will want to investigate further and fix them before proceeding.

TEST 2
Open Loop Step Test Procedure and Loop Analysis

The test objective, is to record a series of open loop step tests which
contain the information necessary to troubleshoot and tune the loop
utilizing the Protunertm WPSA for Windowstm Loop Analysis functions.

Test Procedure:
1. Place the loop in manual with the process at steady state.

NOTE
If the process is not stable in automatic control, (a typical problem with poorly
tuned integrating processes), turn off the controller integral and derivative
settings. This will allow the P only controller to stabilize the process before the
controller is placed in manual.

2. Record equal 10% to 30% step changes in the controller output in the
following sequence; up-down, down-up. After the first step up,
allow the process variable to ramp at least 5%-10% from its initial
value, followed by a second step down of equal size. Make a third
step change in the same direction (down) as the second step, to again
upset the process balance. When the process variable approaches its
original value, make a fourth step in the opposite direction to stop
the test. If at the conclusion of the step sequence the process variable
does not stabilize, the final control element has some hysteresis. To
stabilize the process variable and determine the magnitude of the
hysteresis make small changes in the controller output, in the same
direction as the last step, until the process variable stabilizes.

Loop Analysis

Preparing the Test Data for the PID Tuning Procedure

Application Manual 21
LOOP ANALYSIS - INTEGRATING

The Loop Analysis Tuning screen displays only the PD and PV signals
identified in the Loop Analysis dialog box.

Figure 4.2 - Loop Analysis Screen

The raw step test data displayed on the PD and PV graphs on the Loop
Analysis-Tuning screen, contains the numerous step changes in the
controller process demand and the corresponding responses in the
process variable. The following is the correct step by step procedure to
prepare raw data to calculate optimum tuning parameters.

Loop Analysis Functions:


Noise - Documents the magnitude signal noise present on the PD
and PV signals.
Hysteresis - Documents the normalized hysteresis plus deadband in
the valve.
Tune - Generates the Loop Analysis Report which includes PID
tuning parameter table.

Application Manual 22
LOOP ANALYSIS - INTEGRATING

NOTE
Controller Filtering, Editing, Windowing and choosing the correct Controller
are required for Tuning, Noise and Hysteresis are not.

Noise Analysis

Click the Noise tab to call the Noise Analysis screen. Use the Window
and Zoom functions to define a portion of the data where no changes
were made in the controller output, that appears to be typical signal
noise. Click Estimate to display the signal noise range in percent. The
magnitude of the signal noise calculated is displayed in the Loop
Analysis Report.

Figure 4.3 - Noise Analysis Screen Windowed for Noise Analysis

Noise is defined as random disturbances occurring at frequencies too


high for the control action to correct. In feedback control loops, the noise
is seen by the controller as a random error. The controller action, trying

Application Manual 23
LOOP ANALYSIS - INTEGRATING

to eliminate the error, causes random meaningless controller output


changes to eliminate the error. Integrating processes typically require
large proportional gain settings which makes noise on the PV signal
more of a problem than on typical self-regulating process types.

Hysteresis Analysis

Click the Hysteresis tab to call the Hysteresis Analysis screen. Place the
left window marker to identify the controller output (PD) where the PV
was stable. Place the right window marker on the PD graph near the end
of the test where the PV is stable. Click Integrating, Estimate, to display
the calculated normalized hysteresis in percent.

Figure 4.4 - Hysteresis Analysis Screen Windowed for Hysteresis Analysis


The Integrating normalized hysteresis is calculated using the following
formula:
Hysteresis = PD

Application Manual 24
LOOP ANALYSIS - INTEGRATING

The estimated hysteresis is displayed on the Loop Analysis Report


screen. In the Loop Analysis Report, you can tab the Hysteresis Analysis
screen to view and recalculate the hysteresis estimate.
Estimating the valve hysteresis on integrating loops is extremely difficult,
in that its accuracy relies entirely on the test technique. For an accurate
hysteresis estimate, connect the Techmation stem position transmitter to
the control valve. Conduct an open loop self-regulating test and
determine the hysteresis comparing the controller PD to the stem
position as the PV signal.
NOTE
In integrating processes, hysteresis plus deadband will result in closed loop
cycling if PI or PID tuning is to be used. The controller integral action plus
hysteresis in the valve is the cause of the steady state cycling. Therefore, a valve
positioner is required to minimize hysteresis.

Tuning
Step 1 - Determine the Controller Filter Time Constant if Required
Inspect the response data to determine if the PV signal contains
excessive noise. Excessive signal noise is noise which cannot be
eliminated at its source, and if left unfiltered will result in excessive
control action and ultimately excessive wear and tear on the valve. You
will need to determine a controller filter constant to smooth the noise
before the controller acts on the signal.
NOTE
A controller filter not only smoothes the noise, it also adds lag to the
measurement.

In an integrating process, the filter lag can be effectively canceled by


controller derivative action which adds a lead to the controller.
Therefore, you will want to choose a filter constant which completely
eliminates the noise and PID control where the controller D action
cancels the controller filter dynamics.

Application Manual 25
LOOP ANALYSIS - INTEGRATING

Figure 4.5 - Determining the Correct Controller Filter Constant

1. Click the PV tab to display the PV variable graph.


2. Visually inspect the noise content of the PV graph and decide if
controller filtering is required to smooth the PV signal. That is, is the
noise causing the closed loop control action to excessively cycle the
valve during normal steady state operating conditions.
The decision to use controller filtering is a judgment call.
3. Click the Filter tab to call the Filter dialog box, select the filter type
used in your controller.
4. Start with a small filter time constant and increase it. For each filter
time entered, inspect the raw and filtered signals displayed until you
are satisfied that the controller filter setting completely eliminates the
signal noise.

NOTE
The Tuning Procedure does not use the filtered PV signal in the calculation of
the PID tuning parameters. The Tuning Procedure uses the unfiltered PV signal
to determine the process transfer function and adds the chosen filter constant in
the frequency domain to determine the tuning parameters.

Application Manual 26
LOOP ANALYSIS - INTEGRATING

Step 2 - Window Step Test of True Process Response


Windowing the step test response data is the most important operation
in the analysis procedure. It allows you to identify which step test best
represents the true response of the PV to a step input in the PD. Choosing
the largest PV response will produce the most conservative tuning
parameters.

Figure 4.6 - Windowing the Step Test Data for Tuning Analysis

1. Visually inspect the test data and determine the data that best
represents the true process response.
2. Place the left window marker where both the PV and the PD signals
are stable and the PD step change is made. Place the left window
marker where the PV step appears to stabilize at a steady state ramp.
Click Zoom to view the test data.
3. Refine both the right and left window markers to more precisely
define the response data.

Step 3 - Editing the PD and PV Data If Necessary

Application Manual 27
LOOP ANALYSIS - INTEGRATING

If the unfiltered PV or PD data in the tuning window appears to contain


excessive noise which distorts the data, the graphic editing function can
be used to remove the noise without distorting the true response data.
The Tuning Procedure uses the edited test data to determine the process
transfer function and the tuning parameters.

Figure 4.7 - Graphic Editing the Response Data

1. Click Edit on the menu bar and choose the graph you wish to edit.
2. Use the draw features to edit the graph to remove noise and load
disturbance data.

Step 4 - Execute the Tune Function


The Tune function calls the Tuning Conditions dialog box. Identify the
Loop Type as Integrating and be sure that the controller identification
options match the controller being used. See Appendix for more in-
depth information on controllers.

Application Manual 28
LOOP ANALYSIS - INTEGRATING

Figure 4.8 - Loop Analysis Report Screen

Application Manual 29
LOOP ANALYSIS - INTEGRATING

Application Manual 30
CHAPTER

5
Loop Analysis Report
The Loop Analysis Report documents the results of the Loop Analysis
test procedure. The Loop Analysis Report can be saved as a Record and
used to:
1. Select the optimum P, PI, or PID tuning parameters for the loop.
2. Model the process transfer function for simulation.
3. Analyze the combined and uncombined frequency plots.
4. Store text and graphic comments concerning the test results.
5. Print a Report.

Report Screen
Documents the test results and displays a table of P, PI, and PID tuning
parameters for the controller.

Figure 5.1 - Loop Analysis Report Screen

Application Manual 31
LOOP ANALYSIS REPORT

Choosing the Tuning Parameters from the Table


Setpoint or Load Response
Setpoint response tuning parameters are recommended for control loops
that receive a continuous setpoint change, such as loops configured as
secondary loops in cascade control systems. The Load response tuning is
recommended for loops which have a constant setpoint entered by the
operator.

P, PI, or PID
Loop Type Process Type Suggested Optional
Parameter Parameters
Self- Flow PI 1. PI with DRS (Deadband
Regulating Reset Scheduling for loops
with valves with stick-slip
problems)
2. I only when P gain< 0.1
(PB>1000% (For I only tuning
parameters specify Parallel
algorithm and use the I setting
for PI tuning selections)
Small PI PI with DRS(Deadband Reset
Volume Scheduling for loops with
Pressure valves with stick-slip
problems)
Temperature PID PI control when measurement
is noisy.
Consistency PI
Integrating Level PI 1. I-PD response
2. PI with e2 on I action (For
loops with hysteresis in
valves)
PI with e2 on PI action (For
control of surge and averaging
vessels)
Large Volume PI PID
Pressure
Batch PID 1. P+D only
Temperature 2. PID with conditional integral
configuration

Table 5.1 - Recommended Controller Actions

Application Manual 32
LOOP ANALYSIS REPORT

Slow, Medium or Fast Response


The selection of the Slow, Medium, or Fast response tuning parameters
is based on how you want the loop to respond in closed loop and the
amount of safety factor you feel is required to compensate for the
installed nonlinearities in the installed characteristic. In general, the less
you know about the individual process tested, the larger the required
safety factor (slower tuning), to insure stability under all operating
conditions. The tuning parameters in the table meet the most
conservative of the following specifications.

Response Response Minimum Minimum Minimum


To: Gain Phase Damping
Margin Margin Factor
Load Fast 2.50 400 0.30
Setpoint Fast 2.81 450 0.35
Load Medium 3.98 500 0.60
Setpoint Medium 4.46 550 0.65
Load Slow 6.31 600 0.90
Setpoint Slow 7.08 700 1.00
Table 5.2 - Gain Margin, Phase Margin, and Damping Factors for Protuner
Tuning Table

Application Manual 33
LOOP ANALYSIS REPORT

Loop Plot Screen

Figure 5.2 - Plot Screen Displaying Step Test Data Used For Tuning

Figure 5.2 Displays the step test data windowed for analysis. The PV
graph displays the Raw, Edited, and Filtered PV graphs. Edit the test
Notes and appropriate comments to document the findings.

Noise and Hysteresis


The Noise and Hysteresis tabs display the Noise and Hysteresis Analysis
screens. You can use these screens to either recalculate the noise and
hysteresis or add Comments to document the test results.

Application Manual 34
LOOP ANALYSIS REPORT

Model Tab
The Model tab calls the Process Model screen which displays the Bode
Plots of the process overlaid with the Bode Plots of the first order model.
To refine the process model constants, click Model to call the Process
Model dialog box. Enter the dynamic model constants until the Bode
Gain plots fit and then the DT (deadtime) constant until the Phase Plots
fit.

Figure 5.3 - Model Screen with Model Dialog Box Displayed

Bode Plots
The Bode Plots tab displays the two Bode Plots on a single graph. The
Bode Gain Plot is a plot of process gain in db vs. log frequency in radians
per second. The Bode Phase Plot is a plot of phase shift in degrees vs. log
frequency in radians per second. You can switch between the Process
Only (Uncombined) and Process + Controller (Combined) plots by
clicking on the Combined check box on the Cursor menu bar.

Application Manual 35
LOOP ANALYSIS REPORT

Bode Plots - Process Only (Uncombined)

Figure 5.4 - Bode Plots - Process Only (Uncombined)

The Process Only (Uncombined) Bode Plots represent the solutions to the
process transfer function. The cursor function can be used to display any
value on the plot including the Ultimate Gain and Ultimate Period,
corner frequency. You can use the plots to confirm the Protunertm tuning
parameters and to calculate alternate PID parameters.

Application Manual 36
LOOP ANALYSIS REPORT

Bode Plots - Process + Controller (Combined)

Figure 5.5 - Process + Controller (Combined) Bode Plots - Self-Regulating


Process
The Process + Controller (Combined) Bode Plots display the transfer
function of P(s) x C(s). The Combined Bode Plots provide a valuable tool
to evaluate the PID tuning parameters used for the control of self-
regulating processes. The controller filter along with the controller I and
D tuning parameters are evaluated by simply viewing the Combined
Bode Gain Plot. If the controller filter plus the I and D perfectly cancel
the process poles, the Combined Bode Gain plot will be a straight line
with a slope of minus 20db per decade. You can use the cursor function
to evaluate the closed loop gain margin, phase margin, and the
undamped natural frequency and period of the controller in closed loop.

Application Manual 37
LOOP ANALYSIS REPORT

Bode Gain and Bode Phase Plots


The Bode Gain and Bode Phase display the same Bode Plots discussed
earlier in this Chapter, except the two plots are displayed on separate
graphs.

Nichols Plots
The Nichols Plots displays the Nichols Plot graph of the process transfer
function. The Nichols Plot is a plot of process gain in db vs. phase shift in
degrees with frequency in radians per second as the independent
variable. You can switch between the Process Only (Uncombined) and
Process + Controller (Combined) plot by clicking on the Combined check
box on the Cursor menu bar.

Nichols Plot - Process Only (Uncombined)

Figure 5.6 - Nichols Plot - Process Only (Uncombined)

Application Manual 38
LOOP ANALYSIS REPORT

The Process Only (Uncombined) Nichols Plot represents the solutions to


the process transfer function. The cursor function can be used to display
any value on the plot including the Ultimate Gain and Ultimate Period.
The Nichols Plot is a valuable tool to evaluate the process transfer
function and the quality of the time domain test data used in the
analysis.

Nichols Plot - Process + Controller (Combined)

Figure 5.7 - Nichols Plot Process + Controller (Combined)

The Process + Controller (Combined) Nichols Plot displays the transfer


function of P(s) x C(s). The Combined Bode Plots display is a valuable
tool to evaluate the PID tuning parameters used for control of self-
regulating processes. You can use the cursor to evaluate the closed loop
gain margin, phase margin, undamped natural frequency, and period of
the controller in closed loop.

Application Manual 39
LOOP ANALYSIS REPORT

Nyquist Plots
The Nyquist Plots displays the Nyquist Plot of the process transfer
function. The Nyquist Plot is a circular plot of process gain vs. phase shift
in degrees with frequency in radians per second as the independent
variable. You can switch between the Process Only (Uncombined) and
Process + Controller (Combined) plot by clicking on the Combined check
box on the Cursor menu bar.

Nyquist Plot - Process Only (Uncombined)

Figure 5.8 - Nyquist Plot - Process Only (Uncombined)


The Process Only (Uncombined) Nyquist Plot represents the solutions to
the process transfer function. The cursor function can be used to display
any value on the plot including the Ultimate Gain and Ultimate Period.
The Nyquist Plot is a valuable tool to evaluate the process transfer
function and the quality of the time domain test data used in the
analysis.

Application Manual 40
LOOP ANALYSIS REPORT

Nyquist Plot - Process + Controller (Combined)

Figure 5.9 - Nyquist Plot - Process + Controller (Combined)

The Combined Nyquist plot is a valuable tool in evaluating the gain


margin, phase margin, and damping factor of the closed loop system.
Using this graph you can do M-circle evaluations for robustness and
closed loop damping factor evaluation on a set of proposed tuning
parameters.

Application Manual 41
LOOP ANALYSIS REPORT

Application Manual 42
CHAPTER
SPECIAL NONLINEAR PID CONTROLLERS
6
Special Nonlinear PID Controllers
Where to Apply Them
The Loop Analysis Report screen lists a table of P, PI, and PID tuning
parameters. From the table, you select P, PI, or PID tuning parameters
for slow, medium, or fast response to remote setpoint or load changes in
accordance with the recommendations outlined in Table 5.2 of this
manual. Along with the optimum PID tuning parameters and the
controller filter time constant selection, you must also configure your
controller to work as one of the various nonlinear PID controllers
discussed in this section. Many digital controllers have listed the
available controller types in their configuration options. In other cases,
you may need to program the controller to emulate the controller
algorithms presented in this section.

PID, PI-D, and I-PD


Error Reference on Setpoint Options
In general, a controller has many different requirements. It should have
good transient response to setpoint changes and reject load disturbances.
In the textbook PID controller, an attempt is made to satisfy all the
demands with a single mechanism. (Such controllers are called one
degree of freedom controllers). The introduction of the derivative on PV
and the proportional on PV control options allow for different paths for
the setpoint and load disturbance responses (two degrees of freedom
controllers), which add more flexibility to satisfy the control objectives.
Many controller brands allow the user to select if the controller P and D
tuning parameters operate on (SP-PV) or (-PV). Thus, all the error
reference configurations work exactly the same on load upsets, but
produce dramatically different output changes when a setpoint change is
made.

PID Type
SPECIAL NONLINEAR PID CONTROLLERS

de
e dt + Td dt
1
Output = Kp e +
Ti
PI-D Type
d ( PV )
e dt + Td
1
Output = Kp e +
Ti dt
I-PD Type
d ( PV )

1
Output = Kp PV + e dt + Td
Ti dt
Where:
Kp = Controller Gain
Ti = Integral in Time per Repeat
Td = Derivative in Time
e = Error (SP-PV)
PV = Change in PV
Equation 6.1 - PID, PI-D, and I-PD Controller Types

The PID controller equations illustrated in Equation 6.1 are a simplified


differential equation of how error reference values are implemented.

PID Type
As you can see from the formula for PID implementation, all three PID
terms act on the error. The controller proportional action acts on the
magnitude of the error, and the derivative action acts on the rate of
change of the error. Since setpoint changes are typically abrupt, it is
undesirable in most applications to have derivative action act on setpoint
changes. In normal applications, it is undesirable to kick the valve open
on a small setpoint change when derivative action is used.

PI-D Type
The PI-D type in some controller manuals is referred to as derivative on
PV. In the PI-D implementation, the derivative acts on -PV and not
error. Thus, a step change in the setpoint does not result in an
undesirable kick in the controller output. This implementation is
preferred anytime derivative action is used.
SPECIAL NONLINEAR PID CONTROLLERS

I-PD Type
In the I-PD implementation, both the controller proportional and
derivative act on -PV and not on error. Thus, a step change in the
setpoint does not result in either a proportional or derivative kick in the
controller output. In self-regulating processes, when the process variable
measurement signal contains measurement lag the I-PD controller
prevents setpoint overshoot.
Many self-regulating processes contain lag in the measurement. In
temperature control loops the thermowell adds lag to the temperature
measurement. In flow loops, filtering in either the controller or
transmitter, add lag to the measurement. In both these examples, the
actual measurement of the controlled variable is not seen by the control
system due to the lag in the measurement. The controller tuning will
calculate a large controller gain because the measurement lag will be
much larger than the process deadtime. A step change in setpoint using
a standard PID controller will appear to provide very good control, but
in fact, the real process variable follows the valve position and will
overshoot the setpoint.
In level control loops, the proportional gain setting is often very large.
Even a small setpoint change with a standard PID controller will result
in a very large change in the valve position. If the controlled variable,
which is either feeding the tank or controlling the flow from the tank
cannot stand the large kick, the I-PD controller should be used. Therefore,
you will need to implement the I-PD controller to eliminate the
unmeasured overshoot.
The I-PD controller implementation protects the system from abrupt
operator entered setpoint changes and should be used in all applications,
unless; there is a reason that a fast response to a setpoint change is
required, or no concern with overshoot in the real measurement, or
concern about large and fast changes in the valve position that may
upset other processes in the system.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.1 - Normal PID Response to a Setpoint Change With Filtering in the
Controller

Figure 6.1 illustrates the overshoot in the real unfiltered PV signal and
how the filtered PV variable appearing on the screen appears to provide
stable response with no overshoot when PV filtering is present and a
normal linear PID controller is used.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.2 - I-PD Response to a Setpoint Change With Filtering in the


Controller

Figure 6.2 clearly illustrates the benefits of the I-PD controller when
controller filtering is used. The response of the unfiltered PV to a
setpoint change is fast and responsive with no overshoot of the setpoint.
The filtered PV signal (as seen by the operator) appears to be slow.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.3 - PID Response to a Setpoint Change on a Level Loop

Figure 6.3 illustrates the response to a setpoint change on a level loop


utilizing a standard PID controller. Level loops typically require a large
controller proportional gain and a slow integral. With a normal PID
controller, the large proportional gain setting results in a very large
change in the controller output in response to a step change in the
setpoint. In many applications, this is unacceptable.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.4 - I-PD Response to a Setpoint Change on a Level Loop

Figure 6.4 illustrates the response of a typical level loop to a setpoint


change utilizing the I-PD controller algorithm configuration. As
illustrated, the I-PD algorithm makes the response to a step change in
setpoint very slow, preventing the large change in the controller output.
In many applications, this response minimizes upsets in other loops
being controlled and is the preferred response.

Error Gap Controller


To Solve Rangeability Problems
In a gap controller, a deadband is placed around the error. If the error is
in the gap range, the controller error is set to zero. The error calculations
in a gap action controller are:
SPECIAL NONLINEAR PID CONTROLLERS

de

1
Output = Kp e + e dt + Td
Ti dt
e = (SP PV)
if e egap Then e = 0
if e > egap Then e = e
egap = deadband%
Equation 6.2 - Error Calculation
In some applications, where both wide rangeability and precise control
are required, a small valve and a large valve are used. The small valve is
used to provide the precision and the large valve is used to provide the
rangeability.

Test 1
Protuner Step FIC with GAP
1600PC
in manual

FIC

Test 1
Response in PV

Test 2
SP = 50%
Response in
PV (FIC PD)

GAP Test 2
Step GAP with
FIC in Auto

FT
1

Figure 6.5 - Error Gap Action Controller Application for Large Valve, Small
Valve Rangeability Problem

As illustrated in Figure 6.5, a small valve is used to control the flow. The
gap action controller is used to control the large valve position to keep
SPECIAL NONLINEAR PID CONTROLLERS

the small valve operating near mid-range. The correct tuning procedure
is:
1. Connect the Protunertm to record the three control signals as
illustrated.
2. Record a Loop Analysis test on the FIC loop in manual, calculate the
PID tuning parameters, enter the parameters in the controller, and
place FIC back in automatic control.
3. Place the gap controller in manual and record a Loop Analysis test
on the GAP loop, calculate the PID tuning parameters, enter the
parameters in the controller.
4. Along with the PID tuning parameters, enter an error gap of 10%
into the GAP controller algorithm configuration and set its setpoint
to 50%.

In closed loop operation, the flow controller will control the flow and the
small valve will operate from 40% to 60% travel. If the system demand is
such that the small valve position is required to go below 40% or above
60%, the GAP controller will cause the big valve to make a position
change to bring the small valve back into the desired operating range.
Therefore, the tuning procedure and the tuning parameters are
determined in the same manner as with normal controllers. The PID
controller with the gap function is required to keep the system stable and
to prevent cycling due to interaction.

Error Squared PI Controller


Control of Surge and Averaging Level Loops
It is possible to create a controller with a continuous nonlinear function
whose control action increases with error. This type of controller is called
an error squared controller. The error squared controller configuration
controls averaging level and surge level control systems.
The Error Squared controller is only used for PI control modes and not
PID control. There are two ways to correctly implement the error
squared controller:
SPECIAL NONLINEAR PID CONTROLLERS

e e e
2
Series or Ideal Algorithms Output = Kp
100 + e dt

10000Ti
2
e e e
Parallel Algorithm Output = K p + e dt
100 10000Ti
Where:
|e| = Absolute Value of e to maintain direction
Kp = Controller Gain
Ti = Controller Integral Time
e = Error (SP-PV)

Equation 6.3 - Correct Implementation of the Error Squared PI Control


Algorithm for Integrating Processes

The equations in Equation 6.3 illustrate the correct implementation of


the Error Squared controller algorithm for a PI controller used for the
control of surge and averaging tank level control. Notice, the effective
controller gain becomes smaller as the error approaches zero, the
effective integral time becomes longer. Thus, the effective controller gain,
times the effective controller integral time, remains constant as a function
of error and the tuning parameters remain stable for the integrating
process being controlled.

NOTE
There are a number of control system vendors that offer an Error Squared PI
controller where the implementation is not correct for the control of integrating
processes. In some controller implementations, (e*e)/100 is simply substituted
for the error term in the standard PI controller. This implementation is of course
incorrect, and will result in unstable control because the effective gain, times the
effective integral time, does not remain constant as a function of error. A
number of DCS manufacturers implement an error squared on gain controller.
When using this controller implementation to control surge vessels, be sure to
use P only tuning because any I tuning parameter will result in unstable
control and cyclic control. For an integrating process to be stable, the controller
gain must be able to balance the process faster than the integral can ramp the
valve open. Thus, the integral time is inversely proportional to the controller
gain. It is essential that you check that the PI controller implementation
configured by the manufacturer is correct before using.
SPECIAL NONLINEAR PID CONTROLLERS

Averaging Level Control

In an averaging level control system, you want to both minimize the


movement of the level control valve and have the level control loop settle
out at setpoint after a load disturbance.

Serial Cable
Protuner
Flow In

Protuner
1600PC

Return to PI Controller
Setpoint with Error
after load Squared
change Algorithm

LT PD
LIC
PV

Minimize
Valve Movement

Flow Out

Figure 6.6 - Averaging Level Control Loop


Figure 6.6 illustrates an averaging level control loop on a mixing tank. In
this application, a number of input flows are blended in the tank. The
control objective is to minimize movement of the valve, thus minimize
flow disturbances to the process downstream of the vessel. The other
control objective is not allow the tank to go empty, or overflow during
large disturbances, but to return the level to setpoint after a change in the
load to insure proper mixing in the tank.
The Loop Analysis test procedure for calculating the optimum tuning
parameters of the level controller LIC, uses the same test and analysis
procedures as an ordinary integrating level control loop. Using the
Protunertm calculated tuning parameters displayed on the Loop Analysis
SPECIAL NONLINEAR PID CONTROLLERS

Report screen, and the Error Squared Controller implementation for the
PI controller, will result in the desired control.

Surge Level Control


Surge tanks are intended to absorb process upsets as well as average out
load disturbances in the downstream process. For surge control systems,
there is no need to return the level to setpoint, doing so would actually
reduce the effective capacity of the tank.

Serial Cable
Protuner
Flow In

Level at
High In
Flow Rate Protuner
1600PC
Ponly or
Setpoint PI Controller
with Error
Squared
Algorithm

Level at
Low In LT PD
Flow Rate LIC
PV

Minimize
Valve Movement

Flow Out

Figure 6.7 - Surge Level Control


Consider for example, a surge tank as illustrated in Figure 6.7 feeding a
downstream process. If the production is high, the level in the surge tank
should also be high, the most likely event is a decrease in the inflow to
the tank, then the entire volume of the tank is available to feed the
downstream process. Conversely, if the production is low, the tank level
should be low, so that the maximum volume is available to absorb an
input change which is certain to be positive.
The Loop Analysis test procedure for calculating the optimum tuning
parameters, uses the same test and analysis procedures as an ordinary
SPECIAL NONLINEAR PID CONTROLLERS

integrating level control loop. For surge level control loops, implement P
only control utilizing the Error Squared PI controller.

NOTE
In commissioning any P only controller, you must use the manual reset
function (output bias) to insure that the level is at setpoint under normal load
conditions.

While the Error Squared controller is useful in the control of averaging


and surge vessel control, it is not recommended on boilers, reboilers, and
other vessels where thermal and hydraulic effects are prominent, and
tight control is required under all operating conditions.

Error Squared on I Controller


Solving Hysteresis Cycling Problems in Level Loops
The Error Squared on Integral implementation is another nonlinear PID
controller configuration developed to prevent integral or hysteresis plus
deadband cycling in level control loops. The Error Squared on Integral is
implemented as follows:

Series or Ideal Algorithms


ee
100 dt )
1
Output = Kp(e +
Ti
Parallel Algorithm

ee
100
1
Output = Kp e +
Ti
Where:
|e| = Absolute Value of e to maintain direction
Kp = Controller Gain
Ti = Controller Integral Time

Equation 6.4 - Error Squared on Integral Controller Algorithms

Error Squared on Integral controllers is typically used on level control


loops where fast controller proportional action is required to arrest load
upsets, making the flow out of the tank equal to the flow into the tank.
Integral action in the controller is required to bring the level back to
SPECIAL NONLINEAR PID CONTROLLERS

setpoint. If you are tuning such a process, and the control valve contains
hysteresis or deadband, the integral action in the PI controller will cause
continuous cycling at steady state. Figure 6.8 illustrates the closed loop
response of a level control loop where the control valve has a 2%
deadband subjected to load upset. The controller is a standard PI
controller tuned with the optimum Protunertm calculated PI tuning
parameters.

Figure 6.8 - Linear PI Control of Level with 2% Deadband in the Valve

Figure 6.8 illustrates how integral action in a controller used to control


an integrating process will result in continuous steady state cycling if
there is deadband in the control valve. Making the integral setting
slower will change the frequency of the cycle, but will not eliminate it
completely. Turning off the controller integral action will eliminate the
cycling, but with no integral action in the controller the process variable
will never be at setpoint.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.9 - Error Squared on Integral to Eliminate Cycling

Figure 6.9 illustrates the closed loop response to a 10% load disturbance
utilizing the same PI tuning parameters as shown in the example in
Figure 6.8, implementing the Error Squared on Integral algorithm.
Comparing the results in the two graphs, you will notice that the Error
Squared on Integral controller both returns the process variable to
setpoint and eliminates the steady state cycling. Also, the integral action
in the Error Squared on Integral controller takes a much longer time to
eliminate the error, in that the integral time becomes infinitely slow as
the process variable approaches setpoint. Think of the Error Squared on
Integral controller as a P only controller with automatic manual reset.
SPECIAL NONLINEAR PID CONTROLLERS

Deadband Reset Scheduling Controller


Solving Stick-Slip Cycling
As many as one in five control loops demonstrate a continuous cycling at
steady state when tuned with the optimum PI or PID tuning parameters
calculated using the Protunertm. (The same cycling occurs if tuned using
Lamda, Ziegler Nichols Ultimate Sensitivity, and other tuning
calculation methods). In most cases, the cycling can be directly traced to
the nonlinear behavior of pneumatically actuated control valves. The two
most common types of motion nonlinear control valve responses are
hysteresis plus deadband and stick-slip.
As presented in Error Squared on I in this Chapter, hysteresis plus
deadband will cause steady state cycling in properly tuned integrating
loops, unless the valve is fixed or the Error Squared on Integral
controller configuration is used. Stick-slip action in the control valve will
result in steady state cycling in self-regulating loops unless the valve is
fixed or the Deadband Reset Scheduling controller configuration is used.
Stick-slip response is commonly observed in pneumatically actuated
control valves with positioners. To move the valve, the air pressure in
the actuator must be increased to overcome the friction in the actuator,
linkages, and the valve itself. The air pressure, from the positioner to the
actuator dome, will continue to increase without change in the valve
position during the stick phase. The stored energy in the actuator results
in the valve popping to a new position. This is the slip phase.
The new valve position is beyond the desired setpoint. Pneumatic
positioners are also nonlinear devices and thus contribute to the stick-
slip problem. At a constant ramp input signal to a positoner, the
positioner gain starts out small and thus loads the actuator slowly. When
the magnitude of the ramp input exceeds some predetermined value the
positioner gain increases and loads the actuator dome at a higher rate.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.10 - Typical Closed Loop Stick-Slip Cycling

Stick-slip cycling as illustrated in Figure 6.10 occurs when the controller


integral action continuously increases the controller output without a
corresponding change in the actual valve position. When the valve
finally does move, it pops and the process variable overshoots the
setpoint. At that point, the error becomes negative and the controller
integral action drives the output in the other direction. This results in the
distinctive continuous limit cycle known as a stick-slip cycle. The process
variable appears as a square wave oscillating around the setpoint. The
controller output appears as a triangular wave with a frequency
dependent on the tuning parameters, the valve, and the process gain.
There are three traditional solutions to the stick-slip cycling problem.
The first is to repair or replace the valve. A suspect valve, when removed
from service, will often pass the typical bench tests. Despite the verified
integrity of the valve, maintenance personnel will sometimes install a
new valve and discard the old one. This can be a very expensive
response to the stick-slip problem, and one that does not guarantee
success. The second is to place the controller in manual. This is an
SPECIAL NONLINEAR PID CONTROLLERS

effective approach to eliminate the cycling but is clearly unacceptable.


The third traditional solution is to de-tune the controller integral setting
such that the ramp rate as a function of error is so slow the stick-slip
cycle is eliminated. Unfortunately, substantial de-tuning of control loops
causes problems which are often more serious than the effects of the
cycling itself.
Deadband Reset Scheduling (DRS) is the name given to an algorithm in
which the controller integral setting is adjusted between a fast setting
and a slow setting depending on the size of the control error (SP-PV).

e dt )
1
Output = Kp(e +
Ti * Kr

Where:
if |e| errorgap then Kr = 10 to 20
else Kr = 1
Where:
Kp = Controller Gain
Ti = Controller Integral Time
Kr = Reset Ratio Factor (user adjustable typically
between 10 and 20)
e = Error (SP - PV)
errorgap = Width of Error Deadband around setpoint
Equation 6.5 - PI Time Domain Deadband Reset Scheduling (DRS) Equation

The DRS implementation, essentially increases the controller integral


time setting (Ti) when the actual error (e) is smaller than the errorgap,
thus, de-tuning the controller integral setting when the error is small.
With the controller de-tuned near setpoint, the rate at which the
controller integral action drives the controller output is too slow for a
stick-slip cycle to be maintained. When the controller is forced to
respond to an upset or setpoint change, the optimum tuning is used to
insure fast and stable response.
Determination of the errorgap and Kr constant is done as part of the
standard loop tuning procedure. After entering the Protunertm calculated
tuning parameters, the closed loop response of the process is recorded. If
a stick-slip cycle is detected, use the Statistical Analysis function to
document the range of the cycle on the PV signal. Set the errorgap equal to
SPECIAL NONLINEAR PID CONTROLLERS

the range of oscillation in the PV signal. Since the absolute value of the
error is compared to the errorgap this represents a safety factor of 2X.
Selection of Kr is made by simply observing the cycling at steady state
and increasing Kr until the cycle disappears. Typically, a Kr setting
between 10 to 20 is sufficient to stop the cycling.
Though not intended to replace good valve selection and maintenance
procedures, the Deadband Reset Scheduling DRS algorithm provides a
good compromise between the competing requirements of steady state
stability and speed of response.

Reset Gap Controller


Solving Cycling Problems when Using Motor Operated
Valves on Self-Regulating Processes

Standard integral action in a controller continuously changes the


controller output in an attempt to bring the measured error to zero. All
electric motor actuated valves have a deadband gap and a minimum on
and off time setting. Therefore, when the size of the error (valve position
setpoint - actual valve position) exceeds the deadband gap, the motor is
turned on and changes the valve position at least proportional to the
smallest on time increment.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.11 - Typical Closed Loop Cycle on Motor Operated Valve Using
Standard PI Controller

Figure 6.11 illustrates the closed loop control of water flow in a water
treatment plant. The final control element is a large butterfly valve with
an electric operator. The actuator was set up by the manufacturer with a
minimum deadband and cycle time. As you can see, the cycle looks very
similar to the stick-slip cycle discussed in Deadband Reset Scheduling
in this Chapter. The controller output, as a function of the Protunertm
calculated Medium PI tuning parameters, ramps the controller output to
eliminate the error and the motor turns on and makes the smallest
change possible and overshoots the setpoint.
The new error, causes the controller output to ramp in the opposite
direction. When the motor deadband is again overcome, the actuator
turns on and makes the smallest change. This cycle will continue forever,
or until the motor over heats and shuts off, or the actuator just wears out.
For this reason a continuous integral action in the controller is not
appropriate for this application.
SPECIAL NONLINEAR PID CONTROLLERS

To control this process use the Deaband Reset Scheduling (DRS)


algorithm shown in Equation 6.5 with the reset ratio factor Kr set to
999999 to turn the controller integral action off in the errorgap. Again, the
error gap is found by determining the size of the limit cycle in the PV in
the closed loop test with the Protunertm calculated tuning parameters in
the controller.

Conditional Integration
Solving Overshoot Problems During Setpoint Changes

Batch processes are typically integrating type processes with a large lag
time. The optimum control is a P+D action controller without I. Any I
action tuning, in a controller controlling an integrating process, will
result in an overshoot following a setpoint change. In real processes,
integral action is required following load disturbances to integrate any
small error to maintain the process variable at setpoint. The preferred
controller algorithm is a PID controller with conditional integration. The
conditional integration feature turns off the integral action when the
error is large, and turns on the integral action when the error is
sufficiently small.

e dt + T
1 de
Output = K p (e + d )
Ti dt
100
If e then Ti = Ti
Kp
else Ti = 9999999
Equation 6.6 - Conditional Integration Algorithm

Figure 6.12 illustrates the difference between the setpoint response in a


batch reactor with and without the use of the conditional integration
algorithm.
SPECIAL NONLINEAR PID CONTROLLERS

Figure 6.12 - Setpoint Response Comparing standard PID and a PID


controller with the Conditional Integral Algorithm Response

Split Range Control


Scaling the Protunertm
It is often necessary to configure control loops to control two valves with
a single controller. This technique is called split range control. Split
ranging can be accomplished in a number of ways. The most common is
to split range the signal using the valve positioner calibration. In many
digital control systems, split ranging can be done in software. The
purpose of this section is to cover the two most common split range
implementations and the correct methodology to scale the controller
output signals.
SPECIAL NONLINEAR PID CONTROLLERS

Split Range Implementation In Valve Positioners

Figure 6.13 illustrates a typical chemical reactor temperature control loop


configuration controlling both heating and cooling valves with a single
PID controller. The heating and cooling valves are split ranged in the
field.

Protuner
1600PC

PV
TT
1 TIC PD
M I/P

Return

Coolant

F0
3 - 9 PSIG
Heat
Exchanger

FC Heating
9 -15 PSIG Medium

Figure 6.13 - Split Ranged in Field


In this example, the temperature controller TIC sends a single output
from 0% to 100% to either heat or cool the reactor. The cooling valve is a
fail open with its positioner calibrated for 3 to 9 psi (0.2 to 0.5 bar). The
heating valve is a fail closed design with its positioner calibrated for 9 to
15 psi (0.5 to 1 bar). In this example, the Protunertm is connected to both
the controller PV and PD signals and scaled in the normal manner, this
temperature control loop is an integrating process. The open loop testing
is conducted by first allowing the process to come to a steady state
condition, placing the loop in manual and conducting a standard open
loop integrating step test both above and below 50% controller output to
insure that the loop dynamics are the same for both heating and cooling.
SPECIAL NONLINEAR PID CONTROLLERS

Split Range Implementation in a Digital Controller

Figure 6.14 illustrates a typical chemical reactor temperature control loop


configuration controlling both heating and cooling valves with a single
PID controller. In this example, the controller output is sent in software
to a pair of split range blocks and separate outputs are then sent to the
individual valves.

Protuner
1600PC

Split Range In Software


Requires Special PD Scaling
Input = 50% to 100% I/P
Output = 0% to 100%
PD2
PV PD
TT Input = 0% to 50% I/P
1 TIC Output = 0% to 100%
M
PD1
Return

Coolant

F0
3 - 15 PSIG
Heat
Exchanger

FC Heating
3 -15 PSIG Medium

Figure 6.14 - Split Ranged in the Controller

The Protunertm signal cables are connected as shown to record the


outputs to both valves. Scaling of the output signals is critical to insure
the correct process gain is recorded. To correctly scale the controller
output signals (PD1 and PD2) the following scaling procedure is
recommended:

1. Connect the Protunertm leads as shown.


2. Place the controller output at 50%.
3. Click the Channel corresponding to the cooling valve PD signal.
SPECIAL NONLINEAR PID CONTROLLERS

4. Click Calculate Voltage (Two Point Scaling) enter 50 in the


Controller Output 1.
5. Change the controller output to 45% and enter 45 in Controller
Output 2.
6. Click the Channel corresponding to the heating valve PD signal.
7. Place the controller output at 50%.
8. Click Calculate Voltage (Two Point Scaling) enter 50 in the
Controller Output 1.
9. Change the controller output to 55% and enter 55 in Controller
Output 2.

Scaling the PD signals in this manner will correctly record the percent
change in the controller output vs. the changes in the controller PV
signal. For changes below 50% controller output, the change in PD1 vs.
PV are analyzed, above 50% controller output, PD2 vs. PV are analyzed.
In many batch processes, heating and cooling result in substantially
different process gains. To test the loop use the following steps:
1. Place the controller in manual at 50% output.
2. Step the controller output to 60% and record the rate of temperature
increase.
3. Once the temperature increases, step the controller output back to
50% to stop the test. Use this test data to determine tuning
parameters for heating.
4. Step the output from 50% to 40% and record the rate at which the
process cools.
5. Step the contoller output back to 50% and use this data to calculate
the tuning parameters for the cooling mode.

In many batch applications it is necessary to use different PID tuning for


heating and cooling. Therefore, configure your controller to change PID
settings as a function of controller output.
CHAPTER

7
Cascade Control
Control System Analysis Test Procedure
By definition, cascade control systems consist of two or more control
loops in series. Cascade loops are invariably installed to prevent outside
disturbances from entering the process, and to put the principle
nonlinearities in a much faster responding inner loop controller.
The primary loop controller adjusts the secondary loops controller
setpoint. A cascade control system controls a single variable, and the
inner loop controller is used only to assist in achieving this.
In order for cascade control to be successful, the secondary loop dynamic
response must be much faster than the primary loop. A good rule of
thumb is that the primary loop should be three times slower than the
secondary. Otherwise, the primary loop will need to be de-tuned to such
a degree that cascade control will degrade the overall quality.
The following example illustrates a typical cascade system to control a
heat exchanger.

Protuner
1600PC

Test 2 -Primary with


Secondary in Auto
RSP
PD2
PV1
FIC PD1 TIC PV2
Test 1
FT
Secondary
1

Coolant Flow = Fc

FT TT TT
2 2
1

Process Fluid
Flow = Fp

Figure 7.1 - Preferred Protuner Connection to Test Cascade Control Loops

Application Manual 69
CASCADE CONTROL

The preferred Protunertm connection to test loops configured in cascade,


is to record the controller output (PD signal) and process variable (PV
signal) of both the primary and the secondary controllers as illustrated in
Figure 7.1. In many DCS systems, the PD signal from the primary
controller, which is the remote setpoint (RSP) to the secondary controller
is in software and not available as an analog signal. One solution, is to
program the DCS system to retransmit the controller output (PD) signal
as an analog signal to a spare output point. If for some reason, it is not
possible to connect the Protunertm to record the PD signal of the primary
controller, an alternate connection diagram is illustrated in Figure 7.2.

Protuner
1600PC

FIC
TIC Note:
No connection point for
PD2 is present. When analyzing
PV1 PD1 PV2 TIC step test results use data
PV1 in place of PD2

Test 1
FT
Secondary
1

Coolant Flow = Fc

FT TT TT
2 2
1

Process Fluid
Flow = Fp

Figure 7.2 - Alternate Protuner Connection to Test Cascade Control Loops

As illustrated in Figure 7.2, the alternate Protunertm connection to test


cascade control loops is to simply record the three available signals.

Application Manual 70
CASCADE CONTROL

Control System Analysis Test Procedure


Regardless of the connection type used, the Control System Analysis
procedure to test and optimize the closed loop operation of a cascade
control system is the same. The following outlines the proper procedure:
Test and tune the secondary control loop first.

1. The secondary control loop in this example, is the flow control loop.
Follow the test and analysis procedures in Chapter 3. Enter the
Protunertm tuning parameters in the secondary and place the loop
back in automatic in remote setpoint mode.
2. Next test and tune the primary controller.

The primary loop in this example is the temperature control loop. If the
Protunertm is connected using the alternate cascade connection as shown
in Figure 7.2, you will need to select the signals for analysis as shown:

Process Demand Signal: FIC-PV


(In the alternate connection, TIC-PD was not recorded.)
Process Variable Signal: TIC-PV
Table 7.1 - Signal Analysis

Protection Against Windup


In a single loop controller, reset windup or integrator saturation can
occur if the output saturates and the controller continues to integrate the
error. That is, if the controller output reaches 100% and the valve is fully
open, or if there is an output limit on the controller and the controller
continues to integrate the error. In these cases, the integrator portion of
the PID calculation can calculate some very large values for the controller
output, even though the actual controller output is limited. If this
happens, it can take sometime for the calculated controller output to get
back to an actual controller output value again, this can result in a very
large overshoot.
To prevent this, the controller PID algorithm must be written to
automatically stop the calculation of the controller integral action when
an output limit is reached. Most single loop PID controller algorithms
incorporate some form of anti-reset windup protection.

Application Manual 71
CASCADE CONTROL

In cascade control loops, reset windup or integrator saturation can occur


if the output of the secondary controller saturates and the primary
controller continues to integrate the error. To prevent this, it is necessary
to employ external anti-reset windup protection to automatically stop the
calculation of the primary controller integral action when an output limit
is reached on the secondary controller. In most DCS systems, cascade
controllers have external anti-reset windup protection as part of the
normal cascade control loop configuration. We have found in actual field
testing, that when single loop controllers are configured for cascade
control, external anti-reset windup protection is not setup properly,
Thus, large overshoots occur during startup in response to large setpoint
changes, and following large load disturbances, which could be
prevented.

Application Manual 72
CHAPTER

8
Ratio Control Systems
Implementation and Analysis
Ratio control systems are installed to maintain the relationship between
two variables to control a third variable. Ratio control systems are the
most elementary form of feedforward control. The system load is called
the wild flow and it may be uncontrolled, controlled independently, or
controlled by another controller that responds to variables of pressure,
level, etc.. Ratio control is applied almost exclusively to flows, and there
are correct and incorrect methods of implementation, both of which will
be addressed.
This section will address both the correct and incorrect implementation
of ratio where the ratio calculation is made outside the loop, and the
incorrect implementation where the calculation is made inside the loop.
Also presented is the proper sequence to test and optimize the system
installed dynamic operation.

Correct Implementation of Ratio Control


Consider in this first example, a control system designed to maintain a
certain ratio R of ingredient B and ingredient A:
B
R=
A
The more common and correct way to accomplish this means
manipulating the setpoint of the flow controller controlling the flow of
ingredient B (controlled flow) as a function of the desired ratio, and the
measured flow of ingredient A (wild flow). The setpoint of flow
controller B is calculated:
B = RA
Figure 8.1 illustrates such a system. In this example, the ratio of the two
ingredients is maintained by adjusting the setpoint of the flow controller
FIC-B, using an adjustable gain device known as a ratio station. The
input to the ratio station is the measurement of the wild flow A which is

Application Manual 73
RATIO CONTROL SYSTEMS

multiplied by a constant. The output of the ratio station is then the


setpoint to FIC-B to maintain the desired ratio for the two ingredients.

Wild Flow A

FT
A
A Ratio
X R

Setpoint = R*A
B

PVB
PD
FIC PDB
B

Protuner
1600PC

Controlled Flow B

Figure 8.1 - Ratio Calculation of Controlled Flow B to Controlled Flow A


Outside Closed Loop
Since the ratio calculation is done outside the control loop, it does not
interfere with the loop response.

Protunertm Analysis of the Installed System

The only feedback controller in the system is the controlled flow B. To


tune controller FIC-B connect the Protunertm to measure the input to the
controller PVB and the output of the controller PDB. Place FIC-B in
manual and record a series of step changes in the controller output and
the corresponding response of the flow PVB, in accordance with the
Protunertm test procedure for self-regulating processes documented in
Chapter 3. Use the Protunertm to analyze the test results, troubleshoot the
loop, and determine the tuning parameters for FIC-B. Enter the tuning
parameters from the Loop Analysis Report for response to setpoint to
FIC-B, and place the loop back in automatic control. Control system

Application Manual 74
RATIO CONTROL SYSTEMS

tuning requires tuning the flow control loop for response to setpoint
changes.
To insure proper operation of the system, additional testing should be
done to insure that the flow controller FIC-B can follow the changing
setpoint from the ratio station. Problems can occur when the wild flow A
is changing at a frequency in which no matter how well FIC-B is tuned, it
cannot follow the changes in setpoint. The wild flow A, which is the load
in the ratio control system, may be controlled independently or
manipulated by another controller that responds to other variables such
as level, pressure, etc.. Often, the cause of the cycling found in flow A is
caused by control problems in other loops. Therefore, to optimize the
system, these other loops will need to be identified, tested and
optimized.

Incorrect Implementation of Ratio Control


A second method to calculate the ratio R uses the individual flow
measurements A and B. The calculated ratio then becomes the process
variable measurement to the ratio controller with a manual setpoint.
Changes to the setpoint of the ratio controller RIC would then change the
ratio. Figure 8.2 illustrates such a configuration.

Wild Flow A

FT
A

A
PV = A/B
Ratio Ratio Setpoint
Calcualtor

B
PD
RIC

FT
B

Controlled Flow B

Figure 8.2 - Flow Ratio Divider Placed Inside Closed Loop

Application Manual 75
RATIO CONTROL SYSTEMS

The principle disadvantage of this system is that it places a divider inside


the closed loop. If flow B responds linearly with flow valve B, the process
gain of the loop will vary because of the divider. The following
differential equation explains why this is true:

dR 1 R
= =
dB A B

The process gain varies with the ratio and with flow B. In some cases, the
ratio would not be subject to change, but the process gain of the RIC ratio
control loop varies inversely with flow B. For conditions of low flow
then, the process gain of the RIC loop would be high and at high flow,
the process gain would be low. If RIC were tuned to provide good
response for low flow conditions, the loop response at high flow
conditions would be very sluggish. On the other hand, if RIC was tuned
for good closed loop response at high load conditions, the loop would be
unstable at low load conditions. If the ratio was inverted,
A
R=
B
then

dR A R
= 2 =
dB B B

the results are essentially the same.

The only advantage of using the ratio computing form of implementing a


ratio control system is that the controlled variable (flow ratio), is
constant, and can be recorded to verify control. In the correct
implementation illustrated in Figure 8.1 the two flows would have to be
compared for verification. This is a very poor reason to take what is an
essentially simple linear process and configure the control system as a
highly nonlinear process with major control problems. In short, the ratio
control configuration shown in Figure 8.2 is incorrect and should be
changed if encountered in the field.

Application Manual 76
RATIO CONTROL SYSTEMS

Cascade Ratio Control


This example illustrates two combinations of cascade and ratio control to
illustrate both the design of more complex ratio control systems, and the
proper use of the Protunertm for analyzing the control system dynamics.
Figure 8.3 illustrates a control system where a level controller
manipulates the wild flow and a composition controller manipulates the
flow ratio.

Wild Flow A

FT
A SP
PDL PVL

PVA
PDA LIC

Protuner
RSP
1600PC

FIC-A

X
PDC SP
Protuner
1600PC
RSP AIC

PVC

FIC-B AT

PVB LT
PDB

FT
B Controlled Flow B

Figure 8.3 - Level Controller Manipulates Flow and Composition Controller


Manipulates Ratio

In this system, the two flows A and B, are mixed to control a precise
composition of the combined flow C feeding into a storage vessel, used
as feed to another process. Liquid level in the tank is affected by total
flow, hence, the level controller LIC sets the setpoint of the wild flow
controller FIC-A, which in turn sets the setpoint of the controlled flow
controller FIC-B proportionately.

Application Manual 77
RATIO CONTROL SYSTEMS

Composition on the other hand is not affected by absolute value of either


flow, but only by their ratio. Therefore, to make a change in composition
(ratio), the composition controller AIC adjusts the multiplier of the ratio
station.

Analysis Test Procedure for Installed System

The first step in the analysis of the system is to be sure the proper flows
are chosen as the wild flow and the controlled flow. To minimize the
effect of composition control on the liquid level controlthrough its
proportional manipulation of flow Bflow B should be the smaller of the
two streams.
The second step is to connect the Protunertm to PDB and PVB to record the
input and output of flow controller FIC-B. With the Protunertm
connected, record an open loop analysis test to determine the tuning
parameters for FIC-B. Enter the parameters into the controller and place
the loop in automatic. The next loop to test is FIC-A. Connect the
Protunertm to record PVA and PDA, then test the loop to determine the
tuning parameters for FIC-A. Enter the parameters into the controller
and place the loop in automatic.
With the two flow loops in automatic, the next step is to optimize
operation of the composition controller AIC. Connect the Protunertm to
record the input and output of the composition controller (PVC), the
composition measurement signal, and PDC the ratio station multiplier.
With AIC in manual, record a series of step changes in PDC, and the
corresponding response of PVC, following the proper testing procedure
for testing self-regulating processes. Analysis of the test results will
determine the tuning parameters for entry to AIC.
The last loop to test in this example, is the level control loop LIC. Connect
the Protunertm to measure PVL, the process variable level measurement,
and the output PDL of the level controller, which is the remote setpoint to
FIC-A. The open loop test procedure requires that the loop first be
stabilized in automatic control to balance the system. Then, place LIC in
manual and record a series of step changes in PDL, and the response of
the level measurement PVL, in accordance with the test procedure for
analysis of integrating processes. Analyze the test results to determine
the tuning parameters for LIC.

Application Manual 78
RATIO CONTROL SYSTEMS

Ratio Control Optimization Summary


In each of the examples, the flow measurements need to be linear for the
installed system to be linear. Using head type meters without square root
extractors will result in very nonlinear control problems which can, and
should be, avoided. There are correct and incorrect ways of
implementing ratio control. The correct implementation is where the
ratio calculation is made outside the loop, and the incorrect configuration
is where the ratio calculation is made inside the loop. Unfortunately, the
incorrect configuration is all to commonly found in the field. When this
type of configuration is encountered, a lot of time can be wasted in
designing special nonlinear controllers which will never provide the best
control. The easiest solution, by far, is to reconfigure the loop to place the
ratio calculation outside the closed loop.
No matter how the ratio control is brought about, a computing device
must be used whose scaling requires some consideration. This area has
not been covered in this presentation but needs to be addressed in the
proper setup of any ratio control system. In pneumatic and analog
control systems, the scaling is often a problem. In modern digital
systems, however, most controllers can easily be configured for ratio
operation.

Application Manual 79
RATIO CONTROL SYSTEMS

Application Manual 80
CHAPTER

9
Selective Control Systems
There must be one final control element for each process variable to be
controlled in a system. In many systems however, the process variables
to be controlled outnumber the final control elements. In these
applications, the control system must automatically decide how to share
the final control elements. When this is the case, selective controls can be
employed to switch easily and smoothly between the variables to be
controlled. The purpose of this section is to provide:

1. Insight on how the various selective control strategies are properly


implemented and configured.
2. Guidance on the implementation of external anti-reset windup to
insure the unselected controller integral action is disabled when the
controller is unselected.
3. Information on potential control problems that might be encountered
when the digital control system being used to implement the
selective control strategy, employs the incremental or velocity, and
not the positional form of PID.
4. Protunertm control system analysis test techniques to troubleshoot the
field equipment and determine the optimum tuning parameters for
the controllers.

Selective controls involve the use of signal selectors which choose either
the lowest, median, or highest control signal from two or more signals.
Selective controls are employed in five basic application areas:

Protection of equipment
Variable structuring
Auctioneering
Redundant instrumentation
Valve position control

Each of these selective control application types has its own unique
implementation, control equipment requirements, and test procedure for
optimization.

Application Manual 81
SELECTIVE CONTROL SYSTEMS

Protection of Equipment
In many process control systems, there is a primary process variable that
needs to be controlled along with a second process variable that must not
be exceeded for reasons of economy, efficiency, or safety. The following
example illustrates how a typical selective control strategy is
implemented for the protection of equipment. In this example, the
control of the discharge from a compressor is controlled by manipulating
the setpoint of the motor speed controller. The two variables that require
control are the discharge air flow and pressure. The control strategy uses
two separate controllers and a low signal selector to decide which
controllers output signal will be used for control of motor speed.

PD4
<

PD2 PD3
PIC FIC
PV2 PV3

FT
PT
PV1
SIC

PD1 Discharge

Motor

Compressor

Figure 9.1 - Selective Control Strategy for Protection of Equipment

Figure 9.1 illustrates how the low signal selector is used to select the
lower output of either the pressure or flow controller to manipulate
motor speed. Under normal operating conditions, the discharge pressure
is below its setpoint and the flow controller output is less than the output
from the pressure controller, and is selected by the low signal selector to

Application Manual 82
SELECTIVE CONTROL SYSTEMS

change the motor speed, to control the discharge flow out of the
compressor. Thus, the pressure is allowed to drift below its setpoint
during normal or high load conditions.
During conditions of low loads on the compressor, the pressure
controller reaches its setpoint and its output becomes lower than the
output of the flow controller, and is allowed to assume control, thereby
lowering the flow and controlling the pressure at its setpoint. Decreasing
the motor speed decreases both the flow and the pressure, use of the low
signal selector guards the system against an excess of either.

Protection Against Windup

When one controller is selected from two or more, the others are in an
open loop condition. If the unselected controllers have integral action,
which is often the case, they need to be protected against windup. If not
protected, integral action in the unselected controller will cause its
output to saturate because its output is not able to adjust the final control
element to achieve its setpoint.
One solution to the windup problem is to modify the standard PID
equation by adding a term that integrates the difference between the
controller output and the output of the signal selector. When the
controller is not selected, the controller output and the output of the
signal selector will be different and integration occurs. The new term
cancels out the normal integral action, in effect integral action is stopped
until the controller resumes control. The output of the signal selector
must be fed back to the modified controller, hence the name external
integral feedback.
Unselected controllers can also be protected from windup by setting their
internal high and low limits from the output of the signal selector. The
high limit must be biased slightly above the selected output, and the low
limit slightly below, to avoid interference with the selected controller.
Unselected controllers will windup only to the extent of the bias, which
may result in a small overshoot during transition from one controller to
another.
If external anti-reset windup is not employed, there will be a large
overshoot in the controlled variable during the transition from one

Application Manual 83
SELECTIVE CONTROL SYSTEMS

controller to another, which may be unacceptable. It is therefore


necessary to understand how the controllers you are using work so that
you can be sure that external anti-reset windup is correctly installed.

Digital Algorithm Considerations When Configuring


Protective Selective Control Strategies

How well the selective control system actually works in the field can be a
function of the methodology that is employed in the implementation of
the digital PID algorithm in the controller brand you are using. Selective
controls are limited to the positional implementation of the PID
algorithm. Selecting a controller which uses the incremental or velocity
implementation of the PID algorithm is not equivalent and can lead to
serious problems which may not be easily recognized. This constitutes a
significant limitation in applying the incremental or velocity algorithm,
rather than the positional algorithm.
The first problem is associated with the sample intervals of the
controllers whose outputs are compared. If the signal selector samples
more often than one of the controllers, it will see an output of zero at
times when the control algorithm is not being processed. This is actually
false information, and causes the system to respond differently to
increasing and decreasing signals. In the case of a low selector, the
system will drive downscale at the shortest sample interval of any of the
associated controllers, and upscale at the longest interval. Roles are
reversed for a high selector. If your controller, uses the velocity or
increment form of the PID algorithm, all the controllers and the signal
selectors must sample precisely at the same interval, which is often not
the case. This problem can result in control anomalies that are not easily
recognized or overcome.
The second problem is associated with noise response. The incremental
shows that the output responds to changes in deviation from the last
sample, that is, e. Noise on the input signal will produce a e, even in
steady state, developing a proportional output. The change is even
more pronounced with derivative action. Ordinarily output to the valve
simply reflects the noise level, but a signal selector will pass output of
one sign and reject the next output of the opposite sign. As a result, the

Application Manual 84
SELECTIVE CONTROL SYSTEMS

changes induced by the noise are rectified by the selector, causing the
valve to be driven in the direction preferred by the selector.
The offset caused by noise on the process variable measurement signal
resulting from the noise rectification when an incremental or velocity
controller algorithm is employed, can be estimated by the following
formula:
Ti Td
offset = n 1 +
h h

Where Ti is the integral time, Td is the derivative time, h is the controller


scan rate, and n is the peak to peak noise level on the process variable
signal. As an example, a controller with 0.1 minute per repeat integral
setting, a 1 second scan rate, and a 1% signal noise on the controller
output, will result in an offset from setpoint of 6%. Offset error can be
reduced by reducing both the I and the D, increasing the sample interval
of the controller, or by using process variable measurement filtering to
eliminate the signal noise. However, it remains a fundamental defect of
the incremental or velocity implementation of the PID algorithm which
you need to be aware of when implementing selective controls.

Dynamic Analysis and Tuning

To analyze the dynamics of the installed system with the Protunertm,


requires that you test the system in the proper sequence. The following
steps outline the correct test sequence.

1. Test the system to determine the installed dynamics of the motor


speed control loop.
The motor speed controller case is the inner loop of the cascade
system and its closed loop response must be optimized before the
outer loop controllers are tested. Connect the Protunertm to record
the PV1 (motor speed) and PD1 (output of SIC to the motor). Place
SIC in manual and record the response of PV1 to a series of step
changes in PD1 in accordance with the open loop test procedure for
self-regulating processes as described in Chapter 3. In performing
loop analysis of the test data, pay particular attention to differences
in the rate of change of the motor speed to the step changes in the

Application Manual 85
SELECTIVE CONTROL SYSTEMS

controller output. Many speed controllers allow the user to adjust the
ramp rate of the output change. You will want to assure yourself that
the ramp change rate is not set too slow, and that it is programmed
for the same response for increasing and decreasing speed changes.
If you see that the response is too slow, or there is a difference in the
increasing and decreasing direction, make the appropriate changes
and retest the loop. Once you are convinced the open loop response
data is correct, window the test data and use the Tuning Procedure
to determine the optimum tuning parameters for SIC. Enter the
tuning parameters in SIC and place the loop in automatic closed loop
control.

2. Determine the open loop dynamics of the FIC and PIC controllers.
Connect the Protunertm to record the loops input and output signals.
The signals you will want to record are the PV2 (pressure signal), PV3
(flow signal), PD3 (output of pressure controller), PD2 output of flow
controller, and PD4 (output of the low signal selector which is the
RSP or remote setpoint to the speed controller). With the speed
controller in automatic, record a series of step changes in PD4 in
accordance with the test procedure for self-regulating processes and
the response of both the pressure PV2 and the flow PV3. Use the data
to perform two loop analysis tests. The first test to determine tuning
parameters for PIC, compare the step changes in PD4 to the response
in pressure PV2. The second test to determine the tuning parameters
for the flow controller FIC, compare the step changes in PD4 to the
flow response PV3. Enter the tuning parameters into the controller
and place the loops in automatic control.

With all loops in automatic control, with the Protunertm determined


tuning parameters entered in each controller, record the loop response to
a series of step changes in the setpoint of the selected controller. In
analysis of the data, check to be sure that the anti-reset windup is
operating correctly in the unselected controller by seeing that its output
signal does not integrate to 100% output, but acts as a P only controller
when unselected. If there is noise on the signal, check to be sure that the
noise does not result in an offset in setpoint due to the use of the
incremental or velocity algorithm in the control system.

NOTE

Application Manual 86
SELECTIVE CONTROL SYSTEMS

In some DCS systems, control signals PD2, PD3, and PD4 may be only software
signals and therefore unavailable to be recorded using the Protunertm from the
controller I/O rack. If this is the case, program the system to temporarily re-
transmit the required points to spare outputs.

Variable Structuring
Interlocks automatically shutdown control equipment when hazardous
conditions exist. Shutdowns can be avoided by designing the control
system to take corrective measures before the interlock condition is
reached and keeping the system running at a suboptimal level. The
implementation of this type of selective control system is called variable
structuring.

SP = 50%

< P = 4.0
PD1 When level is below 75%
(PB = 25%)
output is 100% and not
FIC selected.
When level is between
Ponly
75% and 100% output is
PV1 between 100% and 0% and
becomes selected.

FT

Feed

PV2

LIC
PD2 LT

Tails

Figure 9.2 - High Level Override Pinches Feedflow

Figure 9.2 illustrates the control of a distillation column base level control
system where the level is normally controlled by manipulating the tails

Application Manual 87
SELECTIVE CONTROL SYSTEMS

flow out of the column. If the tails line becomes restricted, the tails line
looses its ability to control the level and the level rises. The control
system then selects the feedflow as a secondary manipulated variable.
The output of the level transmitter is fed as both the normal process
variable signal to the bottoms level controller, and as an input to a
reverse acting P only controller with a proportional gain setting of 4. The
output of the P only controller is fed as a signal to the low signal selector
to perform the logic to override the normal feedflow controller and
throttle the feed flow valve to control level.
When the level transmitter output is below 75%, the output of the
proportional gain 4 controller is 100%, and therefore not selected by the
low signal selector. In this case, the control system is performing in a
normal manner. That is, the bottoms level is controlled by the level
controller and the column feed is controlled by the feedflow controller.
When the column level rises above 75% of transmitter span, the output of
the reverse acting proportional gain 4 controller starts to decrease below
100%, and will become 0% when the level is 100%. Somewhere in
between 75% and 100% of the level signal, the output of the proportional
gain 4 controller will be lower than the feedflow controller and will be
selected by the low signal selector. Thus, the feed to the column will be
reduced so that it just keeps up with the tails restriction, and the column
will not flood.
These types of variable structuring control systems are also called
overrides or soft constraint controls. They are used to avoid shutdowns and
are used in the implementation of startup systems. Analysis and tuning
of the flow and level controllers are accomplished using the Protunertm
by performing independent control loop analysis tests on the two
controllers, as if the override control system was not implemented.

Controller Design and Algorithm Considerations

As discussed previously, the effectiveness of the selective control system


is limited to the use of a control system where the PID control algorithms
are implemented in the positional form. Along with the previously
discussed problems with the incremental or velocity implementation of
the PID algorithm when used for selective controls, this type of
application presents another problem. That is, the incremental or velocity
form of the PID algorithm cannot be used for P only control when the P

Application Manual 88
SELECTIVE CONTROL SYSTEMS

only algorithm implementation does not contain a bias term. Therefore, a


P only controller without bias cannot be used to implement this type of
logic.
The incremental or velocity form of the PID algorithm looses track of its
position when integral action is not used. Some manufacturers use the
velocity or incremental algorithm for PI or PID control and the positional
algorithm when P or PD control is used. In any case, it is important that
you understand how the PID digital controller algorithm you are using is
implemented to prevent serious control problems which are not always
easily recognized. It is also important, that the flow controller be
protected against integral windup when unselected.

Dynamic Analysis and Tuning

In this example, the Protunertm would be used to test and tune the FIC
and LIC controllers individually. The following outlines the basic testing
procedure for the installed system. For more detail, refer to Chapters 3
and 4.
1. Connect the Protunertm to measure PD1 and PV1 (the output and
process variable measurement of the flow loop). Place the loop in
manual and record a series of step changes in the controller output
and the response of the flow following the test procedure for self-
regulating processes. Loop analysis of the test data will determine
any control or equipment problems and the optimum tuning
parameters for the controller. Enter the tuning parameters into the
flow controller and test the loop in automatic to verify the closed
loop response.
2. Determine the tuning parameters for the level controller.
Connect the Protunertm to measure the level controller output PD2
and the level signal PV2. Place the level controller in manual and
record step changes in the controller output following the test
procedure for integrating processes.

The control of a level loop is a mass balance problem. Therefore, for the
level to remain constant, the input flow to the system must be equal to
the output flow. It is often the case, when conducting a Protunertm loop
analysis test on a process, that the degree of plugging in the tails line is
unknown. Therefore, it would seem difficult to determine if slow,

Application Manual 89
SELECTIVE CONTROL SYSTEMS

medium, or fast tuning parameters should be selected to insure an


adequate gain and phase margin in the closed loop system.
The best rule to determine which tuning parameters should be selected is
a function of output PD% position of the loop when the system is in
balance at the time of testing. If the output PD% is between 10% and
30%, the tails line is unrestricted and you will want to select the slow
tuning parameters to insure adequate gain and phase margins when the
tails line becomes more restricted. If the output is between 30% and 60%,
the tails line is already potentially restricted and you would want to
select the medium tuning. If the PD% is between 60% and 90%, to
achieve steady state at the time of the testing, you will want to select the
fast tuning parameters.

Feed

Return

PV1
TT

PV2
> TT

TIC
PV3
TT
PD1

Coolent

Product

9.3 - Auctioneering of Process Variable Measurements

Auctioneering is a term used to describe a control system where a


controller selects the highest process variable measurement from a
battery of inputs. An example is the control of the highest temperature in
a reactor. The possibility exists that the location of the highest
temperature may shift. Figure 9.3 illustrates how a high signal selector is

Application Manual 90
SELECTIVE CONTROL SYSTEMS

used to control the peak reactor temperature. In this example,


temperatures all along the reactor are compared and the highest
temperature is used for control.

Loop Analysis Testing

To determine the best possible tuning parameters for the controller,


regardless of the temperature transmitter selected, connect the
Protunertm to measure all three temperatures as the process variable
signals, and the output of the temperature controller as the process
demand signal. Follow the Protunertm loop analysis test procedure, by
first recording the closed loop response of the system under both steady
state operating conditions, and the systems response to small setpoint
changes with the as found tuning parameters in the controller. Then, place
the temperature controller in manual and record the open loop response
of the various temperatures to a series of step changes in the controller
output.
Once the test data is collected, perform the loop analysis and Tuning
Procedure to calculate the tuning parameters for the temperature
controller for all three temperature transmitters. Compare the results of
all three sets of tuning parameters and choose the PID setting which will
assure fast, yet stable, control, regardless which temperature transmitter
is selected. If the process dynamics, and thus the optimum tuning
parameters, vary greatly depending on which transmitter is selected, you
may want to consider a controller which has the option of changing its
tuning parameters. In this type of controller, logic is employed to
download a specific set of tuning parameters based on which transmitter
is selected.

Redundant Instrumentation
Signal selectors are also used to protect a control system from instrument
failures by selecting the valid transmitter signal from among several. In
this section, two examples are presented on how selective controls can be
used to provide redundant process variable measurement signals.

Application Manual 91
SELECTIVE CONTROL SYSTEMS

Analyzers are generally less reliable than other instruments. Figure 9.4
illustrates a system that would allow control to be maintained in the
event of downscale failure of either analyzer. An upscale failure would
be allowed, which would shutdown the reactor, but in a safe condition.

>
AIC

PD1
PV1 PV2

Analyzer Analyzer

Reagent

Product

Feed

Figure 9.4 - High Selector Prevents Analyzer Failure From Damaging the
Reactor

If losing a transmitter is considered unacceptable, three transmitters


must be provided. Three outputs are then compared in a median
selector, which rejects the lowest and the highest signals.

Application Manual 92
SELECTIVE CONTROL SYSTEMS

Median
Selector
AIC
PV1 PV2 PV3
PD1

Analyzer Analyzer Analyzer

Reagent

Product

Feed

Figure 9.5 - Redundant Transmitter Protect Against Failure

Redundant transmitters are commonly used in a hostile environment,


(high temperature, corrosive, dirty, or vibrating surroundings), where
failure rates are high. Three transmitters measure the process, and the
median selector chooses the middle one for control. Redundant
transmitters, or sensors, and the use of medium selectors, keep the
controls working when a transmitter fails, avoiding costly shutdowns.

Loop Analysis Testing

Testing of selective control systems, configured to provide redundant


measurement, is conducted following the same procedure as if the
control system utilizes only a single transmitter. The exception is that the
Protunertm should be connected to measure all the various process
variable signals. On occasion, when recording the test data, a fault may
be found in the transmitter not currently selected. In analysis of the test
data, pay particular attention to both the static, as well as the dynamic,

Application Manual 93
SELECTIVE CONTROL SYSTEMS

response characteristics of the redundant measurements. A variance in


the dynamic response characteristics between the transmitter can
indicate faults in both the setup and installation.

Valve Position Control


In many control systems, there are several controlled variables fed from a
single controlled supply. To minimize energy usage, it is desirable to
keep the most open valve nearly wide open, but still in the controllable
range. Figure 9.6 illustrates an example where a variable speed fan is
used to supply air pressure to two parallel units. Ideally, the variable
speed fan should operate at minimum speed to satisfy both users.

PV1

PD1

FIC

PV3 FT

>
VPC

PD3

PV2
PD2

FIC

FT

Figure 9.6 - High Selector and VPC Minimize Energy

The control system configuration illustrated in Figure 9.6 is designed to


accomplish this. The output of the two flow loop controllers are sent to a
high signal selector which passes the highest signal as the process

Application Manual 94
SELECTIVE CONTROL SYSTEMS

variable signal to the valve position controller (VPC). With the valve
position controller setpoint at 90% (or some other preset value), the VPC
will slow the fan speed in an attempt to keep the most open valve at 90%.
This allows the most open valve to still be in its controllable range and
minimizes the power required to pass the flow demanded by the process.

Tuning the Control Loops

In tuning the control loops with the Protunertm, you must be aware that
the control configuration illustrated in Figure 9.6 introduces a
nonlinearity that needs to be understood in tuning parallel flow loops.
When valve position is controlled, the process gains of the flow loops
varies directly with flow. That is, if linear valves are used for the flow
valves, the installed characteristic will be equal percentage with a low
process gain at low system loads and a high process gain at high loads.
This is not a desirable characteristic for flow control, but an example of
how performance is sacrificed to save energy.
The first step in testing the system is to connect the Protunertm to the test
points illustrated. Place the loops in manual and record individual
control loop analysis step tests on the two flow control loops. Use the test
data to analyze the static and dynamic characteristics of the individual
flow loops.
The results of the Protunertm loop analysis on the individual loops
calculates a table of slow, medium, and fast tuning parameters for entry
into the individual flow controllers. If at the time of testing, the output of
the valve position controller is between 0% and 33% you should choose
the fast tuning parameters, if the output is between 33% and 66% you
should choose the medium tuning parameters, and if the output is
between 66% and 100% you should choose the slow tuning parameters.
The closed loop control will still be fast at low loads, and sluggish or
slow at high loads, but the flow loop tuning parameters will be
optimized based on the use of a linear PID controller.

After the flow loops are tuned and in automatic control, then record the
standard open loop control system analysis step test on the VPC
controller. In performing the test data, compare the step changes in drive
speed (PD3) to the changes in valve position (PV3). The Protunertm

Application Manual 95
SELECTIVE CONTROL SYSTEMS

output will again provide a table of tuning parameters for entry into the
controller.
Before choosing the tuning parameters for the controller, an
understanding of the closed loop requirements of the valve position
controller is required. First, the function of the valve position controller is
to save energy at steady state. Secondly, any changes in the output of
VPC, to keep the selected controller output at 90%, upsets the unselected
flow loops.
Therefore, you want the output of the valve position controller to change
its output slowly. You will want to choose the slowest PI tuning
parameters, or the equivalent I only tuning parameters, if you have the
option for an I only controller available. If you want to use I only control,
simply call Units from the Loop Analysis screen and change the
controller algorithm type to Parallel. Using the Parallel controller
algorithm, the I and D parameters are independent and the controller
gain, the integral setting for the slow PI tuning, can be entered into your I
only controller.

Valve Position Control Strategy


Eliminate Interaction and Linearity Problems

As discussed in the previous section, the use of a valve position


controller results in interaction between the variable speed fan and the
unselected loops, as well as a nonlinearity as a function of load on the
system. This section covers the control strategy where the system
demand frequently changes and sluggish response at high loads is
unacceptable. Figure 9.7 illustrates a modified valve position control
strategy designed to linearize the flow control loops and eliminate
interaction. To compensate for the loop gain variation, the differential
pressure is measured across each flow control valve. The square root of
the differential pressure is then divided into the actual controller output
of each flow controller as the control signal to the valve. Thus, the
denominator of the divider will vary inversely with load just as the loop
process gain varies directly with load. The loop process gain thus
remains constant with flow. The compensator allows fast tuning of the
flow loops and speeds up the response of the entire system by closing the
VPC loop without passing through the flow controllers. This allows the

Application Manual 96
SELECTIVE CONTROL SYSTEMS

selected valve position to return to setpoint following changes in its flow


setpoint, without disturbing the other flow loop.

PV1
PD1
FIC

PV3 FT

>
VPC

PD3 DP

PV2
PD2
FIC

FT

DP

Figure 9.7 - VPC Strategy With Dividers to Improve Response and Eliminate
Interaction

This configuration compensates for the nonlinearities and eliminates the


interaction in the system. The cost is additional dp transmitters for each
flow loop.

Conclusions On Implementing and Tuning


Selective Control Strategies
The function of a control system is to control what the plant produces.
Selective control systems, when installed, setup, and tuned properly,
allow the automatic control system to choose either the lowest, highest,

Application Manual 97
SELECTIVE CONTROL SYSTEMS

or medium control signal from two or more signals, switching easily and
smoothly between them.
Optimum control of these types of systems requires more than simply
tuning the controllers. Effective control requires system analysis testing
to determine the dynamics of both the selected and unselected variables,
correct implementation of anti-reset windup protection, and in the case
of digital controllers, the use of the positional controller algorithm.
Many of the problems identified in field testing installed selective control
systems are associated with; improper setup of the anti-reset windup
protection, hard to recognize problems associated with the use of control
systems which employ the incremental or velocity form of the PID
algorithm, equipment problems, and lack of knowledge of the true
system

Application Manual 98
CHAPTER

10
Feedforward Plus Feedback
Feedforward control is one of the most widely used advanced control
techniques in the process control industry. Feedforward offers many
advantages over feedback control alone. Feedforward can compensate
for load upsets before they are detected by the feedback control system
as an error. In contrast, feedback controllers can only react to correct for a
load upset after an error is detected between the process variable and the
setpoint. Properly tuned feedback/feedforward controllers can reduce
load disturbances to controlled process variable measurement by a factor
of 10, better than feedback control alone.

Basic Concept Of Feedforward Plus Feedback


Control
In a feedback control system, a load upset disturbs the process variable
measurement. The system stays upset until the feedback control brings
the process variable measurement back to setpoint. Feedforward control
can be used to improve the response of the system under these
circumstances. The basic principle of feedforward control is to measure
the disturbances as they occur, and to make adjustments to the process
demand signal of the feedback controller, thus, preventing the
disturbance from upsetting the process variable signal being controlled.

Application Manual 99
FEEDFORWARD PLUS FEEDBACK

Feedback Plus Feedforward Control Of


Self-Regulating Processes

Figure 10.1 - Block Diagram of Feedback Plus Feedforward Control System


of a Self-Regulating Process

As shown in Figure 10.1, the control system has two process demand
signals, that is, two variables that when changed effect the controlled
variable. The two process demand signals are PD1 (the controller output)
and PD2 (the measured load upset). The control system has only one
controlled process variable PV2.
In normal feedback control, the controller determines the error (SP1 -
PV1) and adjusts the controller output PD1, to bring the error to zero as a
function of the PID parameters in the controller. With feedforward
added to the control loop, the output of the controller PD1 is also
changed as a function of a measured load disturbance PD2.
The tuning parameters for the feedforward for self-regulating processes
are typically in the terms of gain, lead/lag, and delay. The feedforward
gain is how much, the lead/lag is how fast, and the delay is when to
change the valve position to correct for a changing load disturbance so
the measured process variable will not be disturbed.

Application Manual 100


FEEDFORWARD PLUS FEEDBACK

The block diagram in Figure 10.1 illustrates that the system can be
defined dynamically by two process transfer functions:

Process Transfer Relating To Changes In


Function Response of
P(s)11 PV1 PD1 (Controller Output)
P(s)12 PV1 PD2 (Load)
Table 10.1 - Variable Table

The process transfer functions are determined in the same manner


described in previous sections in this manual. That is, a series of step
changes are made in the process demand recording the response of the
process variable. One of the step changes, which appears to best describe
the response is Windowed and the Tuning Procedure is run. The Loop
Report displays a table of PID tuning parameters as well as the deadtime,
time constant, and process gain. Following the standard control loop
analysis testing procedure, you will determine.

Process Transfer Transfer Function


Function Constants
P(s)11 (Tuning PID Loop Test) PG11, TC11, DT11
P(s)12 (Tuning Load Upset vs. PV) PG12 TC12 DT12
Table 10.2 - First Order Transfer Functions from Loop Analysis Reports

The Feedforward Decoupler FF(s)12 tuning parameters are found as


follows:

Feedforward Tuning Constants

Feedforward Transfer Gain Lead Lag Delay


Decoupler Function (Seconds) (Seconds) (Seconds)
FF(s)12 P(s)12/ P(s)11 PG12/PG11 TC11 TC12 (DT12-DT11)

Table 10.3 - Calculating Feedforward Tuning Constants

Application Manual 101


FEEDFORWARD PLUS FEEDBACK

Example Of Determining Feedback Plus


Feedforward Tuning Parameters For The Control
Of A Heat Exchanger

Step Changes
=% in PD vs.
Valve Trim Temperature

+ SP
FT TIC
1
+

Coolant Flow
Gain
Lead
Lag
Delay
Step Changes
in Load vs.
Temperature
SP
FIC FT
TT
2
2

Process Fluid
Flow

Figure 10.2 - Heat Exchanger Control with Protuner Connections for


Determining Feedback and Feedforward Tuning Parameters

In this example, we are interested in controlling the outlet temperature of


the heat exchanger by manipulating the coolant valve. Assume that the
process fluid inlet temperature is constant and the primary load
disturbance is a change in flow rate of the process fluid.

Application Manual 102


FEEDFORWARD PLUS FEEDBACK

Recommended Connections for Protuner 1600PC

Channel Channel Name Description

1 TIC-PD Output of temperature controller to


steam valve
2 TIC-PV Controlled temperature
3 Feedflow-LD Process Feedflow (Load upset)
Table 10.4 - Protuner Connection for Tuning Feedback plus Feedforward

Figure 10.3 - Data Analysis Screen Illustrating the Three Recorded Signals

Application Manual 103


FEEDFORWARD PLUS FEEDBACK

Loop Analysis
Tuning Temperature Controller
To tune the temperature control loop and to determine the process
transfer function of both the process and the load disturbance, you will
need to perform two Loop Analysis procedures and save both
procedures as Records. Figures 10.4 and 10.5 display the two Loop
Analysis Reports generated by analysis of the test data illustrated in
Figure 10.3.

Figure 10.4 - Loop Analysis Report Tuning Temperature Control Loop

The Loop Analysis Report in Figure 10.4 contains both the tuning
parameters for the temperature controller and the first order model
constants in the Loop Signature.

Application Manual 104


FEEDFORWARD PLUS FEEDBACK

Figure 10.5 - Loop Analysis Report Tuning Load Upset vs. Temperature

The Loop Signature in the Loop Analysis Report in Figure 10.5 displays
the first order model constants that document the dynamic relationship
between the load upset in the feedflow and the response of the
temperature being controlled.

Calculating Feedforward Tuning Parameters


The feedforward tuning parameters, are calculated using the first order
process model constant found on the Loop Report screen illustrated in
Figures 10.4 and 10.5 above. From the Loop Signature in Figure 10.4 the
process transfer function constants of the temperature control loop are:

Process Gain PGp = -1.19 unitless


Deadtime DTp = 22.11 seconds
Time Constant TCp = 29.91 seconds
Table 10.5 - First Order Model Constants of Process

Application Manual 105


FEEDFORWARD PLUS FEEDBACK

From the Loop Signature in Figure 10.5, the process transfer function
constants for the load upset are:

Process Gain PGL = -1.92 unitless


Deadtime DTL = 45.89 seconds
Time Constant TCL = 36.37 seconds
Table 10.6 - First Order Model Constants of Load Upset
The feedforward tuning parameters are determined by dividing the
transfer function of the load upset by the transfer function of the process.

PG L e DTL
TC L s + 1 PG L (TC P s + 1)
Feedforward = DTP = ( DTL DTP )
PG p e PG p (TC L s + 1)
TC P s + 1

Feedforward Gain -PGL / -PGP 1.61


Lead (seconds) TCP 29.91
Lag (seconds) TCL 36.37
Delay (seconds) DTL - DTP 23.73
Table 10.7 - Calculating Feedforward Tuning Parameters

Enter the feedforward gain, lead/lag, and delay tuning parameters into
the temperature controller. If the delay is negative, enter 0 seconds for
feedforward tuning. The feedforward tuning will typically reduce the
temperature variance by a factor of 10 during changes in the feedflow.

Application Manual 106


FEEDFORWARD PLUS FEEDBACK

Figure 10.6 - Comparing Response to Load Upset With and Without


Feedforward

Application Manual 107


FEEDFORWARD PLUS FEEDBACK

Application Manual 108


CHAPTER

11
Multivariable
Relative Gain Array
Most large process plants have many process variables being controlled
and many process demand signals being manipulated. Ideally, we would
like to have a given process demand manipulated variable affect only its
controlled process variable. Unfortunately, in many cases, a change in
one process demand manipulated variable upsets other controlled
process variables in addition to its own controlled process variable. In
such multivariable control loops, coupling is said to exist. If coupling is
severe, a load disturbance will result in the second loop whenever the
process demand manipulated variable in the first loop changes.
If in addition to coupling from the first loop to the second loop there is
coupling from the second to the first, interaction is said to exist.
Interaction can cause oscillations and even instability. Since coupling or
interaction may exist in multivariable control systems, it is first of all
important which manipulated process demand signals should be
connected to which controlled process variables. It is possible that one
combination of variables may be better than another. However, no
combination may be satisfactory. The purpose of this section is:

1. To explain the Protunertm procedure for testing interactive control


systems.
2. To develop a method to determine the extent of interaction in a
multivariable control system.
3. To determine, based on the information in Step 2, proper pairings of
process demand manipulated variables and process variables to be
controlled.
4. To decide if no pairing is satisfactory and interaction is severe, that a
decoupling control strategy will be required.

Testing a 2x2 Interactive Control System

Application Manual 109


MULTIVARIABLE RELATIVE GAIN ARRAY

Figure 11.1 - Block Diagram of a 2X2 Interactive Control System

As illustrated, in the block diagram of the 2X2 interactive system, a


change in the output (PD1) of controller (C1) causes a change in both
measured process variables PV1 and PV2. Also, a change in the output
(PD2) of controller (C2) causes changes in both PV1 and PV2. The
Protunertm test procedure is as follows:

1. Connect the Protunertm to measure the variables outlined above.


2. Scale and document the input channels.
3. Place both controllers in manual (open loop).
4. Step PD1 in equal steps of approximately 3% to 10% up twice, down
three times and back to its original position. After each step in PD1,
wait for both PV1 (Channel 2) and PV2 (Channel 4) to stabilize at a
new steady state.
5. Step PD2 in equal steps of approximately 3% to 10% up twice, down
three times and back to its original position. After each step in PD1,
wait for both PV1 (Channel 2) and PV2 (Channel 4) to stabilize at a
new steady state.

Application Manual 110


MULTIVARIABLE RELATIVE GAIN ARRAY

Test Data Analysis


The test data gathered with the Protunertm
as outlined above, can now be
analyzed to determine the four process transfer functions that describe
the 2X2 interactive process.
In general, the four process transfer functions are found by windowing
the following test results and executing the Tuning Procedure on each of
the four separate windows.

Process Relates To Changes In


(Tuning Record) Response of
P(s)11 PV1 PD1
P(s)21 PV2 PD1
P(s)22 PV2 PD2
P(s)12 PV1 PD2
Table 11.1 - Loop Analysis Test Variables
The Loop Signature on the four individual Loop Analysis Report screens
display the process transfer function for each of the separate process
responses analyzed. The process model for each process is in the terms of
Process Gain (PG), Time Constant (TC) and Deadtime (DT).
The process transfer function model constants found on each of the Loop
Analysis Report screens are as follows:

Process Model Constants


P(s)11 PG11, TC11, DT11
P(s)21 PG21, TC21, DT21
P(s)22 PG22, TC22, DT22
P(s)12 PG12, TC12, DT12
Table 11.2 - Model Constants from Loop Analysis Tests

Application Manual 111


MULTIVARIABLE RELATIVE GAIN ARRAY

Relative Gain Array (RGA)


Proper Pairing Determination
In most applications you are working on, the individual controllers in the
system are already paired. That is, the decision has already been made
on which controller controls which valve, and which process variable.
The following RGA analysis will help you confirm that the pairing
chosen causes the least interaction when the loops are in automatic. The
purpose of the relative gain is to tell you how much the gain of one loop
changes when the other loops in the interacting system are closed.
The RGA Relative Gain Array matrix for a two-variable system is written
as follows:
PD1 PD2
PV1 RG11 1- RG11
PV 1- RG11 RG11

Where:
PG11 PG22
RG11 =
PG11 PG22 PG12 PG 21

Table 11.3 - Calculation of Relative Gain

NOTE
In calculating the relative gain (RG), it is very important that the sign of each
process gain (PG) used in the equation is correct. The process gain in the Loop
Signature is the absolute value of the process gain. To determine the correct
sign, look at the Plots to see the direction of the step change in the PV vs. the
direction of the PV response.

For each controlled process variable (PVi) the manipulated process


demand (PDi) is chosen having the largest positive relative gain. It is
possible for relative gains to be negative. If a PD and a PV with a
negative relative gain are paired, the system will be uncontrollable and
unstable, each variable will be driven to its limit. Poor control may also
result if the relative gain of the pairing is much larger than 1.

Application Manual 112


MULTIVARIABLE RELATIVE GAIN ARRAY

Example Of Relative Gain Analysis Of A 2x2


Blending System
As shown in Figure 11.2, two liquids are mixed in line to produce a
mixture of desired composition. The total flow is also to be controlled.
The problem is to find the proper pairing of PV controlled variables and
manipulated process demands PD's.

Protuner
1600PC

FT
PD1 FIC PV1

Valve 1 AT

PV2

Valve 2

PD2 AIC

Figure 11.2 - Protuner Connection to the Blending System

For testing purposes, controller C1 is chosen to control the total flow


(PV1) by controlling valve 1 (PD1). Controller C2 then controls the
mixture (PV2) by controlling valve 2 (PD2). The Protunertm is connected
as shown and described in Figure 11.2 above.
The analysis to determine P(s)11 is found by displaying only Channels 1
and 2 on the screen, Windowing in the step test, and running the Tuning
Procedure. P(s)21 is found from the test data on Channels 1 and 4, P(s)22
from test data on Channels 3 and 4, and P(s)12 from test data on Channels
3 and 2.

Application Manual 113


MULTIVARIABLE RELATIVE GAIN ARRAY

From the tuning parameter screens, the process gains for each test are as
follows:

PG11 = 1.0
PG21 = 1.0
PG22 = -0.15
PG12 = 0.35
Table 11.4 - Process Gains from Loop Analysis Tests

The relative gain RG11 is calculated as follows:

PG11 PG22 1 * ( .015)


RG11 = = = 0.3
PG11 PG22 PG12 PG 21 1 * ( 015
. ) (0.35 * 1)

PD1 PD2
PV1 0.3 0.7
PV2 0.7 0.3
Table 11.5 - Relative Gain Table
Since the elements with the largest relative gain are to be paired, this
matrix suggests that valve 2 (PD2) should be paired with the total flow
(PV1) and valve 1 (PD1) should be paired with the mixture (PV2). This is
opposite from the original pairing.
When both loops are in closed loop control the process gain of each loop
will change by; 1/.7 = 1.43 or 43%. Therefore, the tuning parameters
chosen from the Loop Reports screen for the appropriate pairings should
be chosen with at least a 43% gain margin. The medium tuning has a
gain margin of 100% this should provide adequate gain margin for good
control.

Application Manual 114


MULTIVARIABLE RELATIVE GAIN ARRAY

Tuning Coupled Loops


Once you have determined which valve should control which variable
using the relative gain technique, you will want to tune the controllers.
When loops fight each other as in this example, the technique for tuning
the loops is as follows:

1. With the analyzer loop in automatic, place the flow loop in manual
record an open loop test and determine the tuning parameters for the
flow loop. Enter the new PID tuning parameters in the flow loop and
place it back in automatic.
2. Place the analyzer loop in manual and record an open loop analysis
test with the flow loop in auto. Enter the new tuning parameters in
the analyzer loop and place it back in automatic.
3. Again, place the flow loop in manual and record a second open loop
tuning test on the flow loop. Calculate the new tuning parameters
and place the loop back in automatic.
4. Lastly, retest the analyzer loop to determine new tuning parameters
with the flow loop in automatic.

The tuning procedure is to tune each loop with the coupled loops in
automatic. Each time you tune the opposite coupled loop it changes the
tuning parameter of the other loop, it is necessary to go back and fourth
several times to decouple the loops with tuning.

NOTE
Because the loops are coupled, the tuning calculated using the above
methodology is only valid if both loops remain in automatic. If one of the loops is
placed in manual, the tuning parameters in the loop left in automatic will appear
very sluggish.

Application Manual 115


MULTIVARIABLE RELATIVE GAIN ARRAY

Application Manual 116


CHAPTER

12
Decoupling Interactive
Multivariable Control Systems
As discussed in Section 11, if the relative gains are numerically close to
each other, interaction fighting loops in a multivariable control system is
likely to be a problem, particularly if the response times are comparable.
In cases in which cross coupling is severe, the system can become
unstable, and decoupling will be required. This section, will cover how to
implement decoupling strategies in 2X2 and 3X3 interactive multivariable
control systems.

Decoupling A 2x2 Interactive Control System

Figure 12.1 - Block Diagram of a 2X2 Interactive Control System with


Feedforward Decouplers

A 2X2 interactive control system has two controllers, two final control
elements, and variables to be controlled. Interaction is present because
moving either valve causes both process variables to change.

Application Manual 117


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

The Protunertm testing procedure is the same as described in Section 11


The calculation of the decouplers is the same as the calculation of the
feedforward decouplers.
As illustrated, in the block diagram of the 2X2 interactive system, a
change in the output of controller C1 (PD1) causes a change in both
measured process variables PV1 and PV2. Also, a change in the output
(PD2) of controller C2 causes changes in both PV1 and PV2. The
Protunertm test procedure is as follows:
1. Connect the Protunertm to measure the input and output variables for
both loops.
2. Scale and document the input channels.
3. Place both controllers in manual (open loop).
4. Step PD1 in equal steps of approximately 3% to 5% up twice, down
three times and back to its original position. After each step in PD1,
wait for both PV1 and PV2 to stabilize at a new steady state.
5. Step PD2 in equal steps of approximately 3% to 10% up twice, down
three times and back to its original position. After each step in PD1,
wait for both PV1 and PV2 to stabilize at a new steady state value.

Test Data Analysis

The test data gathered with the Protunertm as outlined above, can now be
analyzed to determine the four process transfer functions that describe
the 2X2 interactive process. In general, the four process transfer functions
are found by performing four separate Loop Analysis routines and
saving the results as four separate Reports.

Process Relates To Changes In


(Tuning Record) Response of
P(s)11 PV1 PD1
P(s)21 PV2 PD1
P(s)22 PV2 PD2
P(s)12 PV1 PD2
Table 12.1 - Variables for Loop Analysis Testing 2X2 Process
The Tuning Procedure to determine P(s)11 and P(s)22 determines the PID
tuning parameters for controllers C1 and C2.

Application Manual 118


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

The Loop Signature of the Loop Report screen also displays the process
transfer function for each of the separate process response tests
windowed. The process model in each process is in the terms of Process
Gain (PG), Time Constant (TC) and Deadtime (DT). The process transfer
function model constants found on each of the Loop Reports are as
follows:
Process Model Constants
P(s)11 PG11, TC11, DT11
P(s)21 PG21, TC21, DT21
P(s)22 PG22, TC22, DT22
P(s)12 PG12, TC12, DT12
Table 12.2 - Model Constants from Loop Analysis Test Reports
The dynamic decouplers for the control system are determined as
follows:

Decoupler Tuning Constants

Decoupler Transfer Gain Lead Lag Delay


Function (Unitless) Seconds Seconds Seconds
D(s)12 P(s)12/P(s)11 PG12/PG11 TC11 TC12 (DT12 -DT11)
D(s)21 P(s)21/P(s)22 PG21/PG22 TC22 TC21 (DT21 - DT22)
Table 12.3 - Calculating Decoupler Tuning Parameters

NOTE
The sign for the individual process gains is not displayed on the Loop Analysis
Reports. Therefore, to insure the correct sign for each process gain, look at the
Plots of the test data. If the step change in the PD and the response of the PV are
in the same direction, the process gain is positive. If the response of the PV is in
the opposite direction of the PD step the process gain is negative.

Application Manual 119


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

Decoupling A 3x3 Interactive Control System

Figure 12.2 - Block Diagram of a 3X3 Interactive Control System with


Decouplers

A 3X3 interactive control system has three controllers, three final control
elements, and three variables to be controlled. Interaction is present
because moving any of the three valves causes all three process variables
to change. The Protunertm testing procedure is the same as described in
Decoupling a 2X2 Interactive Control System in this Chapter, except that
control loop analysis testing of all three loops is required. As illustrated,
in the block diagram of the 3X3 interactive system, a change in the
output controllers C1,C2, and C3 causes a change in all three measured
process variables PV1,PV2, and PV3.

Application Manual 120


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

The Protunertm test procedure is as follows:


1. Connect the Protunertm to measure the variables outlined above.
2. Scale and document the input channels.
3. Place all three controllers in manual (open loop).
4. Step PD1 in equal steps of approximately 3% to 5% up twice, down
three times and back to its original position. After each step in PD1
wait for PV2 (Channel 2), PV2 (Channel 4) and PV3 (Channel 5) to
stabilize at a new steady state.
5. Step PD2 in equal steps of approximately 3% to 10% up twice, down
three times and back to its original position. After each step in PD2,
wait for PV1 (Channel 2), PV2 (Channel 4) and PV3 (Channel 5) to
stabilize at a new steady state.
6. Step PD3 in equal steps of approximately 3% to 10% up twice, down
three times and back to its original position. After each step in PD3,
wait for PV1 (Channel 2), PV2 (Channel 4) and PV3 (Channel 5) to
stabilize at a new steady state.

Test Data Analysis


The test data, gathered with the Protunertm as outlined above, can now
be analyzed to determine the nine process transfer functions that
describe the 3X3 interactive process. In general, the nine process transfer
functions are found by performing three separate Loop Analysis routines
and saving the results in nine separate Loop Analysis Records.

Process Relates To Changes


Response of In
P(s)11 PV1 PD1
P(s)21 PV2 PD1
P(s)31 PV3 PD1
P(s)12 PV1 PD2
P(s)22 PV2 PD2
P(s)32 PV3 PD2
P(s)13 PV1 PD3
P(s)23 PV2 PD3
P(s)33 PV3 PD3
Table 12.4 - Variables for Loop Analysis Testing of 3X3 Process

Application Manual 121


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

The Tuning Procedures determine P(s)11, P(s)22, and P(s)33, also the PID
tuning parameters for controllers C1 C2,and C3. The Loop Analysis
Report screen displays the process transfer function for each of the
separate process response tests analyzed. The process model for each
process is in the terms of Process Gain (PG), Time Constant (TC) and
Deadtime (DT).
The process transfer function model constants found on each of the Loop
Analysis Report screens are as follows:
Process Model Constants
P(s)11 PG11, TC11, DT11
P(s)21 PG21, TC21, DT21
P(s)31 PG31, TC31, DT31
P(s)12 PG12, TC12, DT12
P(s)22 PG22, TC22, DT22
P(s)32 PG32, TC32, DT32
P(s)13 PG13, TC13, DT13
P(s)23 PG23, TC23, DT23
P(s)33 PG33, TC33, DT33
Table 12.5 - Model Constants from Loop Analysis Test Reports
The dynamic decouplers for the control system are determined as
follows:

Decoupler Tuning Constants

Feedforward Transfer Gain Lead Lag Delay


Decoupler Function (Unitless) Seconds Seconds Seconds
D(s)12 P(s)12/P(s)11 PG12/PG11 TC11 TC12 (DT12 - DT11)
D(s)13 P(s)13/P(s)33 PG13/PG33 TC33 TC13 (DT13 - DT33)
D(s)14 P(s)14/P(s)44 PG14/PG44 TC44 TC14 (DT14 - DT44)

D(s)21 P(s)21/P(s)22 PG21/PG22 TC22 TC21 (DT21 - DT22)


D(s)23 P(s)23/P(s)33 PG23/PG33 TC33 TC23 (DT23 - DT33)
D(s)24 P(s)24/P(s)44 PG24/PG44 TC44 TC24 (DT24 - DT44)

D(s)31 P(s)31/P(s)11 PG31/PG11 TC11 TC12 (DT12 - DT11)


D(s)32 P(s)32/P(s)22 PG32/PG22 TC22 TC32 (DT32 - DT22)
D(s)34 P(s)34/P(s)44 PG34/PG44 TC44 TC34 (DT34 - DT44)
Table 12.6 - Calculating Decoupler Tuning Parameters for 3X3 Process

Application Manual 122


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

NOTE
Again, the sign for the individual process gains is not displayed on the Loop
Analysis Reports. Therefore, to insure the correct sign for each process gain, look
at the plots test data to determine the correct sign for each process gain in the
table.

Application Manual 123


DECOUPLING INTERACTIVE MULTIVARIABLE CONTROL

Application Manual 124


CHAPTER

13
Statistical Analysis
Statistical analysis of process data can provide valuable insights into the
control of the process.

Statistical Analysis Testing


Statistical analysis testing is a powerful tool to determine performance of
your control system. The procedure for performing statistical analysis on
test data is to simply window the section of test data which you wish to
analyze. Go to Options and select Statistics.

Figure 13.1 - Typical Statistical Analysis Report

Application Manual 125


STATISTICAL ANALYSIS

Noise
Signal noise is short term random variance in the process variable
measurement signal. Statistical variance due to signal noise is an
excellent way to document the quality of the measurement.

Closed Loop Variance


Testing for closed loop variance is typically performed over long periods
of time. The PSA data recording features allow you to set both the scan
rate and sample time, allowing unique flexibility in determining the
quality of control. Statistical data, on critical variables before and after
system optimization, is a powerful way to document the value of your
system optimization efforts.

Statistical Analysis Of Controller Outputs


Feedback control loops are tuned to maintain the process variables at
specific setpoints. Statistical analysis of controller outputs of closed loop
systems, is a powerful tool in understanding the range over which the
final control element must move to control the process at setpoint. The
analysis of how much the valve is required to move under normal
operating conditions gives valuable insight into the load disturbances to
a system.

Application Manual 126


CHAPTER

14
Time Series Analysis
The function of a process control system is to produce uniform end
products in the face of varying raw materials and operating conditions.
Tuning controllers, deals with optimizing the controllers tuning
parameters based on the process dynamics, to provide stable robust
control. In many cases, the job of the control loop is to minimize the
impact of incoming variability on the downstream process and product.
That is, for each control loop in the system to provide minimum variance
control. If the load disturbances are cyclic, at frequencies near the
frequency of the closed loop control system, variance will actually be
increased. Achieving minimum variance control, therefore requires not
only tuning the controllers, but knowledge about the sources, magnitude,
and frequency of the disturbances to the process.

The time series analysis functions, in the Protunertm software, are


powerful tools that can be used as an integral part of a testing program
to improve product quality by identifying and reducing process
variation. Time series analysis refers to techniques dealing with the
analysis of a series of values, collected over time at successive, equally
spaced intervals. The analysis functions generated by the time series
analysis are summarized in Table 14.1. The Protunertm System Analysis
software incorporates both univariate (one variable) and bivariate (two
variable) analysis capabilities.

Application Manual 127


TIME SERIES ANALYSIS

Function Number Of Domain Information Provided


Process
Variables
Power- One Frequency Separates the variation into
Spectrum its frequency components
Characterizes variation as
either high or low frequency
Identifies cyclic variation
Evaluation of control loop
performance
Information regarding
potential improvement from
process control
Auto- One Time Characterizes variation as
correlation either long or short term
Identifies cyclic variation
Evaluation of control loop
performance
Cross- Two Time Indicates the similarity
correlation between two variables
Convo- Two Time The mirror image of
lution Crosscorrelation
Table 14.1 - Time Series Analysis Functions

Feedback Controls With Cyclic Disturbances


To help understand the value of the time series analysis functions in
achieving minimum variance control, it is important to understand, that
a well tuned feedback control loop, can actually increase variability. This
may sound like a contradiction, using the term well tuned with increased
variability, but the following example will assist you in understanding
the concept.

A control loop analysis test was conducted on a control loop to


determine the tuning parameters and the ultimate period of the process.

Application Manual 128


TIME SERIES ANALYSIS

A load disturbance of a constant amplitude at various frequencies was


then introduced. The open loop variance and the closed loop variance
was then compared at each frequency to determine the effects of the
closed loop controller on the variance. Equivalent tests were conducted
using the Protunertm calculated tuning parameters for slow, medium and
fast control. Figure 14.1 illustrates the results of these tests.

Figure 14.1 - Closed Loop Variance Attenuation as a Function of the


Disturbance Frequency
As the graph illustrates, when the period of the disturbance is less than .5
times the ultimate period of the process, the controller in closed loop
neither amplifies or decreases the variance. Therefore, disturbances with
periods less than .5 times the process period, can be considered noise.

NOTE
The controller output will cycle trying to correct the error caused by these fast
disturbances, but will have no effect on the magnitude of the measured variance.

With fast tuning parameters on the controller, disturbances with cyclic


periods between 0.5 and 3.5 times the ultimate period of the process are
amplified by the control loop. In this example, the variance in closed loop
was twice as large as the variance in open loop, when the period of the
disturbance variable was between 1.5 and 2 times that of the ultimate
period of the process.

Application Manual 129


TIME SERIES ANALYSIS

Using medium tuning parameters, disturbances with cyclic periods


between 0.5 and 4.5 times the ultimate period of the process increased
the variance in the controlled variable. The peak being an increase in
variance of 1.55 times the open loop variance at a disturbance period of
two times the ultimate period of the process. With slow tuning
parameters on the controller, disturbances with cyclic periods between
0.5 and 5.5 times the open loop variance were amplified by the control
loop. The maximum amplification being 1.3 times the open loop
variance.
The graphs illustrate, that the faster the tuning parameters, the greater
the amplification of cyclic variances with periods near the ultimate
period of the process, but afford greater attenuation of disturbances with
longer periods.
In conclusion, load disturbances, measured in an open loop process
variable measurement signal, with cyclic periods between 0.5 and 6 times
the ultimate period of the process, are likely to be amplified, not reduced,
by the control loop in automatic. It is therefore, very important, as part of
your control system analysis procedures, to identify such disturbances
and to engineer the appropriate solutions to achieve minimum variance
control. The following examples illustrate how the Protunertm Spectral
Analysis functions can be applied to do just that.

Time Analysis Test Procedure


The first step, in the analysis of cyclic load disturbances, is to perform the
Protunertm Loop Analysis test to determine the optimum tuning
parameters and the ultimate period of the process being controlled.
The second step, is to connect the Protunertm to measure the PV and the
PD signal of the loop being tested, along with other PV measurements
both upstream and downstream of the loop, that maybe a source of the
cyclic variation. Place the control loop under test in manual, with the
other loops in the control system in automatic, and record the normal
control system operation.

Time Series Analysis Of The Test Data

Application Manual 130


TIME SERIES ANALYSIS

Window the test data to be analyzed. The time series analysis functions
can be run on a maximum of 4096 data points. If the data set you want to
analyze exceeds 4096 data points, use the Convert function to create a
duplicate file at the appropriate scan rate for the analysis.
The Power Spectrum of the data is used to determine the frequencies
associated with the cyclic load disturbance on the measurement. This is
done using the cursor to determine the peak frequencies on the Power
Spectrum plot. If the period of the load disturbance is between .5 and 6
times the loop period, the variance will be greater in closed loop than in
manual. If you find this situation, it will be necessary to determine the
source of the disturbance and eliminate it at its source.
The Power Spectrum analysis can be run on each of the other measured
variables to determine if any of them are cycling at the same frequency as
the variable being controlled. The recorded variable, cycling at the same
frequencies as the controlled variable, is very likely the source of the
disturbance.

Figure 14.2 - Power Spectrum

Figure 14.2 illustrates the Power Spectrum performed on cyclic data. Use
the Window, Zoom and Cursor functions to identify and document the

Application Manual 131


TIME SERIES ANALYSIS

peak frequencies. Use the Comment function to document the peak


frequencies found.
If you determine that the cyclic load disturbance contains frequencies
that cannot be controlled by the controller, run the Power Spectrum
analysis on the other interactive variables to determine which variables
are the source of the disturbance.

Application Manual 132


APPENDIX

Commonly Used PID


Controllers
The following table is a list documenting the implementations employed
in some commonly used PID controllers. The list is not complete, and
should only be used as a guideline for Protunertm setup. Refer to the
Instruction Manual for your particular controller to verify the current
implementation.

Controller Algo- Impli- D Set- Anti-Reset PID Units


Description rithm mented Filter point Windup?
Type Action
ABB/Taylor S pos 0.06 PI Yes G R/M M
Mod 30
ESOO
PI only
ABB/Taylor S pos 0.06 I-PD Yes G R/M M
Mod 30
PSPO
ABB/Taylor S pos 0.06 PID Yes G R/M M
MOD 300
Interactive Form = Yes
Prop Action Mode = On
Error
Preact Mode = On Error
ABB/Taylor S pos 0.06 PI-D Yes G R/M M
MOD 300
Interactive Form = Yes
Prop Action Mode = On
Error
Preact Mode = On
Process
ABB/Taylor S pos 0.06 I-PD Yes G R/M M
MOD 300
Interactive Form = Yes
Prop Action Mode = On
Process
Preat Mode = On Process

Application Manual 133


COMMONLY USED PID CONTROLLERS

Controller Algo- Impli- D Set- Anti-Reset PID Units


Description rithm mented Filter point Windup?
ABB/Taylor P pos 0.06 PI-D Yes G R/M M
MOD 300
Interactive Form = No
Prop Action Mode = On
Error
Note: Do not use D on
this controller
ABB/Taylor P pos 0.06 I-PD Yes G R/M M
MOD 300
Interactive Form = No
Prop Action Mode = On
Process Note: Do not use
D on this controller
Allen Bradley I vel 0.00 PID Yes G M/R M
PLC5
ISA
D on Error
Allen Bradley I vel 0.00 PI-D Yes G M/R M
PLC5
ISA
D on Measurement
Allen Bradley P vel 0.00 PID Yes G R/S S
PLC5
Software on PID
D on Error
Allen Bradley P vel 0.00 PI-D Yes G R/S S
PLC5
Software on PID
D on Measurement
Bailey S A None PID No G R/M M
Model 701
Bailey S A None PID No G R/M M
Model 70112
Bailey 820 S A None PID No G R/M M
Bailey P pos 0.10 PID Yes G R/M M
Infi 90
Error Input
Function Code 18
Bailey P pos 0.10 I-PD Yes G R/M M
Infi 90 PV and SP
Function Code 19

Application Manual 134


COMMONLY USED PID CONTROLLERS

Controller Algo- Impli- D Set- Anti-Reset PID Units


Description rithm mented Filter point Windup?
Type Action
Bailey S pos 0.10 PI-D Yes G R/M M
Infi 90
Function Code 156
S18 = 0 Classical
S20 = 0 Normal PID
Bailey S pos 0.10 I-PD Yes G R/M M
Infi 90
Function Code 156
S 18 = 0 Classical
S20 = 1 I only on SP
Bailey P pos 0.10 PI-D Yes G R/M M
Infi 90
Function Code 156
S18 = 1 Noninteracting
S20 = 0 Normal PID
Bailey P pos 0.10 I-PD Yes G R/M M
Infi 90
S18 = 1 Noninteracting
S20 = 1 I only on SP

Bailey P A 0.10 PI-D Yes G R/M M


700 Series
Setpoint Modifier Out
Bailey P A 0.09 I-PD Yes G R/M M
700 Series
Setpoint Modifier In
Bristol Babcock I A None PI-D Yes G R/M M
Rack System
Fischer Porter S pos 0.10 PID Yes % M/R M
Micro DCI I-PD
DPV = 1 when
SP dig-
ital
input
Fischer Porter S pos 0.10 PI-D Yes % M/R M
Micro DCI I-PD
DPV = 0 when
SP dig-
ital
input
Fisher ac2 S A None PID Yes % R/M M
TL Series

Application Manual 135


COMMONLY USED PID CONTROLLERS

Controller Algo- Impli- D Set- Anti- PID Units


Description rithm mente Filter point Reset
Type d Windup?
Fisher S vel 0.1 I-PD Yes G S/R S
DPR 900
Fisher S vel 0.13 PI-D Yes G R/M M
Provox
Configurable
Fisher S vel 0.13 PI-D Yes G R/M M
Provox
Computing
Fisher S pos 0.13 PI-D Yes G R/M M
Provox
UOC/IFC
Foxboro S pos 0.10 PI-D No % M/R R
Type 761
Foxboro S pos 0.10 PI-D Yes % M/R M
Micro Spec
Foxboro S pos 0.10 PI-D Yes % M/R M
IA System
GE Series 90-70 P pos 0.10 PID Yes G R/S S
PIDIND Algorithm 2
Config word XX1
Derivative on Error
GE Series 90-70 P vel None PI-D Yes G R/S S
PIDIND Algorithm 2
Config word XX2
Derivative on PV
GE Series 90-70 S vel None PID Yes G R/S S
PIDISA Algorithm 1
Config word XX1
Derivative on Error
GE Series 90-70 S vel None PI-D Yes G R/S S
PIDISA Algorithm 1
Config word XX2
Derivative on PV
GE Fanuc P vel None PID No P=Gain
Series 5 PLC I=Rep/Scan
I&D Function of Scan
D=Scan
Rate
Genesis P vel 0.00 PI-D No % M/R M
Hartman & Braun I vel 0.25 PI-D Yes G M/R M
Protronic

Application Manual 136


COMMONLY USED PID CONTROLLERS

Controller Algo- Impli- D Set- Anti-Reset PID Units


Description rithm mented Filter point Windup?
Type Action
Honeywell S vel 0.13 PID Yes G M/R M
TDC-Basic
Type A
Honeywell S vel 0.13 I-PD Yes G M/R M
TDC-Basic
Type B
Honeywell S vel 0.13 PID Yes G M/R M
TDC-Extended
Type A
Honeywell S vel 0.13 I-PD Yes G M/R M
TDC-Extended
Type B
Honeywell S vel 0.13 PI-D Yes G M/R M
TDC-Multifunction
Type B
Honeywell S vel 0.13 PID Yes G M/R M
TDC-Process
Manager
Interactive Controller
Type A
Honeywell S vel 0.13 PI-D Yes G M/R M
TDC-Process
Manager
Interactive Controller
Type B
Honeywell S vel 0.13 I-PD Yes G M/R M
TDC-Process
Manager
Interactive Controller
Type C
Honeywell I vel 0.13 PID Yes G M/R M
TDC-Process
Manager
Ideal Controller
Type A
Honeywell I vel 0.13 PI-D Yes G M/R M
TDC-Process
Manager
Ideal Controller

Application Manual 137


COMMONLY USED PID CONTROLLERS

Type B

Controller Algo- Impli- D Set- Anti- PID Units


Description rithm mente Filter point Reset
Type d Windup?
Honeywell I A .125D I-PD No G M/R M
TDC-Process
Manager
Ideal Controller
Type C
L&N S A 0.125 PI-D No G R/M M
DPU 555
L&N S vel 0.10 PI-D Yes % R/M M
ElectroMax V
L&N S A None PI-D No % R/M M
Century
Model 440
L&N S A None PI-D No % R/M M
Century
Model 446-3
Measurex I vel 0.00 PID Yes G S/R S
Vision 2000 Must set Actuator
CCP & PIDP Gain to 100/Span
Modcomp I vel 0.10 PI-D No G S/R S
Modicon I vel 0.50 PID Yes % M/R M
984 to
PID 0.03
Moore Products S vel 0.30 PI-D Yes G M/R M
Type 351/352 to
1.0
Powers I pos 0.10 PID Yes % S/R S
Model 535 & 545
Process Controllers
Rosemount I pos 0.10 PID Yes P = % or G
System 3 I = S/R, = M/R
PD on error =H/R
D = S, = M, =H
Rosemount I pos 0.13 PI-D Yes Same as above
System 3
D on measurement
Rosemount I pos 0.13 I-PD Yes Same as above
System 3
PD on measurement

Application Manual 138


COMMONLY USED PID CONTROLLERS

Controller Algo- Impli- D Set- Anti-Reset PID Units


Description rithm mented Filter point Windup?
Type Action
Rosemount I pos 0.13 ID-P Yes Same as
System 3 above
P on measurement
SATT S vel 0.10 I-PD Yes G S/R S
Instruments
EAC40 & EAC 400
Square D I vel None PI-D No % M/R S
PLC
Texas Inst. I pos or PID Yes G R/S S
D/3 Control System vel
P on error
D on error
Texas Inst. I pos or PI-D Yes G R/S S
D/3 Control System vel
P on error
D on measurement
Texas Inst. I pos or I-PD Yes G R/S S
D/3 Control System vel
P on measurement
D on measurement
Texas Inst. P pos or 0.00 PI-D Yes G R/S S
TI-545 PLC vel
Toshiba S vel 0.10 PID Yes G S/R S
TOSDIC 200 to
0.30
Turnbull I pos 0.25 PI-D Yes % S/R S
TCS 6000 or
% M/R M
Westinghouse P vel 0.50 PID Yes G S/R S
WDPF
MAC 4500
Yokogawa I vel 0.10 I-PD Yes % S/R S
SLPC
CNT5 = 0
Yokogawa I vel 0.10 PID Yes % S/R S
SLPC
CNT5 = 1

Application Manual 139


COMMONLY USED PID CONTROLLERS

KEY TO TABLE
ALGORITHM TYPE I = Ideal
S = Series Algorithm
P = Parallel Algorithm

IMPLEMENTED Vel = Velocity or Incremental Form of Digital PID


Pos = Positional Form of Digital PID
A = Analog PID Controller
DERIVATIVE FILTER *Td = Derivative Filter Time Constant
SETPOINT ACTION PID = P,I, and D act on Error
PI-D = PI act on Error and D acts on -PV
I-PD = I acts on Error and P and D act on -PV
ANTI-RESET WINDUP Yes, some type of anti-reset windup is used
No, no anti-reset windup is used
PID UNITS - FOR P: G = Proportional Gain
% = Proportional Band % (100/Pgain)
FOR I: S/R = Seconds per Repeat
R/S = Repeats per Second
M/R = Minutes per Repeat
R/M = Repeats per Minute
FOR D:
S = Seconds
M = Minutes

Application Manual 140