General
Specifications
<<Contents>> <<Index>>
Models SFE3330, SFH3330
SIH6330, SHW6330
PREDICTROL Packages
Yokogawa Electric Corporation
2-9-32, Nakacho, Musashino-shi, Tokyo, 180-8750 Japan
Tel.: 81-422-52-5634 Fax.: 81-422-52-9802
GS 33G03M23-01E
GS 33G03M23-01E
©Copyright July. 1995(YK)
5th Edition May 1, 2002(YK)
GENERAL
These packages are the model predictive control
software which operate on the CENTUM CS control
station.
The model predictive control package with an
operation model in the controller runs while predicting
the future movement of processes, just as an
experienced operator operates a plant with the
characteristics of a process in mind. This has enabled
automated operation of processes which were used to
be difficult in conventional PID control. This leads to
stable control.
Model predictive control was primarily developed as
a control method for the computer level, and mainly
applied to chemical processes. YOKOGAWA realized
the potential of this control method from early on, and
was the first company in the world to install this in a
DCS. This enabled predictive control without the high
costs of investment. Further, support functions are
available on the ICS or EWS, which allows predictive
control without technical knowledge of computers.
Please note that software with different release
numbers (software version) cannot be used together
on CENTUM CS. When using software, check the
Release No. described as R .
FUNCTION SPECIFICATIONS
Features
• It allows control of a process with lengthy dead time
without any overshoot.
• It offers high stability against fluctuations that are
characteristic of a process and model errors.
• It may be used in the same way as a standard
function block, by installing it as a regulatory control
block.
• It is equipped with 2 auxiliary input terminals for
easy configuration of a control system (feedforward
control) which prevents influence from disturbances
such as load variation.
• It operates on the control station, thus providing high
reliability.
• It is equipped with support functions for generation of
models for control and simulation.
Applicable Processes
• Composition Control of Distillation Columns
Processes which are influenced by disturbances in
feed volumes and with long dead time.
• Temperature Control of Reactors
Processes with fluctuating characteristics, and where
overshoot is undesirable when altering the setpoint.
• Combustion Control and Level Control of Boilers
Processes with reverse response and undefined
characteristics.
• Zone Temperature Control of Furnaces
Processes with interference and large disturbances.
1. Control Functions
Operating Principles
The predictive control block calculates the control
output MV for each control period, as follows.
• It calculates the trajectory of the controlled variable
PV from the present value to a value approaching
the setpoint (reference trajectory). The slope of this
trajectory is specified by closed-loop response time.
• By using the built-in model and PV, and the feedforward
input value, the movement of the controlled
variable PV is calculated within the predictive
horizon.
• The controlled output is calculated so that the
reference trajectory and predictive value of PV
coincide within the coincident zone.
As can be seen from the above algorithm, while PID
control tries to make the control deviation 0 at the
next control output, predictive control tries to make the
control deviation 0 a few steps ahead. This ensures
that even if there are variations or errors in the model,
higher robustness can be maintained.
F01E.ai
Past
1 : Setpoint Value 2 : Reference Trajectory
3 : Predictive Value 4 : Manipulated Value
5 : Coincident Zone
Present Future
Measured Value
Predictive
Horizon
1
2
3
4
5
Input Processing
Same as PID control blocks.
Control Computation Processing
Excluding predictive control computations, it is basically
the same as PID control blocks. With predictive
control blocks, zone control functions rather than
gap operation, is available. In zone control when the
control deviation enters the high-limit or low-limit zone,
the slope of the reference trajectory becomes more
gradual, and prevents unnecessary movement of the
manipulated value. This is particularly effective when
there are variations or errors in sensors, or static.
Specification for reverse action is made by using the
signs for the model.
Calculation and display of the predictive value is carried
out, allowing prior checking of controller movements,
and detection of the impact of the MV output on other
controlled variables.
2
All Rights Reserved. Copyright © 1995, Yokogawa Electric Corporation
<<Contents>> <<Index>>
GS 33G03M23-01E May 1,2002-00
Ouput Processing
Same as PID control blocks.
Function Block Specifications
Items Specifications
Name PRDCT
Control Period 2 to 10,000 seconds
I/O Points 3 inputs per 1 output
Tuning Parameters
Signs Data Names Range
CRT Closed-loop response time 0.1 to 1,000.0
minutes
ZNH Upper zone width setpoint Engineering unit
ZNL Lower zone width setpoint Engineering unit
ZNS Zone control selection switch 0/1 (on at 1)
PRPV Predictive controlled variable Engineering unit
CT Control period 2 to 10,000 seconds
HT Predictive horizon 0 to 60
2. PREDICTROL Support Functions
Designs for modeling of processes or control systems
were restricted to selected control technicians. The
PREDICTROL support functions is a package with
direct connection to instrumentation systems developed
specifically for model predictive control. Thus, it offers
a broad range of functions, from model definition of
processes to the designing of the control system, to
simulation.
Definition of the Simulation Model
This defines the model to be used instead of the actual
process when conducting process identification or
control tests.
Process Identification
This is when a minimal test signal is applied to to the
process, and based on the behavior of the process at
that time, the dynamic characteristic of the process
is obtained. This operation, in the past viewed as
being complicated, can be easily tuned by using the
PREDICTROL support functions.
Designing of the Predictive Control System
The controller for predictive control is designed based
on the dynamic characteristic of the obtained process.
With the PREDICTROL support functions, specifying
the response type of the control response is enough to
design the predictive control system. With the designed
controller, simulation and confirmation of control
capabilities can be carried out.
Definition of Simulation Model
Proceess Identification
• Definition of model identification
• Closed-loop/open-loop identification
• Validation of model identification
Designing of Predictive Control System
• Definition of control model
• Designing of contoroller
• Control test
• Model setting to predictive control block
F02E.ai
Pjt:newpro Stn:FCS1(AFH20S) Area:FCS0102 File:SML003
Simulation Model Degfinition
Comment
Number of Input
Input Character
Integral
95% Resp.Time(min)
Dead Time(min)
Gain
Input operating Point(%)
File Help
Resp.Model Display Resp.Model Edit
: simulation1 Sampling Period(sec):
: ◊1 ◊2
First Input Second Input Third Input
◊3
◊On ◊Off ◊On ◊Off ◊On ◊Off
50,000
50.00 50.00 50.00
0.00 0.00 0.00
1.000 1.000 1.000
50.000 50.000 50.000
60
Output Operating Point(%):
Simulation Model Definition Panel
F03E.ai
1
1
1/1 ▼ ▼ PAUSE CLOSE
100.00
80.00
60.00
40.00
20.00
0.00
10.00
0.00
-10.00
Process Output
Process Input 1
50.00
Process Input 2
50.00
Process Output
Filtered Value
-0.64
Identified Model Output
Filtered Value
-0.75
17:53 18:53 19:53 20:53 21:23 22:53
Open-loop Identification Trend Dialogue
F04E.ai
prctl
Pjt:newpro Stn:FCS1(AFH20S) Area:FCS0102 File:Changed
Controller Design
Resp.Model Display
Tuning
Closed Loop Char.
Robustness
Prd. Horizon 60
File Ch. view Help
Respose Type Choice
◊Smooth Input
Gain Margin
Prd. Horizon
21.500
60
◊Overshoot Minimum
Gain Margin
Prd. Horizon
18.000
9
◊Fast Response
Gain Margin
Prd. Horizon
17.000
3
◊Robostness Marinum
Gain Margin
Prd. Horizon
19.500
1
120
100
80
60
40
20
0
120
100
80
60
40
20
0
140
120
100
80
60
40
20
0
140
120
100
80
60
40
20
0
Controller Designing Panel
3
All Rights Reserved. Copyright © 1995, Yokogawa Electric Corporation
<<Contents>> <<Index>>
GS 33G03M23-01E
OPERATING ENVIRONMENT
Model SFE3330,SFH3330
PREDICTROL Packages
• Hardware Requirements
FCS: AFM10S,AFM10D,AFM20S,AFM20D
AFS10S,AFS10D,AFS20S,AFS20D
• Software Requirements
They operate on the Field Control Station Standard
Control Functions (Models SFE1100,SFH1100).
Models SIH6330,SHW6330
PREDICTROL Support Package
• Hardware Requirements
ICS:
Main Memory: 64 MB or more
(For ICS with stacked CRT, 96 MB or
more)
Auxiliary Memory: 2 GB or more
EWS (HP9000 Series):
Main Memory: 80 MB or more
Auxiliary Memory: 4 GB or more
• Software Requirements
ICS:
The following software packages are necessary.
SIH5100 Standard Builder Functions
SIH5101 System Definition Functions
SIH5102 System Utility
SIH5103 Maintenance Utility
SIH5120 FCS Builder
SIH6330 PREDICTROL Support Package
EWS (HP9000 Series): For the operating
environment, see “System Generation
Builder Functions (GS 33G04K20-01E).”
The following software packages are necessary.
SHW5100 Standard Builder Functions
SHW5101 System Definition Functions
SHW5102 System Utility
SHW5103 Maintenance Utility
SHW5120 FCS Builder
SHW6330 PREDICTROL Support Package
MODELS AND SUFFIX
[Release: R2]
Description
Models
SFE3330 PREDICTROL Package
(for AFM)
SFH3330 PREDICTROL Package
(for AFS)
Suffix
Codes
-S Basic Software License
-C Multiple Software License
(for 2 or more)
1 Always 1
1 English-Language Version
[Release: R2]
Description
Models
SIH6330 PREDICTROL Support Package
(for ICS)
SHW6330 PREDICTROL Support Package
(for HP9000)
Suffix
Codes
-S Basic Software License
-C Multiple Software License
(for 2 or more)
1 Always 1
1 English-Language Version
ORDERING INFORMATION
• Specify model and suffix code.
• When ordering software separately from hardware,
specify ICS or HP9000 ID no. for if to be bundled
with.
TRADEMARKS
• CENTUM is a registered trademark of Yokogawa
Electric Corporation.
• All other company names and product names
mentioned in this GS are registered trademarks of
the respective companies.
May 1,2002-01
Subject to change without notice.
