Professional Documents
Culture Documents
Doc ID LOOP-UM001B-EN-P
ii RSLoop Optimizer User’s Guide
.....
Contacting Technical Support Telephone — 440-646-7800
Rockwell Software Technical Support Fax — 440-646-7801
World Wide Web—www.software.rockwell.com
Copyright Notice ©2000-2001 Rockwell Software Inc. All rights reserved.
Printed in the United States of America
Portions copyrighted by Allen-Bradley Company, Inc., and used with permission.
This manual and any accompanying Rockwell Software products are copyrighted by Rockwell
Software Inc. Any reproduction and/or distribution without prior written consent from Rockwell
Software Inc. is strictly prohibited. Please refer to the license agreement for details.
Trademark Notices WINtelligent Series is a registered trademark, and the Rockwell Software and RSBizWare logos,
RSAlarm, RSAnimator, RSAssistant, RSBizWare Batch, RSBizWare BatchERP,
RSBizWare eProcedure, RSBizWare MaterialTrack, RSBatch, RSBatch eProcedure,
RSBatch Production Manager, RSBizWare BatchCampaign, RSBizWare Historian,
RSBizWare Scheduler, RSBizWare ComplianceTrack, RSBizWare PlantMetrics, RSBreakerBox,
RSButton, RSChart, RSCompare, RSControlRoom, RSData, RSDataPlayer, RSEventMaster,
RSGauge, RSJunctionBox, RSLogix Emulate 5, RSLogix Emulate 500, RSGuardian, RSHarmony,
RSKeys, RSLadder, RSLadder 5, RSLadder 500, RSLinx, RSLinx Gateway, RSLogix, RSLogix 5,
RSLogix 500, RSLogix 5000, RSLogix Frameworks, RSLogix SL5, RSMailman, RSNetWorx,
RSPocketLogix, MaintenanCE Suite, RSMATE, Security Server, RSPortal, RSPower, RSPower32,
RSEnergy, RSPowerCFG, RSPowerRUN, RSPowerTools, RSRules, RSServer32, RSServer,
RSServer OPC Toolkit, RSSidewinderX, RSSlider, RSSnapshot, RSSql, RSToolbox, RSToolPak I,
RSToolPak II, RSTools, RSTrainer, RSTrend, RSTune, RSLoop Optimizer, RSVessel, RSView32,
RSView, RSView32 Messenger, RSView32 TrendX, RSView32 RecipePro, RSView32 SPC,
RSView32 GEMTool, RSView32 SECSHost, RSView32 Active Display System, RSView32 Active
Display Station, RSView32 Active Display Server, RSView32 WebServer, RSView32 DataLyzor,
RSView Enterprise Series, RSView Machine Edition, RSView Supervisory Edition, RSVisualLogix,
RSWheel, RSWire, RSWire Designer, RSWire Detailer, RSWire I/O Builder, RSWire Docx,
RSWorkbench, RSWorkshop, SoftLogix 5, A.I. Series, Advanced Interface (A.I.) Series,
AdvanceDDE, AutomationPak, ControlGuardian, ControlPak, ControlView, INTERCHANGE,
Library Manager, Logic Wizard, Packed DDE, ProcessPak, ProcessPak5, ProcessPak5 for Batch,
View Wizard, WINtelligent, WINtelligent LINX, WINtelligent LOGIC 5, WINtelligent VIEW,
WINtelligent RECIPE, WINtelligent VISION, and WINtelligent VISION2 are trademarks of
Rockwell Software Inc.
ProcessLogix, PLC, PLC-2, PLC-3 and PLC-5 are registered trademarks, and Data Highway Plus,
DH+, DHII, DTL, MicroLogix, Network DTL, PowerText, Pyramid Integrator, PanelBuilder,
PanelView, PLC-5/250, PLC-5/20E, PLC-5/40E, PLC-5/80E, SLC, SLC 5/01, SLC 5/02, SLC 5/ 03,
SLC 5/04, SLC 5/05, SLC 500, and SoftLogix are trademarks of the Allen-Bradley Company, Inc.
Microsoft, MS-DOS, Windows, Windows NT, ActiveX, Visual Basic, Visual C++, Visual J++, and
Visio are registered trademarks, and Microsoft Access and Visual SourceSafe are trademark of the
Microsoft Corporation.
ControlNet is a trademark of ControlNet International.
DeviceNet is a trademark of the Open DeviceNet Vendors Association.
Ethernet is a registered trademark of Digital Equipment Corporation, Intel, and Xerox Corporation.
Pentium is a registered trademark of the Intel Corporation.
Adobe and Acrobat are trademarks of Adobe Systems Incorporated.
IBM is a registered trademark of International Business Machines Corporation. AIX, PowerPC,
Power Series, RISC System/6000 are trademarks of International Business Machines Corporation.
UNIX is a registered trademark in the United States and other countries, licensed exclusively through
X/Open Company Limited.
AutoCAD is a registered trademark of Autodesk, Inc.
All other trademarks are the property of their respective holders and are hereby acknowledged.
Installation 5
Setting up RSLinx for RSLoop Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Setting up ProcessLogix for RSLoop Optimizer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Installing RSLoop Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Starting the RSLoop Optimizer Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Configuring a Loop to Communicate with a Processor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Browsing OPC Item Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Editing and Deleting Loops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Editing an Existing Loop. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Deleting a Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Testing Communications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Troubleshooting Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Quick Start 17
Tuning a Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Guidelines for Optimizing Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Tuning Theory 21
Description of Proportional, Integral, and Derivative Control . . . . . . . . . . . . . . . . . . . . . . . . . . 21
PID Loop Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Proportional Only Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Proportional Plus Integral (PI) Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Proportional Plus Integral Plus Derivative (PID) Control. . . . . . . . . . . . . . . . . . . . . . . . . . 24
RSLoop Optimizer Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Tuning Types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Activation 179
How Activation Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Protecting Your Activation Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Activating RSLoop Optimizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Running the Activation Utilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Finding More Information About Activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Some Common Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Glossary 183
Index xi
.....
Welcome to RSLoop Optimizer™, the application that makes tuning your control loops fast, easy,
and accurate. RSLoop Optimizer also provides methods of analyzing your loops to help ensure
optimal tuning parameters.
This chapter covers:
What is RSLoop Optimizer?
RSLoop Optimizer Features
System and Software Requirements
WHAT
. . . . . . . I.S. . R. .S. L. .O. O. .P. .O. .P.T.I.M. I.Z. E. R. .?. . . . . . . . . . . . . . . . . . . . . . . . . .
RSLoop Optimizer is Rockwell Software’s Windows®-based software for analyzing and tuning
PID control loops in Allen-Bradley® PLC-5®, SLC 500™, MicroLogix 1200/1500, ProcessLogix
and ControlLogix Programmable Logic Controllers.
RSLoop Optimizer provides a complete set of tools with which you can analyze and troubleshoot
your process.
Easy-to-use AutoTune
Seamless connectivity
AutoTune: sequence
to your control reduces
loops: RSLoop
the time
Optimizer
required uses
to tune
RSLinx™
a loop from
hours
Standard,
to minutes.
Professional or OEM (OPC only) for all supported processors. The RSTune family
of products does
Archiving: not archiving
Manual work withofRSLinx Lite.
multiple sets of data allows easy before and after analysis.
AutoTune:
Performance Easy-to-use
increase AutoTune
displays:sequence reduces the
The performance time required
increase to tune
from tuning youra loop
loop is
from
hours to minutes.
displayed on the Faceplate.
Archiving: Manual
PID loop tuning archiving of
categories: multiple sets
Categories can of
be data allows
selected foreasy
load before
tuning and after analysis.
or setpoint tuning
from the simulation
Performance plot. displays: The performance increase from tuning your loop is
increase
displayed on the Faceplate.
Pre-download setting analysis: Allows you to see the performance of your loops before
actually
PID loopdownloading them to the
tuning categories: controller.
Categories can be selected for load tuning or setpoint tuning
from the
Data simulation plot.
optimization: Data can be zoomed, filtered, averaged, and line edited.
Pre-download
Control settingRSLoop
loop testing: analysis: Allows includes
Optimizer you to see the performance
powerful analysis plotsofthat
your loops
provide
before performance
critical actually downloading
informationthem the controller.
to loops.
on your
Hysteresis check: Allows
Data optimization: you
Data can beto determine
zoomed, whether
filtered, your control
averaged, elements
and line edited. (e.g. valves) are
suffering from hysteresis.
Control loop testing: RSLoop Optimizer includes powerful analysis plots that provide
reports: Include
Tuningperformance
critical data, notes,
information andloops.
on your graphics.
Browsing
Hysteresiscapability (version
check: Allows you2.3
to or later): Awhether
determine new feature,
your available
controlonly with OPC
elements (e.g.
communications, whichfrom
valves) are suffering allows you to browse
hysteresis . the network for OPC items such as PID
instruction addresses and control block tag names.
Tuning reports: Include data, notes, and graphics.
Remotely connect to RSLinx (version 2.3 or later): If you are using OPC
communications, you now
Browsing capability have the
(version 2.3option of connecting
or later): to a remote
A new feature, computer
available thatOPC
only with is running
the RSLinx Gateway.
communications, which allows you to browse the network for OPC items such as PID
instruction addresses and control block tag names.
DATA ANALYSIS TOOLS OF RSLOOP OPTIMIZER
Remotely connect to RSLinx (version 2.3 or later): If you are using OPC
Valve Stiction Wizard: Helps with testing for stiction on valves.
communications, you now have the option of connecting to a remote computer that is running
the
PIDRSLinx
TuningGateway.
Table: Provides a concise way of quickly examining several loop tests. From
these, it automatically selects the most conservative tuning values, then it summarizes the
average value of all the tests.
S D S O FAllows R EtoR
.........................................................
YPerformance
S T E M A NSummary: T W A you
and valve wear — at a glance.
E Qoverall
view U I R trade-offs
E M E N TofSperformance, robustness,
Unlimited extra trends and loops can be added: Allows you to monitor additional
variables on the same faceplate trend to see how other process variables affect the loop. Extra
trends and loops help you model these systems:
Multi-variable (2x2, 3x3, 2x3, and so on)
Cascade
Interacting loop
Load processes for feedforward compensation
Valve wear analysis: Allows you to improve and maximize valve life by comparing tuning
and additional filtering.
Updated robustness plots: These plots now use actual process gain and dead time for the
axes. You can see exactly how much dead time or gain increase will produce an unstable loop.
A robustness index has also been added to show the robustness of the loop in one value. The
index can be used to quickly compare how different tunings and filters will affect the loop’s
sensitivity to process changes. Also included: You can dial in the robustness you want by
simply dragging the robustness line.
Histogram plot: Allows you to analyze the grouping of you PV, CO, and errors in histogram
format. Also included are actual valve travels and reversals, IAE (Integrated Absolute Error),
variability index, IAE/time, valve reversals/time, and valve travel/time. In addition, you can
now select time settings of hours, days, weeks, months, or years.
Noise simulation: The response to measurement noise plot allows “what-if” testing to see
how changing tuning or filtering affects controller output wear or jitters to the slave in a
cascade.
Inverse Response Process: Allows RSLoop Optimizer to analyze your process as if it has
negative lead, integrator, and lag.
Characterizer: Helps you linearize your process for optimal performance across the entire
range.
Power spectral density plots: Shows the potential improvement from better tuning and
helps you identify cyclic upsets.
Auto- and cross-correlation graphs: Allows you to determine the relative amount of
interdependency either between two loops (cross-correlation), or the relative amount of
variability that the loop causes with its own control (auto-correlation).
Setpoint filter: Achieves great response to both setpoint changes and load upsets.
Frequency plot adjustments: Gives you the ability to change the maximum number of
frequency points for analysis.
Full support for RSLogix Frameworks PID and Lin function blocks: Allows you to tune
your PID loops and download the parameters to the PID memory table automatically. Use
RSLoop Optimizer’s characterizer to linearize your process and download the X and Y values
directly to the Lin function block.
Extensive reporting capabilities: Gives you the option of adding analysis plots and data
directly into your report and provides the means for you to customize the report template to
suit your needs. Extended reporting includes detailed analysis of hysteresis and linearization.
Also included is an explanation of each analysis, which explains its importance to loop health,
making it more legible to the customer or supervisor.
SYSTEM
. . . . . . . . . A. .N. D. . S. .O. F. .T.W. .A.R. E. . R. .E. Q. .U. I.R. E. M. .E. N. .T.S. . . . . . . . . . . . . . . .
IBM®-compatible 486 or greater (Pentium™ recommended)
Microsoft® Windows® 95, Windows 98, Windows NT™ (4.0, Service Pack 6a or greater),
Windows 2000 (Service Pack 1)
If reporting function will be used, Microsoft® Office 2000 (MS Word)
NOTE: The Lite version of the communications software is not sufficient for communication
with RSLoop Optimizer. You must have at least the OEM version of the communications
software to communicate via OPC or the Professional version to communicate via DDE.
NOTE: If you want to use an OPC link, you must have RSLinx version 2.3 (or higher) or
ProcessLogix 300 (or higher). Earlier versions of RSLinx only support DDE.
.....
This chapter provides information on installing RSLoop Optimizer and setting up the
communications package.
You must have communication software installed and configured for RSLoop Optimizer to
communicate with your control loop. RSLoop Optimizer works with:
RSLinx 2.3 (or higher)
NOTE: The Lite version of the communications software is not sufficient for communication
with RSLoop Optimizer. You must have at least the OEM version of the communications
software to communicate via OPC or the Professional version to communicate via DDE.
You can simulate a control loop using built-in DDE communications without the use of RSLinx.
(See “Using RSLoop Optimizer” on page 27 for more information.)
This chapter covers:
Setting up RSLinx for RSLoop Optimizer
Setting up ProcessLogix for RSLoop Optimizer
Installing RSLoop Optimizer
Starting the RSLoop Optimizer Software
Configuring a Loop to Communicate with a Processor
Editing and Deleting Loops
Testing Communications
Troubleshooting Installation
SETTING
. . . . . . . . . .U. P. . .R. S. .L. I.N. X. . F. .O.R. . R. .S. L. .O. O. .P. .O. .P.T.I.M. I.Z. E. .R. . . . . . . . . .
To have RSLoop Optimizer communicate to your processor, you must have RSLinx configured
and running. For each processor that RSLoop Optimizer will communicate with, you need to have
an RSLinx DDE/OPC Topic defined.
NOTE: This section provides an overview of the steps required in the communication
software. For more information on configuring the software and defining a DDE/OPC topic, see
“DDE Topics” on page 141 or “OPC Topics” on page 145 and the RSLinx documentation.
NOTE: These steps are not needed if you are using the control loop simulator.
1. Install RSLinx.
2. Configure RSLinx to communicate with your processor.
3. Define an RSLinx topic that RSLoop Optimizer can use to communicate with your PLC.
SETTING
. . . . . . . . . .U. P. . .P.R. O. .C. E. .S.S. L. .O.G. .I X. . F. .O.R. . R. .S. L. .O. O. .P. .O. .P.T.I.M. I.Z. E. .R. .
Release 2.0 of RSLoop Optimizer has the capability to connect to ProcessLogix. To have
RSLoop Optimizer communicate to your processor, you must have ProcessLogix configured and
running. You must also have RSLoop Optimizer installed on the OPC server.
NOTE: These steps are not needed if you are using the control loop simulator.
4. Click the Security tab, select “Use custom access permissions”, and then click the Edit
button.
5. You must add an access permission that allows access for the Remote OPC server to
communicate with RSLoop Optimizer. Selecting Everyone includes all users and logins. If
security is a concern, contact your local system administrator for restrictive access.
3. In the Open edit box, type drive:\setup, where drive is the letter of the drive containing the
CD-ROM. Click OK.
4. Follow the directions on the screen.
5. When prompted for the product’s serial number, enter the last six digits of the serial number
(found on the label of the Master Disk).
6. When asked if you want to move activation now, click Yes. Insert the Master Disk into the
disk drive.
7. The utility for moving activation, EvMove, runs. Use the EvMove dialog boxes to move
activation from the Master Disk to your root directory (usually C:). For help using EvMove,
see “How Activation Works” on page 179 or EvMove online help.
8. Remove the Master Disk and follow the directions on the screen. When the setup utility
finishes, an entry for the RSLoop Optimizer application program is displayed in the program
list in the Rockwell Software group.
9. Store the CD-ROM and the Master Disk in a safe place. (For more information on product
activation, see “Activation” on page 179.)
STARTING
. . . . . . . . . . . .T.H. E. . R. .S. L. .O. O. .P. .O. .P.T. I.M. I.Z. E. .R. .S. O. .F.T. W. .A. R. .E. . . . . . . . .
To start RSLoop Optimizer software on a PC:
1. Click Start.
2. Select Programs > Rockwell Software > RSLoop Optimizer.
From this window you can define a new loop, choose an existing loop to either tune or edit, or
delete a loop. Loops that have already been created are listed in the Choose a Loop box in the
main window. For more information, see “Editing and Deleting Loops” on page 14.
NOTE: For important information about programming the PID instruction in your processor,
please see “Recommendations for Programming PID Loops” on page 175.
We recommend using the PD file type when you program the PID loop if you are using a New
Platform PLC-5 processor.
4. Type a name for the loop in the File name box. The file extension .tun is added automatically.
5. Click Save. The RSLoop Optimizer Setup dialog box opens, as shown here.
6. Choose the Processor Type. The information on the dialog box changes to show the available
options for the selected processor.
If you are configuring one of the processors listed in the RSLoop Optimizer Setup window
through ProcessLogix, we recommend that you select the processor you will be connecting to
(such as PLC-5 or SLC 5/02) instead of ProcessLogix.
7. Under Communications Via, select the type of communications you want to use — OPC or
DDE.
8. If you selected:
a) DDE: Continue with step 10.
b) OPC: The Remote computer field is enabled. Double-click the empty field or click the
browse button to the right of the box. The Browse Network Computers dialog box opens.
NOTE: If you are “talking” to the local computer, please leave this field blank.
9. Navigate through the network to locate the remote computer to which you want to establish
communications. Either double-click the computer name or select the computer name and
click OK.
10. Complete the remaining areas on the dialog box as described in the following table:
* If using OPC communications, you can select from a list of available PID instruction addresses and tag names.
(See “Browsing OPC Item Names” on page 13 for more information.)
Scroll bar
Scroll arrows
1. To navigate in the dialog box, scroll down the hierarchy of OPC Item names until you reach
the item name that you wish to use.
2. Click the item name to select it and click OK. The selected item name is transferred to the
Edit Setup dialog box. The topic name appears in brackets next to the item name.
Or, you can double-click the item to simultaneously select it and apply it to the Edit Setup
dialog box.
NOTE: Browsing is only available with OPC communications, not with DDE.
EDITING
. . . . . . . . . .A.N. D. . .D. E. .L.E.T. I.N. G. . .L.O. O. .P. S. . . . . . . . . . . . . . . . . . . . . . . . . .
EDITING AN EXISTING LOOP
The setup parameters for an existing loop can be edited at any time.
1. Click the name of the loop in the Choose a Loop list.
2. Click Edit Setup. The RSLoop Optimizer Setup dialog box opens. See “Configuring a Loop
to Communicate with a Processor” on page 9 for detail on the options in this dialog box.
NOTE: When you are editing an existing control loop, options in the RSLoop Optimizer Setup
dialog box are grayed out if there is archived data stored for that loop.
To select a new PID loop or change the location, do one of the following:
1. Click Save As to create a new loop.
2. Click Close to go to the main window. Click New to create a new loop.
NOTE: If you want to configure a new loop that is similar to another (same PLC), first select
the loop in the Choose a Loop list, then click New Loop. The new loop will be a copy of the
original. Now you can change addresses to complete the setup of the new loop.
DELETING A LOOP
To delete a loop (a .tun file and all associated data files):
1. In the main window, click the loop you want to delete in the Choose a Loop list.
2. Select File > Delete Loop.
3. To:
Delete archived data for this loop: Select Yes in the dialog box.
Keep archived data for this loop: Select No in the dialog box. Data can be deleted
manually later if desired.
TESTING COMMUNICATIONS
.........................................................
The link between RSLoop Optimizer and the server can be either DDE or OPC.
RSLoop Optimizer automatically determines the available communications methods. If there is a
choice, you can choose between DDE or OPC in the Edit Setup dialog box.
NOTE: When you have a choice, OPC is the preferred method of communications.
You can test the communications through the Setup dialog box to see if the server is responding
correctly.
To test communications, click Test.
RSLoop Optimizer attempts to read the process variable for the loop from the server.
RSLoop Optimizer displays either an error message or the current process variable of the loop.
TROUBLESHOOTING INSTALLATION
.........................................................
If RSLoop Optimizer does not start up or run properly, check that the communication software is
installed. RSLoop Optimizer works with RSLinx version 2.3 or later and ProcessLogix version
3.0.0 or later. The communication software provides communication between the programmable
controller and a personal computer in the Microsoft Windows NT, Windows 95, Windows 98 or
Windows 2000 environment.
If you get a LINX Initialization error message in RSLoop Optimizer when you try to call up the
Control Loop Setup window, check these items.
Check to see if RSLinx is running. If not, did you install it? Can you run RSLinx by itself?
Check the configuration of the RSLinx topic for proper station address and communications
device.
Make sure your cable or card is plugged in.
Check the RSLinx topic to be sure that the Station number is set to the node address of the
processor with which you want to communicate.
If you get the error “Foreign application won’t perform DDE method or operation,” you might
have tried to access the faceplate before all of the RSLinx drivers have fully initialized. Wait
and then try again.
Check to see if your Control Block address is pointing to a PID controller.
.....
This chapter gives you a step-by-step approach to get you started using RSLoop Optimizer. More
detailed explanations about the tuning process and how you can edit, verify, and analyze your data
can be found in the remaining chapters of this User’s Guide.
This chapter covers:
Tuning a Loop
Guidelines for Optimizing Loops
TUNING
. . . . . . . . .A. . L. .O. O. .P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Establish and test communications as shown in “Testing Communications” on page 15.
2. Click Faceplate to communicate with your PID loop or software simulation. The Faceplate
window opens.
3. Click AutoTune.
NOTE: You can cancel the AutoTune sequence and return to normal operation at any time by
clicking the End Sequence button.
For each question in AutoTune, help is available by clicking Help. Detailed information on
AutoTune is provided in “Using AutoTune to Collect Data” on page 53.
5. When you have completed the AutoTune sequence, RSLoop Optimizer displays suggested
PID tuning parameters, the Time Data Window for the loop, and the safety factor, derivative,
and filter information. Click Download to send these parameters to the processor or
simulation.
6. A sample of the screen after AutoTune has completed is shown here.
GUIDELINES
. . . . . . . . . . . . . .F.O. R. . O. . P. .T.I M. .I.Z.I .N.G. . L. .O. O. .P.S. . . . . . . . . . . . . . . . . . . .
These guidelines help you optimize loops and identify process equipment problems. They are
especially helpful for a control loop that is difficult to tune.
While controller tuning can attenuate disturbances caused by process equipment problems, if not
corrected, these problems can result in more severe equipment, process, or safety problems over
time.
To completely check the control system:
1. Collect process variable data for some time with the controller in Manual.
2. Watch for any periodic load disturbances.
3. If load disturbances occur, try to identify the sources. Minimizing or eliminating load
disturbances will allow the controller to do a much better job.
4. Collect process variable and controller output data with the controller in Auto under normal
operating conditions.
5. Is the controller output operating at one end of the span?
6. Is a valve operating near its seat? If so, the valve or final control element might need to be
resized to give better controller output resolution.
7. Perform the hysteresis check. (See “Hysteresis Check” on page 78 for more information.)
8. Perform an open loop step test (Manual mode) at several different locations in the controller
output span, for example, at 20%, 40%, 60%, and 80%. (See “Controller in Manual (Open
Loop)” on page 163 for more information.)
9. Check the new tuning parameters and the model identified in the Process Model window. (See
“Options in the Process Model Window” on page 125.)
10. Are the models (or PID tuning parameters) at each step significantly different? If the
parameters are more than a factor of two different, consider trying to linearize the loop. If you
cannot linearize the loop, use the most conservative tuning values.
11. Perform an open loop step test in the opposite direction as in step 8.
12. Compare PID tuning parameters or the models identified in the Process Model window.
13. Does the process respond differently in the up direction versus the down? If so, can you reduce
or eliminate the discrepancy? If not, use the more conservative tuning values.
When tuning a loop, it is important to keep in mind other factors that can affect the control loop.
For example, sensors must be properly located, calibrated correctly, and able to respond quickly
enough to expected process changes. Valves must be sized correctly. The entire design of the
control system affects how well the system can be tuned.
For more information on optimizing your loop, select Help > Optimization Steps on the
Faceplate or the first window in RSLoop Optimizer.
.....
This chapter provides a basic explanation of PID control, including an example of a simple control
loop. Explanations are also provided for the various tuning types that you can select when using
RSLoop Optimizer.
This chapter covers:
Description of Proportional, Integral, and Derivative Control
PID Loop Example
RSLoop Optimizer Theory
Tuning Types:
Load Tuning
Safety Factor
Setpoint Tuning
NOTE: The units for each gain parameter vary depending on the type of processor.
PID loop tuning is the procedure you perform to determine the best possible value for each gain
factor given the process you are controlling.
Determining if a system is well tuned is application-dependent; but in general, a well-tuned system
is one that:
Responds with little or no overshoot
Runs at maximum efficiency, closely tracking the set point
Provides the fastest response to an upset
Compromises between the above three factors
Reaches stability
In some systems, one factor might be more desirable than another. For example, some overshoot
might be tolerable to get a faster response.
Setpoint Error
In PID Equation Out
Process Controller
Variable Output
Temperature
Sensor Heating
Valve
This system might control the heating to a room in a building. The people occupying the room can
adjust the desired temperature by changing the setting (setpoint) on a thermostat. The temperature
sensor sends the actual room temperature (process variable) to the controller.
Room temperature could be affected by load changes — the outside air temperature might get
colder, causing the room to get colder, or many people could come into the room and turn on
computers or copy machines, causing the room to get warmer.
The difference between the setpoint and the actual temperature is called the error. The error signal
is fed to the PID equation and a new output value (controller output) is calculated and sent to the
heating valve to try and bring the temperature back to setpoint.
If the room gets colder, the difference between the current setpoint and the room temperature
increases. The heating control loop would have to respond to this increased error.
If the control loop is tuned properly, the system can respond quickly and efficiently. If the system
is not tuned properly, problems could occur.
If the room is heated too fast, for example, the room temperature might exceed or overshoot the
setpoint. The control loop must back down on the heating, but if it does this too fast, the room
temperature would go below the setpoint. The system might oscillate like this for a long period of
time before it settles down, if it ever settles down.
Heating the room too slowly would also cause problems. The occupants would be uncomfortable,
and keep changing the setpoint to make the system respond, causing the system to keep chasing the
new setpoint.
In either case, an improperly tuned system is inefficient, can cause equipment problems, and will
not achieve the desired goals of the system.
Setpoint
Offset
Process variable
The process variable in the graph does settle out, but it does not reach setpoint. The amount of
offset from setpoint depends on the amount of proportional gain and the conditions affecting the
control loop.
The proportional gain can be increased in order to decrease the amount of offset. However, too
much proportional gain can cause the controller to respond too aggressively, and the process
variable could overshoot the setpoint and then continue to oscillate. An example of this is shown
here.
Setpoint
Offset
Process variable
Setpoint
Process variable
With integral action, the change in controller output is proportional to the amount of time the error
is present. In the example above, the proportional gain responds to the setpoint change, and as time
passes, the integral action works to eliminate the offset and bring the process variable back to
setpoint.
The proper amount of integral action must be used. If there is too little integral action, the system
will be sluggish and take too long to get to setpoint. If there is too much integral action, the system
could end up oscillating and never settle down.
Setpoint
Process variable
The same amount of proportional and integral action was used in this example as was used in the
example of PI control. A small amount of derivative action was added to limit the overshoot and
cause the process variable to settle in to the setpoint a bit faster.
TUNING TYPES
Various types or methods of tuning exist to allow a system to be tuned optimally, depending on the
application and the type of changes to which the system might need to respond. RSLoop Optimizer
allows you to choose between two types of tuning.
Load tuning
Setpoint (Lambda) tuning
For most loops, we recommend that you use PI or PID fastest load tuning. Load tuning gives better
response to load upsets. If, however, you need tuning that gives no overshoot, use setpoint tuning.
The various methods RSLoop Optimizer uses for analysis are discussed in “Control Loop
Analysis” on page 123.
NOTE: You can use RSLoop Optimizer to help you design a setpoint filter. By using a setpoint
filter you can get good response to both setpoint changes and load upsets. (See “Control Loop
Analysis” on page 123 for more information.)
L oad Tuni ng
Load tuning gives PI and PID tuning parameters optimized for load changes at the controller
output. Load changes are the most difficult disturbances for the system to control. They are also
the most common.
There is always a trade-off between fast response and sensitivity to changing process conditions.
RSLoop Optimizer lets you further specify the type of load tuning you want to use, as well as a
safety factor to control the sensitivity of load tuning.
The fastest tuning with the lowest safety factor is the most sensitive to a changing process.
Conversely, the slowest tuning with the highest safety factor is the least sensitive. The sensitivity
of the loop can be analyzed using the Robustness plot, which is discussed in “Control Loop
Analysis” on page 123.
The safety factor setting can be between 1.0 and 5.0. A setting of 1 means the tuning is very
sensitive to small changes in the process but could become unstable if subjected to large or sudden
load changes. A safety factor of 5.0 gives you conservative tuning; the loop will respond somewhat
more slowly but is more stable when subjected to large or sudden load changes.
More conservative tuning is appropriate in most situations, and 2.5 is the default setting. For faster
response decrease the safety factor.
WARNING: Most control loops are somewhat non-linear. To be stable when the process
changes, most loops require tuning with a safety factor larger (more conservative) than 1.
If you enter a Safety Factor that might cause the system to be too sensitive, the message “Warning:
Safety factor makes the loop dangerously sensitive to process changes” is displayed.
You might then want to change the safety factor to a more conservative number, but this is only a
warning. You can still download the more sensitive value.
.....
This chapter provides you with information on the use of the basic windows of RSLoop Optimizer,
including the menu commands, displays, display options, and button functions.
This chapter covers:
Faceplate and Trend Window
Changing the Display of the Faceplate and Trend Window
Changing Controller Settings
Debugging Communications
Using the Off-Line Analysis & PID Tuning Screen
Creating a Report for a Control Loop
Setting Up an Extra Trend
Setting up Extra Loops
Using the Extra Trends and Extra Loops
Step-by-step procedures are also provided in the online help system of the software.
The details of control loop tuning using the methods available in RSLoop Optimizer are covered in
“Tuning Control Loops” on page 53.
NOTE: RSLoop Optimizer comes with a control loop simulation program that can be used to
help you learn how RSLoop Optimizer works without being connected to a process. To use the
control loop simulation:
FACEPLATE
. . . . . . . . . . . . . A. .N.D. .T. .R.E. N. .D. .W. .I N. .D. O. .W. . . . . . . . . . . . . . . . . . . . . . . .
The Faceplate and Trend window is the screen where you begin the process of tuning and testing
your control loops.
The Faceplate and Trend window displays the process variable (PV), setpoint (SP), and controller
output (CO) loop variables in a bar graph, as actual values, and in trend lines. Each is the same
color in each display.
NOTE: Controller output (CO) is sometimes referred to as the controlled variable (CV).
Or
From the Setup window: Click Faceplate.
The Faceplate for ... window opens. The example below is based on the simulate.tun loop.
Span settings
Archive On/Off
button
Menus
Bar graph
display
Real-time
trend display
Controller
mode
PID
parameter
values (see
units by
hovering over Buttons Trend time
values)
Archive On/Off Button: Allows you to turn the archiver on and off with just the click of this
button.
Real-time trend display: Displays real-time data from your processor
PV, CO, or SP display: If you hold the cursor over any point in the real-time trend display,
the PV, CO, or SP values at that time are displayed. The values that are displayed depend on
the Span settings. (See “Changing the Value of the Left and Right Axes” on page 32 for more
information.)
Buttons: Perform various commands
PID and Filter (F) parameter values: The current processor PID values and the new values
that will be downloaded to the process if Download is selected. In RSLoop Optimizer, the PV
filter value is also displayed. This value is also downloaded to the processor.
Controller mode: The current controller mode, auto or manual.
Bar graph display: Displays the loop variables in individual bar graphs and boxes.
TURN ON ARCHIVING
To collect and archive data in RSLoop Optimizer, you must be sure to turn the archiving function
on. To do so, click the Tape Recorder icon (shown at left). Only data that is collected after
archiving is turned on will be recorded.
NOTE: It is helpful to change background colors to remind you that archiving is on. It is
recommended that you set the Normal background color to black and the Archiving
background color to blue.
4. Click OK to save the information from the Trend Options window to your setup file.
NOTE: You can also change these options by clicking the Span button on the top left and right
of the Trend display).
3. Select the tab for the parameter whose variable display characteristics you want to change (PV,
CO, SP). The following options are available on each tab:
Engineering span: These values are for information only. They show the full
engineering range of the selected variable. The PV and CO engineering spans are the
spans used by your controller.
Display span: These spans are for the bar graphs and trend display only. Adjust these to
view the range of data you want displayed on the bar and trend displays. The smaller the
span, the higher the resolution on the display.
The display spans for the variable you select in the “Use XX for right/left axis” box are
shown in that variable’s color at the top and bottom of the trend graph. (See “Changing the
Value of the Left and Right Axes” on page 32 for more information.)
These spans are for the Faceplate display only. The display spans should be within the
engineering span. SP and PV display spans must be the same. If you change either the PV
or SP display span, RSLoop Optimizer prompts you and automatically changes the other.
On a new loop, RSLoop Optimizer automatically sets the display spans to a 25% range
around the current PV and CO values. If you set any span, RSLoop Optimizer will no
longer automatically set any display span for that loop.
Current value: This value is for information only. The current value is shown in its
appropriate color.
Line Color: Allows you to change the line color of the selected variable. Click the Line
Color button or the current value for the variable whose color you wish to change. Select a
new color and click OK. The new color displays in the Current value box on the dialog
box. If you change the PV, CO or SP colors, your color selections take effect for all of
your loops.
Decimal places: This option changes the number of places shown after the decimal point.
The number of decimal places is for the Faceplate window only. It has no effect on the
accuracy of the value stored in the archive files. Use the Current value window to see the
visual effect of changing the number of decimal places. The number of decimal places
must be in the range 0 to 5.
Line width: This option lets you define the width of the lines in the trend display. When
you select OK, your width selections for PV, CO and SP take effect for all your loops (the
width selections are saved in EXPERTUN.INI). The line width must be in the range 1 to 4.
Display on graph: Check or uncheck the box to hide or show the variable on the real
time trend display. If the variable is hidden on the trend display it is still read and
displayed in the digital value window, and archived to disk.
4. Click OK to exit this window and save your changes.
A sample of what you will see for each View option is shown below:
FACEPLATE ONLY
DEBUGGING COMMUNICATIONS
.........................................................
Data Spy allows you to display raw data before scaling, formatting, or adjusting decimal points.
Data Spy is available through the Faceplate and Trend window. To use Data Spy, select
Options > Data Spy. The Spy dialog box opens.
The Mode as ASCII chars box displays each character of the mode string as its ASCII value.
The type of communications being used (DDE or OPC) is shown at the bottom of the dialog box.
This window can remain open while other RSLoop Optimizer windows are active. It always stays
on top.
OPTIONS MENU
.........................................................
This menu contains five choices.
Trend Options
Display Spans, Colors, Decimals
DDE Spy
Bring back Previous PID settings to New
Tuning Report
FACEPLATE BUTTONS
.........................................................
The buttons underneath the Trend window on the Faceplate allow you to choose to tune from
previously archived data, use the AutoTune sequence, or close the Faceplate and Trend window.
Descriptions of each of the buttons are included below. Detailed use of the buttons in various
tuning functions is described in “Tuning Control Loops” on page 53.
The list shows the name of the archive file, the date and time when the data was collected, and
whether the file is an edited version of an archived file.
To work with an archived file, click the filename. The buttons below the archived file list are:
Tune: This button displays the Time data window. From this window, you can have
RSLoop Optimizer determine tuning parameters and perform analysis on the data. (See
“Tuning Control Loops” on page 53 for details on using the Time data window.)
Copy to ASCII: This button allows you to save your data to an ASCII file (extension .asc).
The file can be named and placed in any folder. You can also save the data as a print file (.prn)
or as a data file in comma separated value format (.csv).
Time Plot: The Time data window displays the process variable and controller output data.
Use this window to verify that your data meets tuning requirements, or to edit data to optimize
it before calculating new parameters.
Delete: This button deletes the selected archived data file from your hard disk.
NOTE: Delete removes the selected archived data file for the control loop, not the actual
control loop. To delete the control loop file, see “Editing and Deleting Loops” on page 14.
Back: This button closes the archive data file list window and returns you to the Faceplate
window where you can collect more data.
Change Notes: Allows you to edit the Loop Notes that appear on the right. It is a good idea to
make a small set of notes here for each archive file.
The Loop Notes box to the right of the archive list displays notes that have been entered
concerning the loop. RSLoop Optimizer automatically adds notes to the Loop Notes when a file is
edited. You can also enter notes manually by clicking Change Notes.
1. To edit the Loop Notes for a data set, select the data set.
2. Click Change Notes.
3. Type your changes in the Edit Notes window.
To start a new line in the Edit Notes window, press CTRL + ENTER.
4. Click OK to save your changes or Cancel to abandon them.
AUTOTUNE BUTTON
AutoTune starts the AutoTune sequence. This is a sequence of questions that you can follow to
have RSLoop Optimizer automatically calculate new PID tuning parameters for your control loop.
The AutoTune sequence is described in “Using AutoTune to Collect Data” on page 53.
CLOSE BUTTON
The Close button closes the Faceplate and Trend window and takes you back to the
RSLoop Optimizer main window. It does not close RSLoop Optimizer.
SIMULATE WINDOW
.........................................................
When you open the faceplate for a loop with a processor type of “Software simulation,” the
Simulate window opens minimized. This simulated control loop allows you to gain experience
with RSLoop Optimizer without being connected to a processor. The simulator lets you simulate
flow, level, temperature and pH loops; also typical loop control problems such as hysteresis,
stiction and non-linearity.
NOTE: The information in the Simulate window is the same as the information on the
Faceplate and Trend window, except for the Advanced button. Any changes to information in
the Simulate window will also be changed in the Faceplate and Trend window.
The functions you can do from the basic Simulate window are listed here. For information on the
Advanced button features, see “Advanced Simulate Window” on page 38.
Change the setpoint, Note: The controller output value can only be changed in Manual
controller output, or PID mode.
values Double-click the box.
Delete the old value and type the new value.
Click Enter.
Switch between Auto and Click the drop-down and select Manual or Auto.
Manual modes
Restore the initial tuning Click Initial PID. This option returns the initial tuning values,
parameters regardless of how many settings you have downloaded.
This is different than the Bring back Previous PID settings to
New option on the Faceplate and Trend window Options menu.
The Bring back option restores only the previous PID values, not
the initial values.
CREATING
. . . . . . . . . . . .A. .R. .E.P. O. .R.T. .F. O. .R. .A. .C. .O. N. T. .R.O. .L. .L.O. O. .P. . . . . . . . . . . .
You can use the tuning report to create a record of the tuning results for a control loop. Reports can
only be produced while tuning or for an archived data set. (See “Using Archived Data Files” on
page 57 for more information.)
RSLoop Optimizer creates reports in Microsoft® Word (version 97 SR-1 or higher). Word must be
installed in order to use the reporting feature.
You can create a report while tuning or from archived data.
To create a report:
1. Select tuning for the loop.
2. On the Faceplate and Trend window, select Options > Tuning Report.
3. Wait while the report is being prepared. Doing other work on your computer could interfere
with the insertion of graphics in your report.
4. Maximize Word to see the report.
5. To save the report, select File > Save in Word. RSLoop Optimizer automatically names the file
for you.
NOTE: When you open the report in Word, you will see a message that tells you the document
contains macros and asks what you want to do. If you disable macros, the document will open
as read-only and you will be unable to edit it. You must select Enable Macros to open the
report for editing.
ADDING TO A REPORT
When adding graphics to a report that is open, RSLoop Optimizer adds the item to the report. If a
report for this loop already exists and is not open, you can:
Use the Existing Report: The item is added to the add the item to the existing report.
Write Over: The existing report is replaced with a new report that contains only the item you
are adding. You can add the standard information to this report by using Options > Tuning
Report.
This table describes how to get data for the placeholders that are added to an RSLoop Optimizer
report. It also lists the bookmarks for each type of data, if applicable.
Location in
Where to find the data in RSLoop Optimizer Bookmark to select
report
Initial Tests, 1. Select an archived data file that was collected with the InitialTestAuto
Automatic Mode controller in Auto mode, or create one.
2. Tune from this data file to display the Time data
window.
3. Select Options > Report > Just Add This To
Report.
Initial Tests, Man- 1. Select an archived data file that was collected with the InitialTestManual
ual Mode controller in Manual mode, or create one.
2. Tune from this data file to display the Time data
window.
3. Select Options > Report > Just Add This To
Report.
Hysteresis Check 1. Open an archived data file collected for a hysteresis Hysteresis
check. (See “Hysteresis Check” on page 78.)
2. From the Time data window, select Options >
Hysteresis Check > Automatic.
3. Click OK on the Valve and noise diagnosis and
analysis message to close it.
4. On the Time data window, select Options > Report
> Just Add This To Report.
Hysteresis Check 1. Open an archived data file collected for a hysteresis none
(text data) check. (See “Hysteresis Check” on page 78.)
2. From the Time data window, select Options >
Hysteresis Check > Automatic.
3. On the Valve and noise diagnosis and analysis dialog
box, select Options > Add To Report.
Linearity 1. Open an archived data file collected for none
characterization. (See “Using the Output
Characterizer” on page 91.)
2. From the Time data window, select
Options > Characterizer.
3. Complete the process of building a characterization
curve. (See “Using the Output Characterizer” on
page 91.)
4. On the final Build A Characterizer window, click
Add To Report.
Asymmetry, Pro- 1. Collect data with the controller in manual mode and AsymmetryIncreasingPV
cess Model adjust the controller output (CO) so that the process
Increasing PV variable (PV) rises.
2. Tune the data.
3. Click Analysis on the faceplate.
4. From the Process Model window, select Options >
Add to Report.
Location in
Where to find the data in RSLoop Optimizer Bookmark to select
report
Asymmetry, Pro- 1. Collect data with the controller in manual mode and AsymmetryDecreasingPV
cess Model adjust the controller output (CO) so that the process
Decreasing PV variable (PV) falls.
2. Tune this set of data.
3. Click Analysis on the faceplate.
4. From the Process Model window, select Options >
Add to Report.
Closed Loop 1. Collect data with the old tuning parameters and Before_Setpoint
Before and After include a setpoint change.
(Before, Setpoint 2. Tune the data.
change)
3. From the Time data window, select
Options > Report > Just Add This To Report.
Closed Loop 1. Collect data with the old tuning parameters. Top Statistical analysis
Before and After 2. Tune the data. graphic: Before_Statistics
(Before)
3. From the Time data window, select
Options > Report > Just Add This To Report. Bottom graphic: Before
4. From the Time data window, select
Options > Statistical Analysis.
5. From the Statistical Analysis window, select Options
> AddToReport.
Closed Loop 1. Collect data with the new tuning parameters and After_Setpoint
Before and After include a setpoint change.
(After, Setpoint 2. Tune the data.
change)
3. From the Time data window, select
Options > Report > Just Add This To Report.
Closed Loop 1. Collect data with the new tuning parameters. Top Statistical analysis
Before and After 2. Tune the data. graphic: After_Statistics
(After, Statistical
3. From the Time data window, select
analysis)
Options > Report > Just Add This To Report. Bottom graphic: After
4. From the Time data window, select
Options > Statistical Analysis.
5. From the Statistical Analysis window, select
Options > AddToReport.
Power spectrum 1. From the Time data window, select First graphic: 1
Options > Power and either Power of Deviation, PV,
or CO. Second graphic: 2
2. From the Power Spectral Density window select
Options > Add to Report.
3. Display one of the other Power Spectral Density
graphs. Select Options > Add to Report.
Auto or Cross 1. From the Time data window, select none
Correlation Options > Correlation.
2. Choose one of the correlation graphs.
3. Select Options > Add to Report.
Ed iti ng a Re por t
Once created, you can edit your report to be specific to your company.
Replace all occurrences of “Company Name” with your company’s name. You can double-
click on text in the headers or footers to edit it.
Change the letter on the cover page to suit your needs.
If using RSLoop Optimizer, delete any blank graphs or data that you do not want to use.
Edit or add notes about the graphics in the report.
Summarize your findings in the Conclusions and Recommendations section.
NOTE: When you open the report, you will see a message warning that the document
contains macros and asking what you want to do. If you disable macros, the document will
open as read-only and you will be unable to edit it. You must select Enable Macros to open the
report for editing.
For information on using Word, or for customizing the report template, see the online help in
RSLoop Optimizer or Microsoft Word.
SETTING UP
. . . . . . . . . . . . . .A. N. . E. .X. T. R. .A. .T. R. .E.N. D. . . . . . . . . . . . . . . . . . . . . . . . . . .
You can add extra trend lines to the Faceplate and Trend window. The Extra Trend variable must
be in the same PLC as the loop.
In RSLoop Optimizer you can add as many trends as needed. The data from added trends in
RSLoop Optimizer are saved with the archives, so they can be used for tuning and analysis. Once
trends are created, you can set up extra loops to group the trends for analysis. (See “Setting up
Extra Loops” on page 47 for more information.)
RSLoop Optimizer also allows you to enter the filter address setup. (See “Filter Address Setup” on
page 47 for more information.)
If the Processor Type is set to Software Simulation, you will not see the Extra Trends on the
Faceplate and Trend window.
NOTE: If archives have been collected for a loop, trends cannot be deleted and
communication data cannot be changed. To delete trends or change communication data for a
trend, either:
Delete the existing archives for the loop. (See “Using Archived Data Files” on page 57 for
more information.)
Create a new loop by clicking Save As in the Setup window.
4. Click Extra Trends. The blank Setup Extra Trends window opens.
5. Click Add Trend. The Trend 1 tab is added, as shown here.
NOTE: The options in address info will reflect the communication package and processor
type selected in the RSLoop Optimizer Setup screen.
6. Set Eng span to the full engineering range of the trend variable.
7. Set Inst span to the instrument range of the trend variable. This is the range of the value
reported to RSLoop Optimizer by the server. When gathering data from a PLC, you may need
to set the Inst span to the Numeric range (0 to 40%, for example).
NOTE: When RSLoop Optimizer reads an Extra Trend value, it scales the values from the
instrument span to the engineering span. If the value is not scaled, set Inst span to the same
value as Eng span.
8. In the address info section, complete the boxes as needed for your communication system and
process.
Box Description
Description The name of the trend. This name is used on the Faceplate and Trend
window. PV, CO, and SP cannot be used as names of extra trends.
Display span The range of data that will be displayed on the Faceplate and Trend
window for this trend. The smaller the span, the higher the resolution on
the display.
Decimal places The number of digits shown after the decimal point in the Extra Trend
value displays. This only affects the display. Decimal places can be set
from 0 to 5.
Line width The width of the trend line in the Faceplate and Trend window. Line
width can be set from 1 to 4.
Box Description
Line color The color of the trend line in the Faceplate and Trend window. To
select a color:
Click Line Color.
Select the color.
Click OK.
Display on graph RSLoop Optimizer only. Select this box to display the new trend in the
Faceplate and Trend window. The new trend data is read and archived
even if it is not displayed.
10. Click Test to determine if your server is responding correctly to the address information.
RSLoop Optimizer reads the variable from the server and displays the variable or an error
message.
Both the raw instrument value and the scaled engineering value are shown. The raw instrument
value is the value that is actually reported by the server. The scaled value is in engineering units
and will appear on the Faceplate window. You can change its color or enter a different number of
decimal places. To change the color click the engineering value.
SETTING
. . . . . . . . . .U. P. . .E. X. T. .R.A. . L. .O.O. .P.S. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
In RSLoop Optimizer, you can trend and analyze more than one loop at a time. This allows you to
monitor other processes while collecting data for tuning and analysis. The types of processes that
can be monitored include cascade control, interacting loops, and feedforward control loops.
To monitor other process loops, you must have extra trends set up. (See “Setting Up an Extra
Trend” on page 44 for more information.) Once data has been collected from other process loops,
RSLoop Optimizer can be set up to analyze and tune these other process loops. To define an extra
loop, you must tell RSLoop Optimizer which extra trends belong to which extra loop.
If trends are not assigned to a loop, they are displayed on the Faceplate and Trend window, but
are not available for analysis.
1. In the:
RSLoop Optimizer Setup window: Click Advanced, then click Loop Setup.
2. To:
Add a loop: Click Add Loop. Type the name of the loop in Loop Description. This is the
description of the loop that will be used in the Time data window.
Edit an existing loop: Select the loop from the Loop Description list.
NOTE: The Faceplate loop cannot be changed in the Loop Setup dialog box. It must be
changed in the RSLoop Optimizer Setup dialog box. (See “Editing and Deleting Loops” on
page 14 for more information.)
3. Complete the boxes as needed. Box descriptions are provided in the following table.
Box Description
Process variable goes … when Select up if the process variable (measurement)
controller output goes up: increases when the controller output increases.
Select down if the process variable decreases when
the controller output increases.
Box Description
Controller type Select the controller from the list.
Proportional Band or Gain Display only. Proportional band and gain are related
by:
PROPORTIONAL BAND = 100/GAIN
Units on Integral (or reset) Display only. Shows the time units for Integral or
Reset action. This is the setting of the time action for
the derivative action. If integral is in minutes/repeat
or repeats/minute, derivative units will be in minutes.
P, I, and D Multipliers Display only.
Filter Units Selects the units to be uses by the PV filter analysis. If
you select Same As D, the units for the filter will be
the same as those your controller uses for its
derivative action.
Inverse Response Select this option only if you are sure you have an
inverse response process.
When the Inverse Response Process is selected,
RSLoop Optimizer assumes that the process has a
negative lead and an integrator (or long lag time)
combined with a lag and dead time. This type of
process is also described as having a shrink-swell
phenomena. With inverse response processes, when
the CO changes, the PV initially moves the in the
opposite direction of where it will end up.
Change Displayed Archive Determines which set of data is displayed in the graph
in the lower right of the dialog box. The graph is
scaled based on your display ranges. The display
ranges are saved in the .tun file.
The graph shows the archived data and helps you
choose which trend belongs to which loop. The graph
can also be used to manually examine interactions
between the trended variables.
Box Description
Description, Line, PV, CO, SP Select the PV, CO, and SP of the loop by clicking the
selection box in the appropriate column. A loop must
have a PV and CO. SP is optional.
When the Time data window opens, the trend selected
as the PV will be shown as the top graph. The trend
selected as the CO will be shown as the bottom graph.
(See “Using the Time Data Window” on page 65 for
more information.)
To turn off the display of a line, click its description.
The description button will appeared popped-out, and
the line will disappear.
USING
. . . . . . . .T.H. E. . E. .X. T. R. .A. .T. R. .E. N. .D.S. .A. N. .D. .E. X. .T.R. A. . .L.O. O. .P. S. . . . . . . . . .
In RSLoop Optimizer, the extra trends and extra loops are available through the Faceplate and
Trend window, Time data window, and Analysis windows. To set up extra trends for analysis, you
must configure extra loops.
In the Faceplate and Trend window, the trends are displayed on the graph and the value of one
trend at a time is displayed below the controller output value.
To view the
value of another
trend, select the
trend from the
list.
When the Time data window is open and extra loops are available, a box is added to the top of the
window. Select the loop you want to analyze from the drop-down list. The trend that was selected
as the process variable in the extra Loop Setup window is displayed in the top graph. The trend that
was selected as the controller output is displayed in the bottom graph. (See “Using the Time Data
Window” on page 65 for more information.).
.....
RSLoop Optimizer makes analyzing, optimizing, and tuning control loops fast, accurate, and easy.
You can simply follow an AutoTune sequence, or manually gather data and then have
RSLoop Optimizer calculate the tuning parameters. You can edit the data to optimize the new
parameters, and you can test the parameters before downloading them to your controller.
You can tune data online or offline. Online tuning is done from the Faceplate and Trend window.
Off-line tuning is done from the Off-line Analysis and PID Tuning window, which is selected by
clicking Off-line in the main window.
This chapter covers:
Collecting Data
Using AutoTune to Collect Data
Manually Collecting Data
Using Archived Data Files
Using the PID and PV Filter Display
COLLECTING DATA
.........................................................
With RSLoop Optimizer, you can follow the AutoTune sequence to determine PID tuning
parameters, or you can manually gather data and tune using that information. (See “Data
Collection Methods” on page 163 for more information.)
state settles out, both the process variable and controller output are relatively flat horizontal
lines in the Trend display, moving within the range of normal process noise.
RSLoop Optimizer analyzes process variable and controller output data pairs.
All process variable filtering must be removed from the signal.
WARNING: The data must not be from a load or process upset. Loads must not change
during the test and the range of test data should be as linear as possible. If a load change
occurs during the test, click End Sequence and begin the test again.
NOTE: You can stop the AutoTune sequence at any time by clicking End Sequence.
NOTE: You should allow for a least 30 samples of flat-line data before continuing.
4. When your process data is steady state, click Yes. The Performing Auto-Tune Sequence area
notifies you of the selected setpoint, as shown here.
To use a different value, click Different. You are prompted for a value. You can use negative
numbers if needed. Click Enter.
NOTE: If you are using the Simulate.tun file that comes with RSLoop Optimizer, you might
want to use a larger value. The default is not much larger than the process noise, so you will
get better data if you use 10 or 15.
6. Click OK.
7. If the loop is in Auto mode or you are tuning a simulated loop, go to step 10.
a) If the loop is in Manual mode, the following prompt displays:
To get good data for tuning, RSLoop Optimizer needs to see the process variable respond
to the controller output. The amount of process variable response needs to be at least 4 to 6
times larger than the normal peak-to-peak noise in your process.
c) After the process variable moves by this amount, click Yes. RSLoop Optimizer changes
the controller output back to its original value.
8. The Performing Auto-Tune Sequence area prompts “Is the process settled out?”
WARNING: Plant data taken for RSLoop Optimizer analysis and tuning must have all process
variable filters removed from the signal.
NOTE: The data must not be from a load or process upset. Loads must not change during the
test and the range of test data should be as linear as possible. If a load change occurs during
the test, stop collecting data and start over.
COLLECTING DATA
In some cases, you might want to collect data manually instead of using the AutoTune sequence.
This is a basic procedure to collect data manually. This is a closed loop test with the controller in
the Auto mode.
Examples of other methods of collecting data manually are provided in “Application Notes” on
page 163.
NOTE: The data must not be from a load or process upset. Loads must not change during the
test and the range of test data should be as linear as possible. If a load change occurs during
the test, stop collecting data and start over.
1. Make sure the controller output is not at 0%, 100%, or saturated into a limit. If it is, change the
controller output to between 5% and 95% (or not at a limit). Valves are usually non-linear at
their limits.
2. Let the loop settle out (reach steady state).
3. Select Archive > Archive On (or click the Tape Recorder icon).
4. Change the controller setpoint by about 10%.
5. Wait for the process variable to respond an appreciable amount, then change the setpoint back
to its original value. Skip this step if your process can tolerate a new operating point.
6. Let the loop settle out (reach steady state).
7. Select Archive > Archive Off.
ARCHIVING DATA
Data collected with the AutoTune sequence is automatically archived.
You can also manually archive data:
1. Select Archive > Archive On.
2. RSLoop Optimizer starts archiving data to a file with the same root as the .tun file.
For example, if the .tun file is FIC100.tun, the first archive file will be FIC100.001, the next
archive will be FIC100.002, and so on.
3. When you are done collecting data, select Archive > Archive Off.
When you turn archiving on, RSLoop Optimizer displays the name of the file where the data is
archived in the title bar of the Faceplate and Trend window:
Off Line Analysis and PID Tuning window: In the main window, click Offline. The
Offline button is available after data has been archived.
The list of files looks like this in the Faceplate and Trend window:
It looks like this in the Off Line Analysis and PID Tuning window:
NOTE: The Time and Date shown in the “Collected” column indicate the time and date the
archive was started, based on the computer’s clock.
NOTE: This function does not delete the Control Loop file (.tun). See “Editing and Deleting
Loops” on page 14 for information on deleting a control loop configuration file.
NOTE: Only a small portion of the information that is available is shown in the following
example. Depending on the number of trends created and the amount of archived data
collected, you will normally see more information.
Header Section
==============
Number Of Header Items: 40
Scan Interval: 1
Time Stamp Units: milliseconds
Number Of Channels: 3
Date/Time Data Collected: 10/16/97 11:06AM
Version: 9.06.16
Who Created Me: RSLoop Optimizer
Current P: 2
Current I: .58333
Current D: 0
Current F: 0
Channel 1 Description: PV
Channel 1 Eng. Min. Span: 0
Channel 1 Eng. Max Span: 100
Channel 2 Description: CO
Channel 2 Eng. Min. Span: 0
Channel 2 Eng. Max Span: 100
Channel 3 Description: SP
Channel 3 Eng. Min. Span: 0
Channel 3 Eng. Max Span: 100
Response Direction: up
Controller Algorithm: Allen Bradley PLC-5 PD file ISA, D on PV
Controller Type: ideal
PB Or Gain: gain
Integral Units: min/rep
PB Multiplier: 1
Integral Multiplier: 1
Derivative Multiplier: 1
Mode: 1
PID: 4
TimePlot AutoScale PV: 0
TimePlot AutoScale CO: 0
TimePlot PV Min: 41.63655
TimePlot PV Max: 49.87732
TimePlot CO Min: 39.31247
TimePlot CO Max: 55.48609
TimePlot Number of Lines: 0
TimePlot Line Position1: 0
Timeplot Line Position2: 0
Timeplot Line Position3: 0
Timeplot Line Position4: 0
Timeplot Line Position5: 0
Timeplot Line Position6: 0
Timeplot Line Position7: 0
Timeplot Line Position8: 0
Timeplot Line Position9: 0
Timeplot Line Position10: 0
Timeplot Line Position Hyst1: 0
Timeplot Line Position Hyst2: 0
Data Section
============
PV CO SP
----------- ----------- -----------
41.46768 39.78901 42
41.30512 40.99653 42
41.62125 41.36135 42
41.5027 39.8148 42
41.64124 40.96721 42
41.89695 38.49328 40.85577
42.13604 31.08119 37.29519
42.30678 25.70074 35
42.32682 24.43569 35
42.26803 23.97164 35
41.42966 23.6704 35
39.54797 25.02115 35
The Current PIDF settings are the PIDF values currently being used by the controller. The
New PIDF settings are the values that have been calculated through AutoTune or tuning, or
entered manually. They will be downloaded when you click the Download button. When values
are downloaded, the New settings are updated to reflect changes in the archived data or tuning. If
no changes have been made, N/A displays.
The Current and New F settings are the process variable filter times, which are set during tuning
by the Use... PV Filter option. They can also be manually entered. The filter time units default to
the same as the units for the derivation but can be set in the Loop Setup window. (See “Setting up
Extra Loops” on page 47 for more information.) If the value of P or I is changed, the Performance
Increase is recalculated and the analysis windows update if they are open.
NOTE: Download is only available from the Faceplate, Trend and Tuning window.
WARNING: Downloading different gain values while the processor is in the Run or Remote
Run mode might cause unexpected machine operation, which can result in injury, loss of life,
or machine damage.
To download the new PIDF parameters to the controller or the software simulator:
1. Click Download on the Faceplate and Trend window.
3. Click Yes.
After the values are downloaded, the Current and New parameters should be identical.
NOTE: The Filter parameter must be configured before it can be downloaded. To use a filter,
you must modify the PLC code. (See “Filter Address Setup” on page 47 for more information.)
.....
This chapter covers:
Changing the Time Data Window Display
Calculating Tuning Parameters
Controller Tuning
Editing Data in the Time Data Window
Verifying Data Using the Time Data Window
You can also access the output Characterizer from the Time data window. (See “Using the Output
Characterizer” on page 91 for more information.)
NOTE: Before working with data, it might be helpful to review “Data Pair and Sample Interval
Requirements” on page 57.
The Time data window displays the process variable and controller output data. Use this window
to verify that your data meets tuning requirements, or to edit data to optimize it before calculating
new parameters.
This window opens when:
The AutoTune sequence is complete
You select an archived data file and click Tune
You are in the Off-line Analysis and PIDF Tuning window and click Time Plot or Tune
This window can be moved anywhere on your screen and remains on top of other
RSLoop Optimizer windows. A sample Time data window is shown here.
Zoom > Zoom out to see all data Zoom out to see all data
Options > Report > Full Tuning Report Full Tuning Report
CHANGING
. . . . . . . . . . . .T. H. .E. .T. I.M. E. . .D. A. .T.A. .W. .I.N.D. O. .W. . D. .I.S.P. L. .A.Y. . . . . . . . . . .
You can scale the Y-axis or change the line weight and type of graph that is displayed.
2. If:
Auto Scaling is checked (the default), select Auto Scaling. This turns off auto scaling and
opens the Set Scale dialog box.
Scaling is not checked, select Set Min, Max Scale.
3. Type the values that you want for the y-axis.
4. Click OK.
When the Time data window opens, RSLoop Optimizer calculates and displays New PIDF
parameters for the control loop. It also adds the Controller Tuning, Probable Performance Increase
(or Are Current Settings Stable), and Analysis options to the Faceplate and Trend window.
The PIDF parameters are based on the data collected for your loop. More specifically, the data
shown in the Time data window is used for all calculations. If you modify or zoom in this
window, the results will change.
Before the loop is tuned, the New column displays N/A (not applicable) for each parameter. The
parameters in the New column are recalculated as you make changes in the Controller Tuning
section of the Faceplate and Trend window and as you make changes to data in the Time data
window.
You can also change any of the New tuning parameters yourself. Click the box of the parameter
you want to change, edit the value, and press Enter.
The Probable Performance Increase is updated as the PIDF parameters change and tells you how
much better the controller will respond to a load upset with the new PIDF parameters. It is the
percent probable performance increase in integrated absolute error (IAE) to a load upset assuming
the controller does not overshoot to a load upset with either the current or new parameters. If the
controller overshoots, the performance increase is an approximation that is roughly proportional to
the IAE improvement. It is still a good predictor of performance.
This box says None if the Probable Performance Increase is insignificant. Adding derivative might
increase the performance in these cases. (Click the Use derivative if possible box.)
CONTROLLER TUNING
.........................................................
When the Time data window opens, you can select the controller tuning factors that you want to
use for this loop.
In RSLoop Optimizer, the tuning factors include a filter option that allows you to experiment with
types and sizes of filters.
NOTE: See “Tuning Theory” on page 21 for definitions of load tuning, setpoint tuning, safety
factor, and other RSLoop Optimizer parameters.
Set Point Tuning: Provides good setpoint response at the expense of good regulation.
Lamba tuning: Select lag rule or Intg (integrator) rule. These options are grayed out if
they are not applicable to the process you are working on.
NOTE: To see what the PIDF settings would be for each type of tuning, click PID Grid, where
you will be able to change the settings and see the effects in one window.
3. For:
Load Tuning: Type a Safety Factor between 1 and 5, with 1 being the most sensitive.
Set Point Tuning: Type a Response time (in seconds). For a faster response, use a
smaller response time. For slower response, use a larger response time. The default value
that RSLoop Optimizer sets is conservative to prevent overshoot on setpoint change.
Lambda Tuning: Type a Lambda time (sec). The lambda time sets the speed of
response. Adjust the lambda for the first order response time to setpoint you want. For
faster response, use a smaller lambda. For slower response, use a larger lambda.
4. To use derivative gain, select Use derivative if possible.
5. If you want to use filtering:
a) Select the Use PV Filter checkbox.
b) Select the type of filter from the PV Filter list.
RSLoop Optimizer calculates the largest possible filter without degrading performance.
Simulation and robustness plots update instantly to reflect your filter type and size.
6. Click Done tuning. The Time data window closes. If you are using the Off-line Analysis and
PID Tuning window, the PIDF parameter display is also closed.
USING
. . . . . . .T. H. E. . PID
. . . . G. . R. I.D. .T.O. .C. H. .E.C. K. .R. .E.L.A. T. I.V.E. .R. E. .S.P.O. N. .S.E. T. .I M. .E. .
Relative response time (RRT) is an indicator of the speed of the control loop. The smaller the
relative response, the faster the loop; the larger the response, the slower the loop. You can use this
number for comparisons.
The values for RRT are provided in the PID Tuning Grid. To open the grid, open the Time data
window and click PID Grid.
To derive the relative response time, RSLoop Optimizer:
Calculates the closed loop frequency response to a load upset
Finds the frequency value where the amplitude ratio peaks
Converts this frequency value to the period
If you use the faster tunings (Load fastest or Load fast), the PV might oscillate around the setpoint.
The period of this oscillation in the time response is roughly the same as the relative response time.
Change the relative response time by adjusting the safety factor or lambda time.
You can adjust the safety factor and the lambda time directly from the PID Grid making it easier to
directly see the effects on the RRT. Simply click on the set of values you would like to work with
then change them at will.
To prevent interaction in interacting loops, adjust the safety factor or lambda value (setpoint
speed) so that the relative response time is different by a factor of 5. For example, if you have 3
interacting loops and the fastest one has a period of 1 second, adjust tuning in the other two to
result in a relative response time of at least 5 and 25.
If you have a cascade loop, the master loop should have a relative response time that is 3 times
slower (larger) than the slave loop. Tune the slave loop and then tune the master. If the master has
a relative response time less than 3 times the slave loop, detune the master to get a relative
response time of 3 times the slave.
EDITING DATA
. . . . . . . . . . . . . . . .I.N. .T. H. .E. .T.I.M. E. . D. .A. T. .A. .W. .I .N.D. O. .W. . . . . . . . . . . . . .
Data in the Time data window can be zoomed, averaged, edited by line, and filtered.
After you edit the data, RSLoop Optimizer automatically recalculates and displays the PIDF
tuning, based on the changes.
ZOOMING
After collecting data, you might want to zoom in on the data you want to use for tuning. This is
typically the smallest window of data that starts at a steady state condition and ends within about
5% of steady state. You can zoom manually or use Auto Zoom. When Auto Zoom is used,
RSLoop Optimizer automatically tries to zoom in on the data that offers the best tuning.
3. Click again where you want to end zooming. The portion of the graph you selected is
highlighted.
4. To change the zoomed area, click a new spot. You can do this as many times as needed.
Clicking the graph anywhere to the right of the midpoint moves the right edge of the
highlighted area. Clicking to the left of the midpoint moves the left edge.
To include all of the data points to the left, click just outside the left side of the graph. To
include all of the data points to the right, click just outside the right side of the graph.
To stop zooming, click Cancel Zoom.
5. Click Zoom Now or click the Zoom In button.
You can zoom in as many times as needed.
To return to the previous view, select Zoom > Zoom Back to Previous.
To return to the view of all data, select Zoom > Zoom Out or click the Zoom Out button.
Us ing Au to Zoom
To use Auto Zoom:
1. Select Zoom > Auto Zoom. The data is highlighted.
2. Click Zoom Now to accept the data.
Auto Zoom attempts to zoom based on the current data. If RSLoop Optimizer does not have the
proper data to perform Auto Zoom, the error message “Cannot automatically find a zoom location
in the data” is displayed. You can then either zoom manually, or recollect data and retry Auto
Zoom. (See “Data Collection Methods” on page 163 for methods of collecting data.)
AVERAGING DATA
RSLoop Optimizer uses your initial data as the steady state basis for all tuning and modeling.
Ideally, this data should start and end at steady state. However, if your initial data is noisy or
cycling, averaging might improve your tuning results. Average the controller output and the
process variable data to just where it starts changing due to the bump introduced as part of the test.
Process
Variable
4. To change the area, click at the new spot. You can do this as many times as needed. Clicking
the graph anywhere to the right of the midpoint moves the right edge of the highlighted area.
Clicking to the left of the midpoint moves the left edge.
To include all of the data points to the left, click just outside the left side of the graph. To
include all of the data points to the right, click just outside the right side of the graph.
To stop averaging, click Cancel Avg.
5. To modify your data, click on the graph at the new beginning or end point of the data you wish
to average.
6. To accept the change click Avg now.
You can undo the last averaging operation by selecting Edit > Undo or clicking the Undo button.
4. RSLoop Optimizer changes the data to be a line between the two points you picked. You can
edit as many data points as needed.
Noise spikes
have been
edited into a line.
5. Click OK.
To undo the line editing session, select Edit > Undo or click the Undo button. You can edit both
the PV and OP data.
FILTERING DATA
1. Select Edit > Filter (or click the Filter button). The Process Variable Filtering dialog box
opens.
2. Select the Filter type from the list. For filter descriptions, see the online help.
3. Type the Filter time constant in seconds. Pick a filter time large enough to eliminate as much
noise as possible without degrading loop performance. If the filter time constant is too large, it
can hurt loop performance, because the control system will try to control the filter.
WARNING: Filtering modifies your plant data. If you use a PV filter, subsequent tuning will be
based on the filtered process variable.
4. Click Test to test the effects of filtering. The graph in the Time data window is redrawn with
the filter values applied.
5. Click OK to accept the filter values or Cancel to ignore the changes.
To undo the filtering change, select Edit > Undo or click the Undo button.
SAVING CHANGES
After making editing changing in the Time data window, you can save the changes by selecting
Edit > Save changes. The edited data is automatically saved in a new archive data file.
RSLoop Optimizer keeps both the original and the edited data file or files. The edited file is
indicated by “Yes” in the Edited column. RSLoop Optimizer automatically adds a note that
identifies the original file.
The file that the
edited file was
created from.
Yes indicates
that data in this
file was edited.
STATISTICAL ANALYSIS
Statistical analysis can be used to check the improvement in control with the new tuning
parameters. Before tuning, collect the normal operating data (with no setpoint changes) with the
controller in Auto. Download the new tuning parameters and collect normal operating data again.
Compare the statistics for both sets of data.
Statistical analysis will be applied to the selection in the Time data window. When using
statistics, be sure to compare apples to apples. For example, you may wish to compare statistics
for steady-state performance with old tuning and new tuning. In this case, you will want to select
data where there are no setpoint changes.
NOTE: Be sure to turn Archiving on when collecting data so you can save and compare the
two sets of data. Select Archive > On to start archiving. Select Archive > Off when data
collection is complete.
5. Select Options > Statistical Analysis. The Statistical Analysis dialog box opens, along with
the Histogram of PV window. (See “Hysteresis Check” on page 78 for more information.)
The Time data window also changes to show the statistics as mean, standard deviation, and 2x
standard deviation lines on the process variable graph.
2x Standard
deviation (green
lines)
Standard deviation
(blue lines)
The statistical information can be added to the loop report by selecting Options > Add to
Report.You will be given the option to add the data to the Before Statistics or the After Statistics
section of the report.
Definitions of the statistics are:
Sample (raw): The data collection sample interval in the archive.
∑
2
( mean – x ( i ) )
sample variance = -------------------------------------------------------------
number of data points – 1
IAE - Integrated Absolute Error: Performance of a loop is judged by how closely the
controlled variable is held to the setpoint, both for setpoint changes and load changes. There
are many mathematical ways of judging this, and IAE is one type. IAE is a measure of
controller error defined by the integral of the absolute value of a time dependent error
function. When comparing IAE statistics, be sure to select equivalent time ranges for
comparison. The IAE calculation is not normalized for time.
CO Statistics: This capability is available in RSLoop Optimizer only. Travel and Reversals:
is the amount of valve travel and reversals. Travel is the total distance the valve has traveled in
the CO showing in the time graph. Reversals is the number of times the valve has changed
direction in CO showing in the time graph. Use these numbers along with the time simulations
to analyze and reduce valve wear. You should collect data from normal operating conditions in
automatic for these tests.
Set Travel and Reversals per Hour, Day, Week, Month, or Year to see how much your valve
would have travelled or reversed during that time period. For example, if your travel is
100%/day this means the valve would move a distance of its entire range in one day: it could
move from 0% to 100%, or it could move back and forth from 50% to 60%, 5 times each way.
HISTOGRAM
The histogram option allows you to see the statistical spread of your data. Move the cursor over a
bar on the Histogram window to see the percentage of your data that is in this bar.
You can display a histogram of error, PV (as shown below), or CO. With a histogram of error, if
the data are grouped on one side of zero, a non-linearity is indicated.
In general, you will expect to see a bell-shaped curve for the histogram. A differently shaped
curve, such as the one shown here, bears further investigation. For example, a bi-modal curve (two
bell shapes separated by a flat spot in the middle) is an indicator of stiction in the valve.
HYSTERESIS CHECK
RSLoop Optimizer can perform a hysteresis check on your data. Hysteresis is a dead band in the
movement of a device upon a change of direction.
For example, if a valve is moving in one direction, it continues to move if the air signal keeps
moving in the same direction. But when the air signal changes direction, the valve has to stop and
reverse motion. Conditions like static friction and lost motion in linkages take time to overcome.
This time, or hysteresis, must be compensated for by the air signal increasing to an amount equal to
the hysteresis before the valve will move in the other direction.
If RSLoop Optimizer cannot calculate hysteresis from your data, an error message is displayed.
Click Yes to this message to display information on how to properly gather data to perform a
hysteresis check.
If your hysteresis is more than 1% for valves with positioners or 3% for valves without positioners,
consider repairing or changing equipment to try to reduce the hysteresis.
NOTE: To properly assess hysteresis, the size of the bumps must be larger than the
hysteresis. The trouble is, you don’t know how much hysteresis exists before you do the tests.
If you make a CO bump, and nothing happens, then that bump does not count. You must
make another bump.
4. To check automatically: Select Options > Hysteresis check > Automatic or click the
Hysteresis button. The time data graph changes as shown below.
5. To check manually: Select Options > Hysteresis Check > Manually Choose.
Four vertical lines are displayed on the graph. Drag the first three lines to a settled area
before each of the three controller output changes. Place the fourth line to a settled area
near the end of the graph (after the last CO change). To move a line, move the cursor over
it until the cursor turns into a double-sided arrow ( ). Hold down the left mouse button
and drag the line.
Click Find.
6. The areas RSLoop Optimizer used to calculate hysteresis are indicated on the graph.
RSLoop Optimizer finds each of the three areas where steady state occurred in the process
variable after a step change. The areas are:
p2: From the first steady state area to the second
Valve Oversized
An oversized valve cannot provide very fine control. Any valve has a rangeability factor that
determines how precisely it regulates. A good control valve might have a rangeability of 100 to 1,
which means you can make 1% adjustments in the output. The valve can also be referred to as
having a resolution of 1%. This allows finer control with less chance of inducing hunting or
cycling.
If a valve is oversized, the usable operating range of the valve gets smaller. For example, a valve
with a rangeability of 100 to 1 could be oversized for an application to the point that its effective
usable operating range is only 50 to 1 (process gain of 2). This translates to a resolution of 2%.
The oversized valve can only be adjusted within 2% instead of 1%. It no longer has tight control,
and the process might be hunting within the valve resolution. The oversized valve could also be
operating on its seat. A command could be sent to the valve, but it might not move off its seat since
control isn’t precise.
The solution to an oversized valve is to change the valve trim or replace the valve with one of the
appropriate size.
Valve Undersized
An undersized valve can be a safety problem since it might not have the range to control the
process. The solution to an undersized valve is to change the valve trim or replace the valve with
one of the appropriate size.
Resolution
of 0.024% Span of 100 to 500
Resolution of 0.1 degree
You are measuring temperature and one of the transmitters has a span of 0 to 4096. With the
12-bit converter described above, you are only able to measure and control within 1 degree:
4096 * 0.024% ≈ 1
The other transmitter has a span of 100 to 500 degrees. Your resolution would improve
approximately 0.1 degrees: 500 – 100 = 400 400 * 0.024% ≈ 0.1
Now your system could detect a change of 0.1 degrees instead of 1 degree.
If you only need to detect and control a measurement within a certain range, choose a transmitter
span that represents what you need.
NOTE: If a PV signal is scaled within the PLC ladder logic, you may lose some resolution for
the control. You should take extra care to ensure that the PV signal reaching the PID block
has adequate resolution for your control needs.
Options > Report > Just Add This to Report: Places this graphic in the tuning report for
this loop. RSLoop Optimizer displays a list of options for placing the time plot into the report.
Options > Copy to Clipboard: Copies the data to the clipboard from which it can be pasted
into the appropriate section of the report.
(See “Creating a Report for a Control Loop” on page 40 for more information on reporting.)
Cumulative Power is only available if Linear is selected. This option displays the Cumulative
Power Spectral Density graph, which quickly shows you how many of the cyclic disturbances
are below a certain frequency. Cumulative power is used to detect frequency ranges where the
power increases. It is also useful for determining the frequency at which the power stops
increasing. It is a sum of all the power normalized so that the total is 100%. (The Deviation,
PV, or CO on the Y-axis is displayed as a percentage.) The standard power spectrum might not
easily show the contribution of a frequency to the total noise present. A frequency on the
power spectrum could be 10% of total noise or 90%; Power Spectral Density alone does not
show this. Cumulative Power shows that a certain percentage of the cyclic disturbances are
below a specific frequency.
Lines: Display the graph using lines. Lines are useful for finding tendencies and trends. This
can be misleading, however, if the distance between points is not constant, as when the
frequency is displayed in seconds. This is an example of a display using lines. In this case, it
might be better to use bars (see the example for Bars, below).
Bars: Bars are useful for seeing the exact location of the points, the number of points, and the
value for each point.
Report > Add to Report: Places this graphic in the tuning report for this loop. (See
“Creating a Report for a Control Loop” on page 40 for more information.)
Omit Zero Freq (DC) omits frequencies close to zero from the graph and adjusts the Y-axis
accordingly. This zooms in on the data as shown in the examples below.
USING CROSS
. . . . . . . . . . . . . . . .A.N. D. . .A. U. .T.O. .C. .O. R. .R.E. L. A. .T.I.O. N. . . . . . . . . . . . . . . . .
The correlation options (listed below) help you check loop interaction and detect hidden cycling.
Cross Correlation of PV and CO: Cross Correlation can help you determine whether two
loops are interacting with each other.
Auto Correlation of PV and CO: Auto Correlation helps detect hidden cycling. Oscillations
are more evident in the Auto Correlation graph than in the Time data window.
To display a correlation graph:
1. From the Time data window, click Options > Correlation.
2. Select one of the following options:
Cross Correlation of PV and CO
Auto Correlation of PV
Auto Correlation of CO
In the cross correlation graph, the correlation shown on the graph should be near zero if there are
no dependencies. The closer the correlation is to 1 or –1, the greater the interaction between the
two loops. For most normal situations, the correlation bars should shrink away to inside the blue
bars within 10 dead times.
In the auto-correlation graph, a reduction in area under the auto-correlation curve tells you if
tuning has improved the control by making it more random.
The Loop Summary Table calculates the average tuning values and also the most conservative
tuning values. Some loops are non-linear and behave differently at different ends of the range.
Some loops behave differently in one direction to the other. For example, temperature loops
usually heat up quicker than they cool down. With these loops, you should analyze the loop at
different ends of the range and/or in both directions and use the most conservative tuning values.
COLUMN DESCRIPTIONS
Archive column: The archive file number from which the tuning values were obtained.
Zoom start, end columns: The portion of the archive file from which the tuning values were
obtained.
P, I, D, F columns: The recommended tuning values.
Lag1 column: If RSLoop Optimizer modelled the process to have a single or double lag the first
or only process lag is shown.
Lag2 or Lead column: If RSLoop Optimizer modelled the process to have a double lag or a lead,
the second process lag or the lead time is shown. A process lead is only shown if the loop is an
inverse response process.
Intg column: If RSLoop Optimizer modelled the process as an integrating process, ‘yes’
displays.
Stability column: Relative Stability Index. This shows the most conservative or most stable
tuning as 100. All other tuning rows are listed relative to the most conservative row. They have
values less than 100 indicating the amount they are less conservative. For example, a value of 50
indicates that tuning is twice as aggressive as the most conservative tuning. When you add another
row of tuning values to the table, the entire column may change.
RRT column: The relative response time of the loop with these tuning values.
Units row: The second row of the table (in grey) shows the units for the model. If different models
contain different units then the Loop Summary table converts the models with longer time units to
those of shorter ones. For example, if one model had units in seconds and another in minutes, the
model with units in minutes would be converted to seconds.
Add button: Click the Add button to add an extra line of tuning values to the Loop Summary
table. The values added correspond to the PID tuning values currently visible in the column
headed ‘New’.
Remove button: Click the Remove button to remove the line that is currently highlighted.
Copy to New button: Click this button to have RSLoop Optimizer copy the PID and F values on
the currently highlighted line to the ‘New’ column. It is the reverse of the Add button.
Double-clicking on a line of PID and F values has the same effect as clicking the Copy to New
button.
Right-click on the window for other methods of copying PID and F values to the New column:
Add the window to a Microsoft Word document
Copy the window to the clipboard as a bitmap so that it can be pasted into any document.
.....
RSLoop Optimizer provides both output and pH (input) Characterizers. This chapter only
discusses the output Characterizer. (See “Using the pH Linearizer” on page 109 for more
information.)
This chapter covers:
Example of When to Use the Output Characterizer
Building a Characterization Curve
Using Characterizer with RSLogix Frameworks
Examples of Using the Characterizer Code in RSLogix 5
Many control loops are difficult to tune because they are non-linear. This means that the process
gain changes as a function of the measurement or controller output. Without any linearization, the
controller needs to be tuned for the condition where the process gain is the highest. This results in
sluggish tuning everywhere else. Linearizing these loops improves control because the controller is
better tuned over the complete operating range.
The output Characterizer takes the output of the controller and transforms it to a value so that the
control loop is linear over the entire range of control. A control loop with a characterizer should
allow you to have robust optimal tuning parameters anywhere over the CO or PV range.
Use the output characterizer:
If servo control is important
If you need to linearize flow loops, jacket temperature in split-range chemical reactors, or
slave loops in cascades
For any loop where the setpoint will change
If the linearity of the loop changes with load or production rate
Characterizer can supply code in FORTRAN, Basic, C, or X-Y line endpoints. If you are using
RSLogix Frameworks, you can download the X and Y values directly to the Lin function. Use the
code to program the PID loop in your ladder logic. (See “Using Characterizer with RSLogix
Frameworks” on page 100 for ladder logic examples.)
PV Output
to valve
With a characterizer, you might be able to get uniform performance across the entire range of your
process. This allows your process to run optimally without de-tuning for oscillations.
Characterizer tests the final control element with the rest of the process to guarantee a good fit
between the characterization curve and the installed characteristic of the final control element and
process.
BUILDING
. . . . . . . . . . . A. . C. .H. A. .R.A. C. .T.E. R. .I Z. .A.T. I.O. N. . .C. U. .R.V. E. . . . . . . . . . . . . . . . .
Before using the output Characterizer, make sure the controller is tuned to stabilize the loop at that
value of controller output where the loop gain is highest.
You must complete these steps to use the Characterizer:
Collect process data
Start the Characterizer
Run the Build Characterizer Wizard
Manual mode: Change the controller output to 10% of the process variable.
STARTING CHARACTERIZER
1. On the Faceplate and Trend window, click Tune from archived data.
2. Select the archive file you just created.
3. Click Tune.
4. In the Time data window, select Options > Characterizer.
5. Place a vertical line over each steady state area and double-click. This places a line at that
point, as shown in the example below.
NOTE: RSLoop Optimizer saves your characterization curve. Therefore, if you have already
built a characterization curve for this archive, the lines are already placed. You can go directly
to the Build Characterizer Wizard or adjust the line positions first.
Added lines at
each steady
state area
Adding a line: Place the cursor where you want the line and double-click.
Moving a line: Position the cursor over a line (cursor changes to a double ended arrow),
hold down the left mouse button and drag.
Deleting a line: Position the cursor over the line (cursor changes to a double ended
arrow) and double-click.
6. Click Build. Follow the instructions in “Run the Build Characterizer Wizard,” below.
R unn ing th e B uild C har acte rize r Wizar d wit h P iec ewi se L inea r f it
1. If you haven’t done so, on the Time data window, click Build.
NOTE: If you have already built a characterization curve for this archive, the final characterizer
screen, shown on page 96, is displayed.
2. The Build Characterizer Wizard starts and displays a graph of PV vs. CO, similar to the one
shown here.
Each point plotted (with a +) on the graph corresponds to one of the lines you placed on the
time graph. Look at this graph to verify your process characteristic.
Does the curve represent your process? If not, click Cancel and collect another set of data.
Below the graph, RSLoop Optimizer lists the process’s lowest and highest gains and their
ratio. The ratio of the highest gain to lowest gain should be less than two, with an absolute
maximum of three. RSLoop Optimizer comments on the linearity of your loop and
provides some direction on the benefit of characterization for your loop. If the loop does
not require characterization, click Cancel.
If characterization seems appropriate, click Next to continue. (You might see some graphs
flash briefly across the screen.)
3. Select Piecewise Linear as the type of fit.
4. Click Next.
5. Use the Lines to Use list box to select the number of line segments to use.
6. Click Next.
A graph is displayed that shows both the characterizer lines and the process characteristic.
Drag the red and blue boxes on the characterizer so that the line matches the process
characteristic.
To move a box, place the cursor over the box. The cursor changes to crosshairs (+) over red
boxed or a double-ended arrow over a blue box. Drag the box as needed.
After moving the red and blue boxes, the graph looks something like:
NOTE: Use the Auto Adjust button to automatically move the red design points between the
actual process characteristics. With the cursor, space the red design points as you want them
and click Auto-Adjust to move them up or down to the proper location.
7. To add a box, place the cursor where you want the new box. Double-click.
8. To delete a box, place the cursor over the box and double-click.
9. Click Next.
You might not be able to get the curves to match exactly at this point. The last wizard screen
will allow you to expand the graph above and below the current endpoints, giving you more
room to move the blue endpoints. This can help you achieve a closer match.
10. Select the programming language for your characterizer code. Click Finish.
If you selected X-Y pair list, be aware that X is the vertical axis (the characterizer input).
11. On the final screen, you can make changes to the characterizer and copy the characterizer
code. The buttons and options are described in “Final Characterizer Wizard Screen” on
page 99.
R unn ing th e B uild C har acte rize r Wizar d w it h H ype rb olic Fit
1. If you haven’t done so, on the Time data window, click Build.
NOTE: If you have already built a characterization curve for this archive, the final characterizer
screen, shown on page 96, is displayed.
2. The Build Characterizer Wizard starts and displays a graph of PV vs. CO, similar to the one
shown here.
Each point plotted (with a +) on the graph corresponds to one of the lines you placed on the
time graph. Look at this graph to verify your process characteristic.
Does the curve represent your process? If not, click Cancel and collect another set of data.
Below the graph, RSLoop Optimizer lists the process’s lowest and highest gains and their
ratio. The ratio of the highest gain to lowest gain should be less than two, with an absolute
maximum of three. RSLoop Optimizer comments on the linearity of your loop and
provides some direction on the benefit of characterization for your loop. If the loop does
not require characterization, click Cancel.
If characterization seems appropriate, click Next > to continue. (You might see some
graphs flash briefly across the screen.)
3. Select Hyperbolic fit.
4. Click Next.
A graph is displayed that shows both the characterizer line and the process characteristic.
5. Drag the red and blue boxes on the characterizer so that the line matches the process
characteristic.
NOTE: Use the Auto Adjust button to automatically move the red design points between the
actual process characteristics. With the cursor, space the red design points as you want them
and click Auto-Adjust to move them up or down to the proper location.
After you adjust the line, the graph looks something like this.
NOTE: You might not be able to get the curves to match exactly at this point. The last wizard
screen allows you to adjust the graph to get a closer match.
6. Click Next.
7. Select the programming language for your characterizer code.
8. Click Finish.
9. On the final screen, you can make changes to the characterizer and copy the characterizer
code. The buttons and options are described in the following.
Process characteristic and characterizer: You can edit the graph in the same manner as
in step 6.
Choose The Type of Fit: Change the fit type to Piecewise Linear or Hyperbolic. You need
to edit the graph again as described in the specific procedures.
Hyperbolic Fit: (Hyperbolic Fit Characterizer only.) The linearity constant for the hyperbolic
equation.
Zoom: Click Expand Top or Expand Bottom to extend the display of the graph above or
below, respectively. You can then drag, create, and delete boxes above or below the original
graph in the same manner as in step 6. Once expanded, additional buttons become available:
Shrink Top returns the top of the graph to its original position (any changes made in the area
above the graph remain). Shrink Bottom performs the same function as Shrink Top, except
on the bottom of the graph, and Expand All or Shrink All expands or contracts both.
Auto Adjust: The Auto Adjust button automatically moves all the red design points between
the actual process characteristics. With the cursor, space the red design points as you want
them and click Auto Adjust to move them up or down to the proper location.
Language for Characterizer Code: Select FORTRAN, Basic, C, X-Y pair list (Piecewise
Linear Fit only) or Structured text for the programming language.
NOTE: With the Structured Text option, you can copy the resulting code from the Code to
use area into a control block that is using structured text language and your loop will be
linearized.
Create Code: Updates the code if you made changes to the characterizer graph.
Report: Creates a report with, or adds to an open report, a graphic of this dialog. (See “Using
RSLoop Optimizer” on page 27 for more information about reporting.)
Close: Closes the dialog.
Gains: Displays Linearity Info (maximum and minimum gain, loop linearity, etc.).
The information on this screen is saved so that the next time you run the characterizer for the same
archived data file, it is immediately display so that you do not have to complete all the steps in the
Build Characterizer Wizard again.
USING CHARACTERIZER
. . . . . . . . . . . . . . . . . . . . . . . . . .W. I.T. H. . R. .S. .L.O. G. .I .X. .F. R. A. .M. E. .W. O. .R. K. .S. .
The X,Y values can be directly downloaded to the Frameworks Lin function block.
In RSLogix Frameworks:
Connect the CV_EU output of your PID function block to a Lin function block.
Configure your Lin function block as a straight-through transfer function. Set SegTot=1,
X1=0, X2=100, Y1=0, Y2=100.
Name your Lin function after the PID function block. For example, if you named your PID
block TIC1, name the Lin function block TIC1_Lin.
Make sure the Engineering Units Min and Max for the CV_EU output of the PID function
block are 0 and 100.
PID
PID
0000
Control Block PD12:0
Process Variable N10:4
Tieback 0
Control Variable N13:0
Setup Screen
0012 (END )
Y = CO / 100
X = Y / (0.3492 + (1 – 0.3492)*Y)
Valve = X* 100 + 0
PID
PID
0000
Control Block PD12:0
Process Variable N10:4
Tieback 0
Control Variable N13:0
Setup Screen
CPT
0001 Compute
Dest N7:0
0<
Expression N13:0 | 100
CPT
0002 Compute
Dest N7:1
0<
Expression N7:0 | (0.3492+((1.0-0.3492)*N7:0))
0008 (END )
.....
Many control loops are difficult to tune because they are non-linear. This means that the process
gain changes as a function of the measurement or controller output. Without any linearization, the
controller needs to be tuned for the condition where the process gain is the highest. This results in
sluggish tuning everywhere else. Linearizing these loops improves control because the controller is
better tuned over the complete operating range.
The pH Linearizer takes the input of the controller and transforms it to a value so that the control
loop is linear over the entire range of control. A control loop with a characterizer should allow you
to have robust optimal tuning parameters anywhere over the PV range.
Use the pH (input) Linearizer for:
pH loops
Oxidation-reduction loops
Temperature control loops on distillation columns, where the temperature profile is steep in
the center
pH Linearizer can supply code in FORTRAN, Basic, C, or X-Y line endpoints. If you are
programming the PID loop in ladder logic, see “Examples of Using the Linearizer Code in
RSLogix 5” on page 114.
This chapter covers:
Using the pH Linearizer
Programming the pH Linearizer into Your PLC
Examples of Using the Linearizer Code in RSLogix 5
USING
. . . . . . . .T.H. E. . P. .H. . L. .I N. .E.A. R. .I Z. .E.R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The pH Linearizer wizard walks you through designing a pH Characterizer.
Before using the pH Linearizer, you need to:
Know the ranges for your pH transmitter
Collect data
To use the pH Linearizer:
1. On the main window, select the loop.
2. Select Options > pH Linearizer. The Enter pH transmitter ranges dialog box opens.
NOTE: Use Remove Last Row to delete any empty rows before clicking Next.
NOTE: After you enter your titration data and click Next, your titration data is saved to the .tun
file. Returning to the pH characterizer will bring up the last data you entered, even if you
cancel before completing the pH Linearizer wizard.
6. Click Next. The Build a Characterizer dialog box opens. An example is shown here.
Process
variable (pH)
scale
Titrated
volume scale
Each point plotted (with a +) on the graph corresponds to one of the values you entered. Check
this graph to verify your process characteristic.
Does the curve represent your process? If not, click Cancel and enter another set of data.
Below the graph, RSLoop Optimizer lists the process’s slopes. The lowest slope is the
lowest gain and the highest is the highest gain. The ratio (Max/Min) of the highest gain to
lowest gain should be less than three, preferably less than 2. RSLoop Optimizer also
indicates the linearity of your loop and provides suggestions for how to improve it. If the
loop does not require characterization, click Cancel.
If characterization seems appropriate, go to the next step.
7. Click Next. (You might see some graphs flash briefly across the screen.) The Build a
Characterizer dialog box opens.
8. In the Build a Characterizer dialog box, you can drag the red and blue boxes as needed to
change the process characteristic.
NOTE: Use the Auto Adjust button to automatically move the red design points between the
actual process characteristics. With the cursor, space the red design points as you want them
and click Auto-Adjust to move them up or down to the proper location.
NOTE: If you selected X-Y pair list, be aware that X (the characterizer input) is the vertical
axis.
11. Click Finish. On the final screen, you can make changes to the characterizer and copy the
characterizer code.
Process characteristic and characterizer: You can edit the graph in the same manner
as in step 8.
Language for Characterizer Code: Select FORTRAN, Basic, C, X-Y pair list
(Piecewise Linear Fit only) or Structured text for the programming language.
NOTE: With the Structured Text option, you can copy the resulting code from the Code to
use area into a control block that is using structured text language and your loop will be
linearized.
Create Code: Updates the code if you made changes to the characterizer graph.
Report: Creates a report with, or adds to an open report, a graphic of this dialog. (See
“Creating a Report for a Control Loop” on page 40 for more information about report
functions.)
Gains: Displays the gains of this loop.
PROGRAMMING
. . . . . . . . . . . . . . . . .T.H.E. . P. H. . .L. I.N. E. .A. R. I. Z. E. .R. .I .N.T. O. . Y. .O. U. .R. .P. L. .C. . . .
The pH Linearizer characterizer must be applied to both the pH signal and the setpoint. The PLC
must see both of these signals characterized.
To program the pH Linearizer into your PLC:
1. Scale your controller between 0 and 100%.
2. Display the uncharacterized pH.
3. Set the setpoint entry to a point before characterization.
4. Setpoint and process variable values for display and operation should be reading values before
the characterizer block.
Setpoint
Characterizer Controller
Setpoint
Operator
Setpoint
PID Controller
Operator
Process Controller
Variable Process Variable
Process
Variable
Characterizer
PID
0000 PID
Control Block PIDA
Process Variable PV_LIN
Tieback 0
Control Variable PIDA_CV
Setup Screen
LES CPT
0001 Less Than (A<B)
Compute
Source A PV_PH
Dest PV_LIN
0<
0<
Source B 2.301
Expression (36.8 * PV_PH)-73.56
2.301<
.....
The word stiction is made from the words ‘stick’ and ‘friction’. It is the resistance to the start of
motion. Stiction or stick-slip occurs because of tight packing, undersized actuators, or corroded
stems. Sticking at the seat is very common especially for valves designed for tight shutoff.
For an example of stiction, think of moving a desk on a hardwood floor a quarter of an inch. When
you start pushing on the side of the desk the pressure builds up until it finally breaks loose and
moves farther than you wanted it to. Valve stiction is similar — where pressure builds up and the
valve finally moves, resulting in the valve settling in the wrong place. The effect is that stiction
always produces an oscillating process variable.
Of all the valve problems, stiction is the most harmful to a process, since the controller will push to
move the valve, until the process variable reaches the set point. If stiction is present, the valve will
move too much and the process variable will go too far. Then, the controller output will reverse
direction and the same thing will happen again. The limit cycle caused when the control valve
sticks and then suddenly slips during a change in input signal is called stick/slip cycle.
This chapter covers:
The Stiction Wizard
Stiction Check
Using the Stiction Wizard
STICTION CHECK
.........................................................
Using RSLoop Optimizer’s stiction checker, you can get a quantitative measurement of the stiction
in a loop. This makes it easier for you to diagnose lack of performance.
2. After the process variable has settled out, collect a little more data (45 seconds or more) to get
normal process noise, then double-click Controller Output and quickly increase it by 5%.
3. Wait until the process variable (PV) settles out (wait at least 30 seconds) and increase the
controller output (CO) quickly by another 0.2%.
4. Repeat Step 3 until the PV moves from the change you made.
5. In the RSLoop Optimizer PID Analyzer/Tuner, zoom in on the steps, being sure to get initial
steady state data, and select Stiction Check from the Options menu.
After you have collected the data, the archived data should have small steps similar to the sample
below.
You will be able to see the stiction where the PV line suddenly realizes that the CO has been
increasing all along while it has been stuck on the same path. It will then jump up to the CO line
and resume on track (the jump is circled in the illustration above).
USING
. . . . . . . .T.H. E. . S. .T. I.C. T. I. O. .N. .W. .I .Z.A. R. .D. . . . . . . . . . . . . . . . . . . . . . . . . . .
1. On the Faceplate and Trend window, click Tune from archived data.
2. Select the archive file you just created.
3. Click Tune.
4. In the Time data window, select Options > Stiction Check.
5. A message box prompts you to drag lines before each CO change. There are already lines
placed on the graph because RSLoop Optimizer makes an attempt to place them for you
initially, you are just “fine-tuning” the settings.
6. To Add A Line: Move the cursor to a settled area you would like to mark. Double-click or
right-click and select Add. Add a line for each settled area in the data. The dashed line is
currently selected.
To Move A Line: To position a line, drag it with the mouse: Position the cursor over a line
(cursor changes to a double-headed arrow: ↔), hold down the left mouse button and move the
cursor.
To Delete A Line: Position the cursor over the line (cursor changes to a double-headed arrow:
↔). Double-click or right-click and select Delete.
7. Once your lines are placed correctly click Next on the upper left corner of the graphs.
8. Another message box prompts you to drag the lines to mark the stiction band (once again
RSLoop Optimizer has attempted to place them for you).
9. Red line: The length of the red line indicates the time the PV has been averaged. If you feel
the red lines are not long enough, repeat the test increasing the time between each small CO
change. The height of the red line is the average value of PV during this time.
Yellow line: The yellow lines extend as far as the previous red line so you can easily see if the
PV has moved.
10. Mark Stiction Band: RSLoop Optimizer has placed two vertical lines on the graph,
attempting to automatically mark the COs of the stiction band. Move these COs to accurately
mark the stiction band, if necessary. The first line should mark the settled time before the first
small CO change. RSLoop Optimizer has probably placed this line correctly. The second CO
should mark the settled time that the PV actually moved from one of the small steps you made.
11. Click the Next button at the top left of the graphs.
12. The results are displayed, giving you the % Stiction, along with the High controller output and
the Low controller output used in the calculation of the % Stiction.
13. The stiction information can be added to the loop report by clicking the Report button.
.....
RSLoop Optimizer provides advanced analysis tools that you can use to test your tuning
parameters before downloading them to the controller. Five of the analysis tools are available by
clicking Analysis on the Faceplate. They include:
Process Model
Process Frequency Response (Bode) Plot
Control Loop Simulation Plot
Robustness Plot
Performance Summary
These tools can be used to do “what if” analysis of your tuning parameters before you download
them to the controller.
RSLoop Optimizer also provides these analysis tools:
Time Line Analysis
Performance Evaluation
Power Spectral Density plots
Auto- and Cross-Correlation
NOTE: The quality of the information in the analysis windows depends on the quality and
relevance of the data that you collected. RSLoop Optimizer models are based on frequency
response. Frequency response is based on the data shown in the Time data window.
USING
. . . . . . . .T.H. E. . S. .T. A. .N.D. A. .R.D. . A. .N. A. .L.Y. S. I. S. . T. .O. O. .L.S. . . . . . . . . . . . . . . .
1. To use the Process Model, Process Frequency Response, Control Loop Simulation, Time Line
Analysis and Robustness plots:
2. Open the Faceplate and click Tune from archived data or
3. Select the loop name and click Off-line
4. Select the archive file to tune.
5. Click Tune. The Time data window opens.
NOTE: You must work with an archive file that contains at least one valid bump. The bump
can be from data collected in Manual or Auto mode.
Analysis button
(Check your
data by clicking
this button)
NOTE: If you close one of the analysis windows, all of the analysis windows close.
When you change a PID tuning value, the Control Loop Simulation and Robustness plots
update immediately. See “Robustness Plot” on page 133.
7. Begin working with the plots by selecting a model from the Process Model window.
PROCESS MODEL
.........................................................
The Process Model window shows the process model type identified for your control loop.
RSLoop Optimizer uses your process frequency response to identify the model as first order,
second order, or integrating process.
NOTE: The quality of your models, simulation, and robustness depend on the quality and
relevance of your time data.
1. Adjust the plot windows so that you can see the Process Model, Process Frequency Response,
and Robustness Plot.
2. Check the data on the Process Frequency Response and Robustness Plots.
3. In the Process Model window, select a different Model type from the list (Force steady state
gain or Allow gain to float). The plots change to reflect the new model.
4. Check the data on the Process Frequency Response and Robustness Plots.
5. In the Process Model window, select the Model type that fits your data the best.
NOTE: The Process Frequency Response plots both the actual and the model data. Choose
the model that is closest to the actual data of Amplitude Ratio and phase. The most important
frequencies are where the phase is between approximately 90 and 180 degrees (phase lead)
or –90 and – 180 degrees (phase lag).
The Robustness Plot shows the trade-off between tight tuning and sensitivity to process
changes. If you have two sets of tuning with about the same robustness, you can generally
choose the faster response.
6. Once you’ve determined the best Process Model, you can use the Process Simulation window
and the Robustness Plot to analyze the calculated and current PID parameters.
OPTIONS
. . . . . . . . . . I.N. . T. H. .E. .P. R. .O. C. .E.S. S. . M. . O. .D.E. L. . W. . I. N. .D.O. W. . . . . . . . . . . . . .
The table in this window shows the gain, dead time, and time constant values for the selected
Model type. If RSLoop Optimizer identifies the model as second order, you see two Time constant
entries in the table: 1st Time constant and 2nd Time constant. If your process type is integrator, the
first order time constant value displays the word integrator. (The integrator process is discussed
more in “Application Notes” on page 163) If the model is second order and the roots are
imaginary, the time constant displays the word imaginary.
MODEL TYPE
All of the analysis windows depend on the model type chosen. Two models are available. When
you choose a different model, all the analysis windows update. The model types are:
Allow gain to float: Causes RSLoop Optimizer to ignore steady state information and fit the
model based on higher frequencies. These higher frequencies are more important to optimal
and stable closed loop control.
Force steady state gain: Forces the steady state gain to be equal to the amplitude ratio at the
lowest frequency. The model is fit using the remaining good frequencies. The lowest (or zero)
frequency values for amplitude ratio and phase are not plotted on the frequency plot, so at the
lowest plotted frequency, the model frequency response might not appear to match the
process.
If you selected the Inverse Response option in the Advanced window during loop setup, the model
will also show a lead time and integrator.
ADDING TO A REPORT
Adds the information from this window to the report. Y This option is available by selecting Add
to Report from the Options menu. See “Creating a Report for a Control Loop” on page 40.
The Sample interval is the interval between moves or between coefficients of the step response
vector. Enter a number equal to or greater than the sample interval of your process data.
Click Save vector to create a text file with the coefficient vector representing the step response of
the process model.The standard Windows Save dialog box is displayed. You can accept the default
file name and location or specify your own. The file contains three columns of numbers. The first
column contains the vector coefficients, the second column contains the coefficient counter or
index, and the third column contains the time.
The left vertical axis is the scale for the amplitude ratio in decibels. Two plots of amplitude ratio
are provided, one based on actual data and the other based on the Model type.
The right vertical axis is the scale for phase angle in degrees. Two plots of the phase angle are
provided, one based on actual data and the other based on the Model type.
The amplitude ratio and phase angle are plotted against frequency in radians/time (logarithmic
scale) on the horizontal axis. The lines are color coded.
The process frequencies most important to process control are those between about minus 90 and
minus 180 degrees of phase.
In Process Frequency Response plots for first order time constants, the corner frequency (or cut-off
frequency) is where the amplitude ratio starts to turn and the phase lag equals 45 degrees. Consider
a sine wave input to a process at this corner frequency. The output of a first order process has half
the input power at the corner frequency.
A lag time or first order time constant of 10 has a corner frequency of 1/10. A PID controller with
an integral time of 10 also has a corner frequency of 1/10. A PID controller with a derivative time
of 10 has a corner frequency of 1/10; however, for derivative, this represents a phase lead of 45
degrees rather than phase lag.
To find these frequencies:
1. Look at the red scale on the right of the plot. Locate –90 and –180.
2. Find the stretch of the red phase angle line between –90 and –180. These are the most
important phases.
3. See where the –90 and –180 phase points line up vertically with the blue amplitude ratio line.
These are the most important amplitude ratios.
The Process Frequency Response plot is based on process data only, and therefore does not update
if the PID tuning changes. It only updates when the Model changes. This can happen if you change
the model type, or if you zoom in to a different bump in the Time data window.
The Options menu on the Process Frequency Response window allows you to:
Display the plot in Radians/sec. This is the default.
Display the plot in Cycles/sec.
Show High Frequencies. With this option checked, RSLoop Optimizer shows amplitude ratio
and phases for frequencies with phases lower than –270 degrees.
Add to Report. See “Creating a Report for a Control Loop” on page 40 for more information
on reports.
If you change the model of your process, or change a PID tuning parameter, the simulation is
instantly recalculated to reflect your changes.
All of the plots show the response for both the current and new PID parameters for comparison.
New PID parameters are shown in red and current parameters are shown in blue.
Options > Add to Report: See “Creating a Report for a Control Loop” on page 40 for more
information on reports.
Time Range > Expand or Shrink: The scale on the Control Loop Simulation is initially set
by RSLoop Optimizer. To increase the scale, select Expand. To decrease the scale, select
Shrink.
SETPOINT PLOT
The Setpoint Plot shows the response to a setpoint change for the current settings and the new
settings.
This is the first plot displayed if you selected Setpoint tuning on the Faceplate. If it is not
displayed, select Set Point plot from the Options menu.
A setpoint plot is shown here.
S etp oint Fi lt er
When the Set Point Control Loop Simulation plot is selected, you can also apply and change the
Set Point Filter.
You can achieve good response to both setpoint changes and upsets with the setpoint filter. With
many loops, tuning that is good for reducing the effects of process upsets results in setpoint
response with lots of overshoot. The Set Point Filter conditions the setpoint command before it
reaches the controller in order to dampen the overshoot.
1. Make sure Set Point Plot is selected on the Control Loop Simulation window Options menu.
2. Select Options > Set Point Filter > Use Lead/Lag Set Point Filter. RSLoop Optimizer
applies a filter to the setpoint of the simulated loop and modifies the Control Loop Simulation
plots to show the results. The filter is applied to the simulation of the New PID settings.
The PV plot displays the filtered setpoint in black. The PV and CO with the filter applied are
shown in magenta.
RSLoop Optimizer calculates the lead time and displays lead time, lag time, and the lead/lag
ratio at the top of the Control Loop Simulation window.
3. You can also adjust the ratio of lead to lag time. RSLoop Optimizer automatically sets the lag
time to approximately the integral time (in seconds). To adjust the ratio, select Options > Set
Point Filter > Adjust Lead/Lag Set Point Filter. Enter a number between 0 (full filtering)
and 1 (no filter) with up to four places after the decimal point (for example, 0.0625).
Variable Description
SPuserSet User entered setpoint
LagTime lag time (from Control Loop Simulation window)
LeadTime lead time (from Control Loop Simulation window)
SampleInterval calculation interval for this lead lag code (same as controller scan rate)
y1 last value of y
LOAD PLOT
The Load plot shows a simulation of a process upset. RSLoop Optimizer applies a step load
change to the simulation.
This is the first plot displayed if you selected Load Tuning on the Faceplate. If it is not displayed,
select Load plot from the Options menu.
This example shows the response to a process upset. The upper portion of the window shows the
process variable response and the lower portion shows the controller output response.
To reduce wear, use PV filtering or adjust the PID values. If your system allows it, try adding the
filter value that RSLoop Optimizer recommends first. If this does not reduce wear enough,
consider modifying the tuning parameters. When adjusting the PID values, keep in mind that:
Removing derivative action results in the greatest reduction in valve movement.
Reducing proportional action will usually be more effective than reducing integral action.
For the least wear, select PI tuning with a large safety factor or lambda. Although the
robustness will be good, there will be poor response to setpoint changes and upsets.
There is always a tradeoff between fast tuning, robustness, and valve wear. For least wear,
select PI tuning with a large safety factor or lambda.
ROBUSTNESS PLOT
.........................................................
The RSLoop Optimizer Robustness plot shows how sensitive your loop is to process gain or
process deadtime changes. The Robustness plot graphically shows the tradeoff between tight
tuning and stability. If you have two sets of tuning parameters with about the same robustness, you
can generally choose the faster response. An example Robustness plot is shown here.
New
Current
The two axes of the plot are process gain and dead time.
The plot has a region of stability and a region of instability. The red (new tuning parameters) and
blue (current tuning parameters) lines on the robustness plot are the limit of stability. To the right
and above the lines, the closed loop process is unstable. To the left and below the lines, the closed
loop system is stable.
The crosshairs in the safety factor area are at the actual process gain and deadtime. Generally, a
safety factor or divisor of 2 is reasonable for a control loop. For practical system stability, keep the
limit of stability line outside the safety factor area.
For example, in the Robustness plot shown above, the line for the current tuning parameters is
inside the safety factor area. These parameters offer very tight tuning, but if the system encounters
a high gain ratio, it will be in the region of instability. The line representing the new tuning
parameters is well outside the safety factor area, offering more stability.
Robustness plots are calculated using the Process Model and controller tuning values. The
accuracy of the plot depends on the accuracy of the model.
Whenever you choose a different model in the Process Model window, the Robustness plot
refreshes. This plot also updates when a new Safety Factor value is entered on the faceplate.
RSLoop Optimizer conservatively estimates process dead time when dead time gets close to the
sample interval. If your plant has very small dead time, the model appears less stable and the verge
of stability lines is lower. The simulations (which use the model) also appear less stable than your
plant.
process gain
gain ratio = -------------------------------------------------------------------------------------------------
process gain the controller was tuned for
process deadtime
deadtime ratio = -------------------------------------------------------------------------------------------------------------
process deadtime the controller was tuned for
At the cross, where both ratios are 1, the process gain and deadtime are at the process values you
tuned for.
Ad d t o R epor t
See “Creating a Report for a Control Loop” on page 40 for information on reports.
PERFORMANCE SUMMARY
.........................................................
The performance summary (shown below) allows you to view overall trade-off’s of performance,
robustness and valve wear at a glance. This table provides a listing of Performance increase,
robustness increase, valve travel index, valve reversal index, Current RRT, and New RRT.
Valve Reversal Index is the percent improvement in controller output reversals from the current
to new. If the new PID tuning parameters (and filter) result in more reversals than the current ones,
the Valve Reversal Index will be negative.
Current RRT (sec) is your current relative response time. See the glossary for a definition of
RRT.
New RRT (sec) is the value RSLoop Optimizer uses to replace the current relative response time.
PERFORMANCE
The red (new) performance bar is inversely related to the current safety factor.
The red performance bar does not take into account any manual changes you may have made to the
new values, however the blue bar is always set relative to the red one. The blue bar is set based on
the performance increase. If the performance increase is negative then the blue bar will be higher
than the red one. In this case, the current settings will probably perform better to load upsets than
the new ones.
ROBUSTNESS
RSLoop Optimizer sets the bars for robustness equal to the closest distance for the current and new
tunings.
VALVE DUTY
To get the best Valve Duty analysis, set the Control Loop Simulation to Measurement Noise
Response so the valve travel and reversals correspond to an operating control loop. The bars show
valve travel and reversals
NOTE: The time line analysis can be added to the full loop report in Microsoft Word.
In general, for optimal performance, the times for each item should be in the following order on the
time line:
Sample Interval
Filter Time
Derivative Time
Dead Time
Equivalent Dead Time (if a second order process)
Integral Time
Relative Response Time (RRT)
Each of these items is marked on the time line if they are not zero.
So that you can easily compare the time of each component of the control loop, RSLoop Optimizer
converts each component's time into seconds. The Integral and Derivative times are taken from the
New column. For easy comparison, the controller Integral and Derivative times have been
converted from Series or Parallel to equivalent Ideal controller Integral and Derivative tuning
constants (unless your controller already uses an Ideal structure).
If the process is second order (contains 2 lags) then an equivalent dead time is calculated and
marked. For the Time Line analysis equivalent, dead time is defined as:
Equivalent Dead Time = Dead Time + Smaller of the 2 lags
The time axis is on a logarithmic scale. To see precise times, hover the mouse cursor over the
graph and the Tool Tip Help will list the time where the mouse is currently located.
1. With off-line data collected, click Tune, then the Analysis button on the faceplate (as shown
below).
TIME EVALUATIONS
In general, the relative times of each item should be in the order listed above. Where possible,
RSLoop Optimizer evaluates appropriate relative times and places a colored bar with a message
between the times. The colors have the following meanings:
Green - Excellent or OK
Red - Improvements can be made to significantly increase performance (see example below)
Sometimes there is not much room to see the text in the colored evaluations. To see the text for the
evaluation, hover the mouse cursor over the text and the Tool Tip Help will display the full
message. Making the Time Line graph wider may also help to see a message. Make the Time Line
taller if some of the evaluations are cut off. RSLoop Optimizer evaluates the following:
1. Sample Interval compared to Dead Time
2. PV Filter compared to Controller Derivative Time
3. Controller Derivative compared to Controller Integral Time
4. PV Filter compared to Dead Time or Equivalent Dead Time (if second order)
NOTE: RSLoop Optimizer only performs evaluations when the values are not zero.
.....
RSLoop Optimizer requires a DDE or OPC topic for each control loop you need to communicate
with. See “OPC Topics” on page 145 for information on using OPC topics.
WHAT
. . . . . . . I.S. . A. . D. .D. .E. .T. O. .P. I.C. ?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A topic is a path from Windows’ Dynamic Data Exchange (DDE) to a processor. When a DDE-
compatible Windows application makes a DDE request for information from a communication
package such as RSLinx, it must specify a topic name and an item. The information stored in the
topic determines:
Which communications Driver Type and Driver Number to use
Whether to access a local or remote station
What station number to access
What kind of processor is at that station number
How to communicate with the processor: poll or unsolicited
How long to wait before registering a time-out error
You can also have multiple topics to communicate with several different processors at once. The
minimum number of topics needed in a project is one for each different processor with which you
wish to communicate. For example, you might have a topic named Poll 25 that communicates with
a PLC-5 processor at station 25 and a topic named Poll 18 that communicates with a PLC-5/250
processor at station 18.
Multiple topics can be used in reference to the same processor. One topic called Poll 25 might poll
station 25 for data. Another topic called Wait 25 might wait for an unsolicited message from
station 25. You could even have additional topics that poll station 25 at different poll rates.
NOTE: Different topics can be used simultaneously, and different DDE-compatible Windows
applications can use the same topics simultaneously.
Computer
1784-KT
10
PLC-5
50
SETTING UP
. . . . . . . . . . . . . .A. .D. .D. E. . .T.O. .P.I .C. .I N. . .R. S. .L. I.N. X. . . . . . . . . . . . . . . . . . .
1. Open RSLinx.
2. Select DDE/OPC > Topic Configuration.
3. Browse to and highlight the PLC you wish to communicate with.
4. Click New.
.....
RSLoop Optimizer requires an OPC or a DDE topic for each control loop you need to
communicate with. See “DDE Topics” on page 79 for information on working with DDE topics.
WHAT
. . . . . . . I.S. . A. .N. .O. P. .C. . T. .O. P. .I C. .?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A topic is a path from Windows’ OLE for Process Control (OPC) to a processor. OPC is a
communications protocol that provides a defined set of COM interfaces that, in turn, provide data
access functions. OPC can be used by the server as a more flexible and powerful alternative to the
Dynamic Data Exchange (DDE) communications protocol.
When an OPC-compatible Windows application makes an OPC request for information from a
communication package such as RSLinx, it must specify a computer name, topic name and an
item. The information stored in the topic determines:
Which communications Driver Type and Driver Number to use
Whether to access a local or remote station
What station number to access
What kind of processor is at that station number
How to communicate with the processor: poll or unsolicited
How long to wait before registering a time-out error
You can also have multiple topics to communicate with several different processors at once. The
minimum number of topics needed in a project is one for each different processor with which you
wish to communicate. For example, you might have a topic named Poll 25 that communicates with
a PLC-5 processor at station 25 and a topic named Poll 18 that communicates with a PLC-5/250
processor at station 18.
Multiple topics can be used in reference to the same processor. One topic called Poll 25 might poll
station 25 for data. Another topic called Wait 25 might wait for an unsolicited message from
station 25. You could even have additional topics that poll station 25 at different poll rates.
NOTE: Different topics can be used simultaneously, and different OPC-compatible Windows
applications can use the same topics simultaneously.
Computer
1784-KT
10
PLC-5
50
SETTING UP
. . . . . . . . . . . . . .A. N. . O. . P. .C. . T. .O.P. I.C. . I.N. .R. .S. L. .I N. .X. . . . . . . . . . . . . . . . .
1. Open RSLinx.
2. Select DDE/OPC > Topic Configuration.
3. Browse to and highlight the PLC you wish to communicate with.
4. Click New.
.....
G RAPHS
This appendix contains examples that demonstrate practical uses of power spectral density to
answer the following questions:
Is Control Improved By Tuning?
Is Valve Life Extended?
Is an Oscillation Hidden in the Noise from a Flow Loop?
The statistical analysis shows a small improvement in standard deviation from Current to New, but
Manual mode has the smallest. Statistical analysis does not always provide the best evidence of
improvement. The power spectral plots are clearer.
The cumulative power plots below show that the new tuning parameters are the best. The objective
is to keep the power at long periods as low as possible.
The following graphs confirm that the new tuning parameters are the best. The objective, again, is
to keep the power low at low frequencies (long periods). Power at low frequencies is lowest with
the new parameters. The difference might not be immediately apparent because the scale on the
graphs is different. Note that for current parameters, the vertical axis has 6 decades, Manual mode
has 5, and New has only 4.
Finally, bumping the system confirms the power spectral data. The new tuning responds much
better, as the power spectral plots predicted.
The new tuning parameters reduce the effort made by the valve and increase the valve’s life.
According to the Process Frequency Response plot (shown below), if the tuning parameters are not
aggressive, the bandwidth will be about 0.2 Hz (natural frequency is 0.4 Hz). All the frequencies
over 0.2 Hz are useless (5s) if the setpoint does not change.
The cycling is probably caused by the positioner because the frequency of the sine wave does not
vary with the amplitude of the PV.
NON-SINUSOIDAL CYCLE
Multiple peaks occurring at harmonics in a power spectrum indicate a non-sinusoidal cycle. Find
the lowest frequency peak and look for equipment cycling at that frequency. In many cases, the
origin of these kind of peaks is another loop tuned too aggressively.
T ime P lot
P ow er Sp ectr um
.....
This chapter contains examples of data collection, diagnostics for various types of conditions, and
explanations of handling problem loops when using RSLoop Optimizer. (See “Guidelines for
Optimizing Loops” on page 18 for more information.)
Data Collection Methods
Examples of Data Editing
Integrating (Non-Self-Regulating) Loops
Temperature Control of Extruders
Cascading Loops
WARNING: Plant data taken for RSLoop Optimizer analysis and tuning must have all process
variables filters removed from the signal.
NOTE: If the controller returns the valve to valve limits (0% or 100%), repeat this test with a
different set point. When the valve is saturated, you will not get accurate tuning results.
The data collected in these two graphs was collected from the
same loop, except the one on the right included a step change. In
this case, the doublet pulse saved 25 minutes.
EXAMPLES
. . . . . . . . . . . . O. .F. .D. A. .T.A. .E. .D.I .T.I .N.G. . . . . . . . . . . . . . . . . . . . . . . . . . . .
Once you collect data, you must decide if you need to edit that data to get the best possible results.
Here are some examples of editing using real plant data collected from a variety of process plants.
Ideally you should use data from one process response. Zoom in on the data from that response and
edit out any noise spikes or invalid data. In some cases, you might not want to zoom in on the data.
This could occur if you need some initial data to average to get a good initial steady state.
Sometimes data requires multiple editing steps to have optimal data for tuning. The next example
demonstrates this situation.
The zooming, editing and averaging in this example would look like:
When you edit out the noise spikes, the actual range of movement for the process variable and the
controller output can now be seen more easily on the Time data window graphs.
This data allows RSLoop Optimizer to calculate optimal tuning parameters.
This process responds faster in the positive (increasing) direction than it does in the negative
(decreasing) direction. To find the optimal tuning parameters for this data:
1. Zoom in on the data when it is increasing.
2. Make your desired controller tuning selections.
3. Note the new tuning parameters calculated by RSLoop Optimizer.
4. Zoom out on the data to return it to its original form.
5. Zoom in on the data when it is decreasing.
6. Make your desired controller tuning selections.
TEMPERATURE CONTROL
. . . . . . . . . . . . . . . . . . . . . . . . . . .O. F. . E. .X.T. R. .U.D. E. .R.S. . . . . . . . . . . . . . . .
Temperature loops on extruders tend to be dominant lag or integrator-like loops. They take a long
time to reach the desired temperature.
When collecting data for these loops, use a small safety factor (about 1.5). Use derivative action if
the output goes to a heater. This might cause the output to oscillate, but provides better response.
To collect data for this type of loop in Manual mode, see “Controller in Manual (Fast Plant Test)”
on page 164.
To collect data for this type of loop in Auto mode:
Possible initial setting: Proportional value of about 80, or gain of 1.25
Possible initial setting: Derivative value of about 1 min
Do the step test with no integral action in the controller. The loop settles faster.
Leave derivative action in.
CASCADING LOOPS
.........................................................
In cascading loops, the output of one control loop (or controller) manipulates the set point of
another. Using a cascade design can greatly improve performance of the control system when the
process can be split into separate control levels.
A typical application of a cascading loop is trying to control the level of a liquid in a vessel. The
master loop measures the liquid level in the vessel, compares it to a set point, and calculates a new
output command. This output command becomes the new set point for the slave controller, which
measures the flow, compares it to the set point sent by the master, and then adjusts the flow rate,
maintaining the desired level in the vessel.
LT LC
Flow (Slave) Controller
FC
Level
Transmitter
Flow
FT Transmitter
NOTE: In RSLoop Optimizer you can use the Relative Response Time feature to check for
interactions between the master and slave loops. (See “Using the PID Grid to Check Relative
Response Time” on page 70 for more information.)
7. See online help for instruction on tuning and analyzing cascade loops simultaneously.
.....
Use this chapter to review the sources of additional information about RSLoop Optimizer
software, including online help and technical support. This chapter helps you to find what you
need efficiently.
You can find more information about RSLoop Optimizer through:
Online Help
Online Books
Technical Support Services
If you need quick help to get started using RSLoop Optimizer software, the RSLoop Optimizer
User’s Guide can help you to learn what you need to know to see results fast. Use it with the online
help system to get information about features and functions in RSLoop Optimizer.
SUPPLEMENTAL READING
.........................................................
These Rockwell Software materials might be useful when setting up and using RSLoop Optimizer.
ONLINE HELP
.........................................................
RSLoop Optimizer includes online help, a tutorial, and optimization steps that provide general
overview information, comprehensive step-by-step procedures, and context-sensitive control
definitions. To view online help while using RSLoop Optimizer:
Choose Help > Contents or Help > Search for Help on from the main window or Faceplate
and Trend window
Click the Help button in dialog boxes
Move the cursor to the box, button, or item you want help with and press F1
The tutorial guides you through the basics of using RSLoop Optimizer. It is available from the
main RSLoop Optimizer window. To use the Tutorial, select Tutorial > Tutorial.
The optimization steps provide detailed information on optimizing your loop. It is available
through the Faceplate and Trend window. To use the optimization steps, select Help >
Optimization Steps.
ONLINE BOOKS
.........................................................
In RSLoop Optimizer, the Online Books feature allows you to immediately access and search the
product documentation for RSLoop Optimizer from the Help menu. This feature contains the
User’s Guide in electronic book format. From the main window, open the User’s Guide by
selecting Help > Online Books.
.....
P ROGRAMMING PID L OOPS
This appendix provides information on using RSLoop Optimizer with the PID instruction used in
the Allen Bradley PLC-5 and SLC 500 families of processors.
PROCESSOR CONSIDERATIONS
.........................................................
PLC-5 PROCESSORS
We recommend that you use the PD file type with dependent gain’s derivative calculation on PV.
The PD file type is only available in New Platform PLC-5 processors.
De ad ba nd
PLC-5 processors allow you to set a dead band range in which the controller output will not
respond to the error. We recommend that you do not use a dead band as it can cause cycling.
De riv ati ve ga in
If you are using a PLC-5 processor with firmware Series/Rev earlier than B/M, C/L, D/C, or E/A
and derivative gain, be aware that switching from manual to auto might cause a spike in the output
of the controller. To avoid the spike, set D (derivative gain) to zero, switch to auto mode, then set
D to the desired value.
The following equations describe the output spike (D is the derivative value and T is the loop
update time).
If D is on error:
D
Output spike = [ ( error now ) – ( error when you switched mode ) ] × ----
T
If D is on PV:
D
Output spike = [ ( PV now ) – ( PV when you switched mode ) ] × ----
T
We recommend that you select the “D on PV” option.
CONTROLLOGIX PROCESSORS
We recommend you use PID with the dependent gains and derivative calculation on PV.
I nteg er cont r ol
Integer control requires a false to true rung transition to execute. This Integer PID operation is
called a rung transitional execution.
Integer control uses 23 consecutive elements of data table memory per PID instruction. One
integer element equals one word of data table memory. For example, if you use N10:0 as your
control address, elements N10:0 through N10:22 are reserved for the PID instruction.
WARNING: If the PV goes from 4095 to 4096, the instruction sees zero for the PV. If the PV
goes from 0 to –1, the PID instruction sees 4095 for the PV.
P D cont ro l
PD control executes as long as the rung is true. PD control does not require a false to true
transition. This PD PID operation is called a rung conditional execution. PD control uses one
element of data table memory per PID instruction. One PD element equals 82 words of data table
memory. For example, if you use PD10:0 as your control address, your next available address is
PD10:1.
Use the done bit of a timer to execute the PID block; do not use a block transfer.
This is the
value to enter.
PV OR SP ENGINEERING UNITS
These are the units displayed by the AutoTune sequence when it asks for a setpoint change. These
are for your display purposes only.
The suggested PV engineering units are:
PLC-5 PD file type: The PV Engineering units should match the Engineering Units min/max
boxes in the PID instruction.
PLC-5 integer file type: If setpoint scaling is on, the PV Engineering units should match the
scaled engineering units. If setpoint scaling is off, the PV Engineering units should be
unscaled units.
SLC 500: If setpoint scaling is on, the PV Engineering units should match the scaled
engineering units. If setpoint scaling is off, they should be unscaled units.
ControlLogix 5550 PID type: The PV Engineering units should match the Engineering Units
min/max boxes in the PID instruction.
ControlLogix 5550 PIDE type: The PV Engineering units should match the PVEUMax and
PVEUMin fields in the PIDE block.
.....
Rockwell Software products are copy-protected. Only a computer with access to the activation key
can run the software. The key is located in an activation file, which is located on the Master Disk
supplied with the program. The activation file contains one activation key per product. Each key
contains one or more licenses, depending on how many copies of the product you purchased.
NOTE: Store your Master Disk in a safe place. If your activation becomes damaged, the
Master Disk might be the only means to run your software in an emergency.
NOTE: Systems attached to extensive networks can take quite a while to search for activation
files on all available drives. You can use the CHECKDRIVES environment variable to specify
or limit the drives your software checks for activation files and to specify the order in which they
are checked. Refer to the activation utilities online help file, COPYPROT.HLP.
To avoid damaging your activation files, do not perform the following operations with activation
files on the hard drive.
Restore from backup
Upgrade the operating system
Reinstall the same version of DOS
Uninstall DOS
Compress or uncompress the hard drive
Turn off Windows for Workgroups 32-bit file access. If activation files were moved to a hard
drive with 32-bit file access on, turning off 32-bit file access results in activation files being
lost. (You can turn on 32-bit disk access and 32-bit file access in Windows for Workgroups
without harming the activation file.)
Defragmentation utilities will not harm activation files.
Before running any type of utility that might modify the structure or organization of the hard drive,
remove activation from the hard drive:
1. Use the Move Activation utility (EvMove) to move activation files from the hard drive to an
activation disk. Run EVMOVE.EXE from your Master Disk or your hard drive (typically
located in C:\PROGRAM FILES\ROCKWELL SOFTWARE\RSUTIL).
2. Perform the hard disk operation
3. Move the activation files back to the hard drive.
NOTE: You must use the move utility, EvMove, to move activation files. Attempts to copy,
move or e-mail an activation file by other means will damage the file.
Disk drive. The activation key resides on a floppy disk (activation disk). Use this method if you
will typically use RSLoop Optimizer on more than one computer — for example, if you want to
run RSLoop Optimizer on a desktop computer at some times and a portable computer at others.
Network drive. The activation keys reside on a network drive. Use this method if you have
purchased multiple licenses of the software and want several users to be able to activate the
software simultaneously over a network. See the online help for instructions on moving activation
to a network drive.
Network activation. Move activation to a network server to allow multiple users access to the
activation.
Moving activation. See detailed instructions for moving activation.
Resetting activation. See detailed instructions for using the Reset utility to repair a damaged
activation file.
Troubleshooting. Look up error messages and get problem-solving suggestions.
You can access online help:
From the Help button on one of the EvMove or Reset dialog boxes
From the RSLoop Optimizer main window: Select Help > Copy protection
Without running either RSLoop Optimizer or the activation utilities. Find and run the file
COPYPROT.HLP on either your Master Disk or your hard disk.
.....
Activation disk. Any disk (floppy or hard) containing an activation file. An activation disk can be
used to activate the software. This is different than a key disk (Master Disk) in that at least one
license of the software must be available on the activation disk to activate the software.
Activation file. A hidden read-only system file that “activates” a Rockwell Software product. The
software will run properly only if your system can find the correct activation file.
Activation key. Activation files contain a database of activation keys. Each key is particular to a
certain product and must be accessible on a local or remote drive for that product to run.
Archiving. The action of saving data collected from a particular control loop to the hard drive.
This data file is saved to the same directory as your control loop (.tun) file.
Auto mode. In Auto mode, the controller or processor automatically controls the output
commands based on its program. This is a closed loop.
AutoTune. The RSLoop Optimizer AutoTune sequence automates the process of collecting data
for PID tuning. Answer the questions to the RSLoop Optimizer AutoTune sequence and
RSLoop Optimizer finds optimal tuning for your loop.
Averaging filter. The averaging filter is a moving average filter — every sample interval it gives
the average of the previous time constant of values. The effect of the time constant is about half
that of the first order filter. To get the equivalent first order time, use an averaging filter with an
averaging window length of twice a first order constant. The averaging filter is a good filter for
reducing frequencies of noise occurring at the filter time. If you have a specific frequency you
want to remove, setting the average filter time equal to the period of that noise should completely
remove it (period is the inverse of frequency). In general, it is not as good as the Butterworth at
higher frequencies.
Butterworth filter. See Second order Butterworth filter.
Cascading loops. The output of one control loop manipulates the setpoint of another control
loop. Includes a master loop and a slave loop(s).
Closed loop. A control loop with feedback. The process variable is compared to a setpoint and a
new output command is sent. This occurs with the controller in the Auto mode.
Control Block. A file that stores PID status and control bits, constants, variables, and internally
used parameters. The Control Block type affects how you should program the logic to execute the
PID instruction.
Controller gain. The ratio of a change in the output to a change in either the controlled variable or
the setpoint.
Controller output. The current output command of the controller. It is sometimes referred to as
the Controlled Variable.
Cycling (oscillating). A condition in a control loop where the process variable does not settle out
at setpoint, but continues to change, often going both above and below setpoint.
DDE. Dynamic Data Exchange. A method of transferring data from one Windows application to
another. For example, RSLinx uses DDE to provide PLC data to RSLoop Optimizer.
Dead band. A user-specified range above and below the setpoint where controller output does not
respond to the error. The dead band lets you control how closely the process variable matches the
setpoint without changing the output. We recommend that you refrain from using dead band as it
can cause cycling.
Dead time. The amount of time that occurs before the process variable moves appreciably after
the controller output changes.
Derivative gain. Causes the change in controller output to be proportional to the change in the
rate-of-change of the error; also called rate.
Disturbance. An undesired change in a variable applied to a system which tends to affect
adversely the value of a controlled variable.
Disturbance, cyclic. Disturbances that occur on a repeatable, periodic basis.
DMC (dynamic matrix control). A multi-variable control algorithm that uses plant models to
achieve optimal control. In DMC a model of the individual pieces of the process are represented in
a coefficient matrix. Each process in the matrix is modeled by using the open loop step response of
the process.
Error. In a control loop, the difference between the setpoint and the process variable.
First order filter. The most commonly used filter. Also called a first order lag. You can
implement it with the equation:
PV + LagTime ( PVf1 – PV )
PVf = -------------------------------------------------------------------
LagTime + SampleInterval
First order process. A process whose describing differential equation is a first order equation.
The time constant for a first order process is called lag time.
Frequency response plot. If you introduce a sine wave into a linear process, the output of the
process is also a sine wave. However, the output might not be the same size and might be shifted in
time relative to the input. The relative size of the output sine wave compared to the input is called
the dynamic process gain or amplitude ratio and the amount of time shift is called the phase. A
different sine wave frequency introduced in your process might give a different amplitude ratio
and phase. If you tried this for a broad range of different sine wave frequencies, you would get a
spectrum of amplitude ratio and phase data at each frequency from which a frequency plot like the
one in RSLoop Optimizer could be made.
Gain. See Controller gain, Derivative gain, Integral gain, Process gain, Proportional gain.
IAE - Integrated Absolute Error. Performance of a loop is judged by how closely the controlled
variable is held to the setpoint, both for setpoint changes and load changes. There are many
mathematical ways of judging this, and IAE is one type. IAE is a measure of controller error
defined by the integral of the absolute value of a time dependent error function.
Improvement Indices. See valve travel index, valve reversal index and Robustness Increase.
Integral gain. Causes the change in controller output to be proportional to the amount of time the
error is present; also called reset.
Integrator. If a process settles out after being given a step change it is called self-regulating. If it
does not, it is called non-self-regulating or integrating. This is much like the integral action of a
controller, which keeps going if the error remains. An integrating process can take a long time to
(or might never) reach steady state (settle out) after a manual step change in the controller output.
Interval (sample interval). The time in seconds that RSLoop Optimizer uses to collect data. This
sample interval is also used to update the trend and bar graphs.
Key disk. A floppy disk that can be used to activate the software even if that disk contains zero
licenses. The Master Disk is the only key disk. This differs from an activation disk in that an
activation disk must contain at least one license.
Lag. The effect of something happening after a change occurs. For example, the valve moves after
the command is sent. The amount of lag time varies depending on the system equipment and
design.
Lag time. The amount of time after the dead time that the process variable takes to move 63.3% of
its final value after a step change in valve position. Lag time is the time constant for a first order
process.
Lambda. See Setpoint tuning.
Manual mode. The mode of the controller where the output command is under the control of the
operator, and not the control loop. This is an open loop.
Master Disk. This disk is supplied with the software. It contains a database of keys in an
“activation file” that enables the software to run. Be sure to store your Master Disk in a safe place.
If your activation file becomes damaged, the only way you can run your software (until the
activation is reset) is with your Master Disk.
Modeling. A mathematical approximation of the control loop. It is based on actual data and loop
response. The model can then be used for testing the response of the loop with different tuning
parameters and under different conditions.
Noise. Unwanted spikes or disturbances in the system. Noise could be seen on an active
transmitter measuring temperature, if for example, the transmitter was subjected to electrical noise
from nearby contactors or similar equipment. Noise should be edited out of data that is to be used
to calculate tuning parameters.
See also Normal process noise.
Noiseband. When performing a hysteresis check, the amount of noise as a percent of the process
variable range. More than 2% might be an indication that filtering should be used.
Normal process noise. Minor variations in the process variable (PV) that are not caused by the
loop’s control. Normal process noise can come from sources such as electrical interference,
magnetic fields, turbulence in flow loops, or waves in a tank in level loops.
Offset. In a proportional only control loop, the difference between the setpoint and the settled out
value of the process variable.
Open loop. A control loop without feedback. The controller does not send new output commands.
The operator can change the output command. Occurs with the controller in the Manual mode.
Overshoot. If a process variable is responding to a new setpoint or to a load change, and initially
goes beyond (above or below) the goal, it is known as overshoot.
Power Spectral Density. Useful for seeing the potential improvement from better tuning. Shows
the relative power in the signal at various frequencies ranging from twice the sample time to twice
the data collection period.
Process gain. The steady state ratio of the output of a process to a change in its input.
Process upset. An upset that is not from changing the setpoint. A process upset is something that
causes your process variable to deviate from setpoint when you were running at steady state. An
example in a flow loop would be a change in downstream pressure. In a level loop controlled by
outflow, an increase in inflow would be an upset.
Process variable. The variable that the control loop is trying to control. It is used as feedback for
the loop.
Proportional gain. When the change in controller output is proportional to the change in error.
Quarter amplitude damping. A control loop response in which the size (amplitude) of the
overshoot or undershoot of each successive cycle is one fourth the size of the previous cycle is
called quarter amplitude damping.
Rangeability. Any valve has a rangeability factor that determines how precisely it will regulate. A
good control valve might have a rangeability of 100 to 1, which means that you can make 1%
adjustments in the output. The valve can also be referred to as having a resolution of 1%.
Rate. See Derivative gain.
Relative Response Time (RRT). is a relative indicator of the speed of the control loop. The
smaller the relative response, the faster the loop. The higher the relative response value, the slower
the loop. Use this number for comparison purposes.
The Relative Response Time is dependant on the speed of response: you can change it by adjusting
the safety factor or lambda time.
Reset. See Integral gain.
Reversals. The number of times the valve has changed direction in CO showing in the time
graph.
Robustness. An indication of how sensitive your loop is to process gain or process deadtime
changes. A very robust loop tends to remain stable under most conditions. It is not very sensitive,
however, and might respond slowly or not hold control tightly to normal changes. The Robustness
plot graphically shows the tradeoff between tight tuning and stability. See also Sensitivity.
Safety factor. Controls sensitivity when using Load type tuning. The setting can be between 1
and 2.5. A setting of 1 means the tuning is very sensitive to small changes in the process but could
become unstable if subjected to large or sudden load changes. A safety factor of 2.5 gives you
conservative tuning; the loop will respond somewhat slower but is more stable when subjected to
large or sudden load changes.
Second order Butterworth filter. This is a high performance second order filter. It is a good
filter to eliminate noise, since it causes the least amount of phase addition for the most amount of
noise reduction compared to other filters. You can implement it with this equation:
v1 + SampleInterval ( PV – PVf1 )
v = --------------------------------------------------------------------------------
Tau + SampleInterval
Server. RSLoop Optimizer uses RSLinx or the ProcessLogix OPC server as the server for
communications to the controller. The link between RSLoop Optimizer and the server can be
either DDE or OPC. RSLinx version 1 use a DDE link. RSLinx version 2 or later and
ProcessLogix use an OPC link.
Setpoint. The desired operating point of the control loop.
Setpoint tuning. Used to match the setpoint response to a first order time constant (or lag time)
that you enter. System response is first delayed by the process dead time. This method is popular in
the paper industry on some loops when operators want no overshoot. Also known as lambda
tuning.
Settled out. See Steady state.
Simulation. Simulation of a control loop is accomplished by using a model of the loop based on
actual data collected from the loop. The model can be used to simulate how the loop will react
under different tuning parameters or conditions.
Steady state. The condition when both the process variable and controller output are relatively
flat horizontal lines in the Trend display, moving within the range of normal process noise. A
process variable or controller output that is ramping or in a transient condition is not steady state.
The process variable does not have to be at setpoint.
Step change. A change of the setpoint or controller output from one steady state value to another.
Step test. A change made to the setpoint or controller output in order to analyze the response of
the control loop.
Stiction. The word stiction is made from the words stick and friction. It is the resistance to the
start of motion. Stiction or stick-slip occurs because of tight packing, undersized actuators, or
corroded stems. Sticking at the seat is very common especially for valves designed for tight
shutoff.
Time constant. For a first order lag filter, the time constant is the time to reach 63.2% of the final
value in response to a step change.
Travel. The total distance the valve has traveled in the CO showing in the time graph.
Valve characteristic. The type of response a valve has to a command is its characteristic. It is a
function of the valve plug design. The correct characteristic should be chosen for the specific
application.
Valve sizing. Choosing a valve to have the proper range and flow characteristics for a specific
application.
Valve Travel Index. The percent improvement in the controller output travel from the current to
new. If the new PID tuning parameters (and filter) result in more travel than the current ones, the
Valve Travel index will be negative.
Valve Reversal Index. The percent improvement in controller output reversals from the current
to new. If the new PID tuning parameters (and filter) result in more reversals than the current ones,
the Valve Reversal Index will be negative.
Verify. The process of using the Time data window to confirm that the control loop data collected
is optimal for calculating tuning parameters.
example H
cycling and noisy data 166
noisy data 166
hidden cycling, detecting 86
PID loop 22
histogram 78
plant test 165
hyperbolic fit, output Characterizer 97
process responds faster in one direction 167
hysteresis 78, 185
expand control loop simulation 129
notes 80
extra
performing check 79
loop, adding 47
trend 44
setting up 44
extruders 169 I
temperature control 169
IAE 77
see integrated absolute error 185
F information, additional 171
initialization error
faceplate 17, 28, 35 LINX 15
RSLinx 15
and trend window 28
bargraph display 29 input characterizer 109
features 2 installation 5, 8
integer control 177
files, delete archive 59
filter 184 integral gain 21, 24, 185
average 183 integrated absolute error. 185
integrating loops 125, 168
first order 184
option 70 integrator 185
PV 70 interaction, preventing 71
interval 185
second order 187
Butterworth 187 sample 57
time constant 74 inverse response process 49
first order 124, 184
filter 184
FORTRAN programming language 99, 112 K
frequency
response 25, 127, 184 key disk 182, 185
KEYDISK environment variable 181
G
L
gain 24
controller 183 ladder logic 175
derivative 21, 70, 176, 184 lag 185
integral 21, 24, 185 time 185
process 81, 124, 186 lambda
proportional 21, 23, 187 tuning 26, 188
graph value 71
changing type (Time data window) 67 lead, process 49
Measurement Noise Response 132 license 179, 185
line color 31
line edit 73
linear (Power Spectral Density) 83
trend 28, 35 V
adding 44
display 29
valve
display, changing 31
characteristics 188
extra 44
life 154
options 30, 31
oversized 81
troubleshooting 172
reversals 132
activation 181, 182
sizing 188
communications 15
travel 132
tuning 17, 25, 69
undersized 81
automatically 37
wear
controller 69
analysis 132
determine optimal parameters 69
indicators 132
from archived data 36, 59
Valve Reversal Index 136, 189
lambda 26
Valve Travel Index 135, 188
offline 40, 53, 69
View menu 32
parameters, calculating 68
report 40
setpoint 26
turn on archiving 29 W
WINtelligent LINX
U see RSLinx 11
Z
zoom 71